JavaScript’s rising popularity has brought with it a lot of changes, and the face of web development today is dramatically different. The things that we can do on the web nowadays with JavaScript running on the server, as well as in the browser, were hard to imagine just several years ago, or were encapsulated within sandboxed environments like Flash or Java Applets.

Before digging into Node.js, you might want to read up on the benefits of using JavaScript across the stack which unifies the language and data format (JSON), allowing you to optimally reuse developer resources. As this is more a benefit of JavaScript than Node.js specifically, we won’t discuss it much here. But it’s a key advantage to incorporating Node in your stack.

As Wikipedia states: “Node.js is a packaged compilation of Google’s V8 JavaScript engine, the libuv platform abstraction layer, and a core library, which is itself primarily written in JavaScript.” Beyond that, it’s worth noting that Ryan Dahl, the creator of Node.js, was aiming to create real-time websites with push capability, “inspired by applications like Gmail”. In Node.js, he gave developers a tool for working in the non-blocking, event-driven I/O paradigm.

After over 20 years of stateless-web based on the stateless request-response paradigm, we finally have web applications with real-time, two-way connections.

In one sentence: Node.js shines in real-time web applications employing push technology over websockets. What is so revolutionary about that? Well, after over 20 years of stateless-web based on the stateless request-response paradigm, we finally have web applications with real-time, two-way connections, where both the client and server can initiate communication, allowing them to exchange data freely. This is in stark contrast to the typical web response paradigm, where the client always initiates communication. Additionally, it’s all based on the open web stack (HTML, CSS and JS) running over the standard port 80.

One might argue that we’ve had this for years in the form of Flash and Java Applets—but in reality, those were just sandboxed environments using the web as a transport protocol to be delivered to the client. Plus, they were run in isolation and often operated over non-standard ports, which may have required extra permissions and such.

With all of its advantages, Node.js now plays a critical role in the technology stack of many high-profile companies who depend on its unique benefits. The Node.js Foundation has consolidated all the best thinking around why enterprises should consider Node.js in a short presentation that can be found on the Node.js Foundation’s Case Studies page.

In this post, I’ll discuss not only how these advantages are accomplished, but also why you might want to use Node.js—and why not—using some of the classic web application models as examples.

How Does It Work?

The main idea of Node.js: use non-blocking, event-driven I/O to remain lightweight and efficient in the face of data-intensive real-time applications that run across distributed devices.

That’s a mouthful.

What it really means is that Node.js is not a silver-bullet new platform that will dominate the web development world. Instead, it’s a platform that fills a particular need.

What it really means is that Node.js is not a silver-bullet new platform that will dominate the web development world. Instead, it’s a platform that fills a particular need. And understanding this is absolutely essential. You definitely don’t want to use Node.js for CPU-intensive operations; in fact, using it for heavy computation will annul nearly all of its advantages. Where Node really shines is in building fast, scalable network applications, as it’s capable of handling a huge number of simultaneous connections with high throughput, which equates to high scalability.

How it works under-the-hood is pretty interesting. Compared to traditional web-serving techniques where each connection (request) spawns a new thread, taking up system RAM and eventually maxing-out at the amount of RAM available, Node.js operates on a single-thread, using non-blocking I/O calls, allowing it to support tens of thousands of concurrent connections (held in the event loop).

Diagram of traditional vs. Node.js server thread

A quick calculation: assuming that each thread potentially has an accompanying 2 MB of memory with it, running on a system with 8 GB of RAM puts us at a theoretical maximum of 4,000 concurrent connections (calculations taken from Michael Abernethy’s article “Just what is Node.js?”, published on IBM developerWorks in 2011; unfortunately, the article is not available anymore), plus the cost of context-switching between threads. That’s the scenario you typically deal with in traditional web-serving techniques. By avoiding all that, Node.js achieves scalability levels of over 1M concurrent connections, and over 600k concurrent websockets connections.

There is, of course, the question of sharing a single thread between all clients requests, and it is a potential pitfall of writing Node.js applications. Firstly, heavy computation could choke up Node’s single thread and cause problems for all clients (more on this later) as incoming requests would be blocked until said computation was completed. Secondly, developers need to be really careful not to allow an exception bubbling up to the core (topmost) Node.js event loop, which will cause the Node.js instance to terminate (effectively crashing the program).

The technique used to avoid exceptions bubbling up to the surface is passing errors back to the caller as callback parameters (instead of throwing them, like in other environments). Even if some unhandled exception manages to bubble up, tools have been developed to monitor the Node.js process and perform the necessary recovery of a crashed instance (although you probably won’t be able to recover the current state of the user session), the most common being the Forever module, or using a different approach with external system tools upstart and monit, or even just upstart.

NPM: The Node Package Manager

When discussing Node.js, one thing that definitely should not be omitted is built-in support for package management using the NPM tool that comes by default with every Node.js installation. The idea of NPM modules is quite similar to that of Ruby Gems: a set of publicly available, reusable components, available through easy installation via an online repository, with version and dependency management.

A full list of packaged modules can be found on the npm website, or accessed using the npm CLI tool that automatically gets installed with Node.js. The module ecosystem is open to all, and anyone can publish their own module that will be listed in the npm repository. A brief introduction to npm can be found in a Beginner’s Guide, and details on publishing modules in the npm Publishing Tutorial.

Some of the most useful npm modules today are:

  • express - Express.js, a Sinatra-inspired web development framework for Node.js, and the de-facto standard for the majority of Node.js applications out there today.
  • hapi - a very modular and simple to use configuration-centric framework for building web and services applications
  • connect - Connect is an extensible HTTP server framework for Node.js, providing a collection of high performance “plugins” known as middleware; serves as a base foundation for Express.
  • and sockjs - Server-side component of the two most common websockets components out there today.
  • pug (formerly Jade) - One of the popular templating engines, inspired by HAML, a default in Express.js.
  • mongodb and mongojs - MongoDB wrappers to provide the API for MongoDB object databases in Node.js.
  • redis - Redis client library.
  • lodash (underscore, lazy.js) - The JavaScript utility belt. Underscore initiated the game, but got overthrown by one of its two counterparts, mainly due to better performance and modular implementation.
  • forever - Probably the most common utility for ensuring that a given node script runs continuously. Keeps your Node.js process up in production in the face of any unexpected failures.
  • bluebird - A full featured Promises/A+ implementation with exceptionally good performance
  • moment - A lightweight JavaScript date library for parsing, validating, manipulating, and formatting dates.

The list goes on. There are tons of really useful packages out there, available to all (no offense to those that I’ve omitted here).

Examples of Where Node.js Should Be Used


Chat is the most typical real-time, multi-user application. From IRC (back in the day), through many proprietary and open protocols running on non-standard ports, to the ability to implement everything today in Node.js with websockets running over the standard port 80.

The chat application is really the sweet-spot example for Node.js: it’s a lightweight, high traffic, data-intensive (but low processing/computation) application that runs across distributed devices. It’s also a great use-case for learning too, as it’s simple, yet it covers most of the paradigms you’ll ever use in a typical Node.js application.

Let’s try to depict how it works.

In the simplest example, we have a single chatroom on our website where people come and can exchange messages in one-to-many (actually all) fashion. For instance, say we have three people on the website all connected to our message board.

On the server-side, we have a simple Express.js application which implements two things: 1) a GET ‘/’ request handler which serves the webpage containing both a message board and a ‘Send’ button to initialize new message input, and 2) a websockets server that listens for new messages emitted by websocket clients.

On the client-side, we have an HTML page with a couple of handlers set up, one for the ‘Send’ button click event, which picks up the input message and sends it down the websocket, and another that listens for new incoming messages on the websockets client (i.e., messages sent by other users, which the server now wants the client to display).

When one of the clients posts a message, here’s what happens:

  1. Browser catches the ‘Send’ button click through a JavaScript handler, picks up the value from the input field (i.e., the message text), and emits a websocket message using the websocket client connected to our server (initialized on web page initialization).
  2. Server-side component of the websocket connection receives the message and forwards it to all other connected clients using the broadcast method.
  3. All clients receive the new message as a push message via a websockets client-side component running within the web page. They then pick up the message content and update the web page in-place by appending the new message to the board.

Diagram of client and server websockets in a Node.js application

This is the simplest example. For a more robust solution, you might use a simple cache based on the Redis store. Or in an even more advanced solution, a message queue to handle the routing of messages to clients and a more robust delivery mechanism which may cover for temporary connection losses or storing messages for registered clients while they’re offline. But regardless of the improvements that you make, Node.js will still be operating under the same basic principles: reacting to events, handling many concurrent connections, and maintaining fluidity in the user experience.


Although Node.js really shines with real-time applications, it’s quite a natural fit for exposing the data from object DBs (e.g. MongoDB). JSON stored data allow Node.js to function without the impedance mismatch and data conversion.

For instance, if you’re using Rails, you would convert from JSON to binary models, then expose them back as JSON over the HTTP when the data is consumed by Backbone.js, Angular.js, etc., or even plain jQuery AJAX calls. With Node.js, you can simply expose your JSON objects with a REST API for the client to consume. Additionally, you don’t need to worry about converting between JSON and whatever else when reading or writing from your database (if you’re using MongoDB). In sum, you can avoid the need for multiple conversions by using a uniform data serialization format across the client, server, and database.


If you’re receiving a high amount of concurrent data, your database can become a bottleneck. As depicted above, Node.js can easily handle the concurrent connections themselves. But because database access is a blocking operation (in this case), we run into trouble. The solution is to acknowledge the client’s behavior before the data is truly written to the database.

With that approach, the system maintains its responsiveness under a heavy load, which is particularly useful when the client doesn’t need firm confirmation of a the successful data write. Typical examples include: the logging or writing of user-tracking data, processed in batches and not used until a later time; as well as operations that don’t need to be reflected instantly (like updating a ‘Likes’ count on Facebook) where eventual consistency (so often used in NoSQL world) is acceptable.

Data gets queued through some kind of caching or message queuing infrastructure (e.g., RabbitMQ, ZeroMQ) and digested by a separate database batch-write process, or computation intensive processing backend services, written in a better performing platform for such tasks. Similar behavior can be implemented with other languages/frameworks, but not on the same hardware, with the same high, maintained throughput.

Diagram of a database batch-write in Node.js with message queuing

In short: with Node, you can push the database writes off to the side and deal with them later, proceeding as if they succeeded.


In more traditional web platforms, HTTP requests and responses are treated like isolated event; in fact, they’re actually streams. This observation can be utilized in Node.js to build some cool features. For example, it’s possible to process files while they’re still being uploaded, as the data comes in through a stream and we can process it in an online fashion. This could be done for real-time audio or video encoding, and proxying between different data sources (see next section).


Node.js is easily employed as a server-side proxy where it can handle a large amount of simultaneous connections in a non-blocking manner. It’s especially useful for proxying different services with different response times, or collecting data from multiple source points.

An example: consider a server-side application communicating with third-party resources, pulling in data from different sources, or storing assets like images and videos to third-party cloud services.

Although dedicated proxy servers do exist, using Node instead might be helpful if your proxying infrastructure is non-existent or if you need a solution for local development. By this, I mean that you could build a client-side app with a Node.js development server for assets and proxying/stubbing API requests, while in production you’d handle such interactions with a dedicated proxy service (nginx, HAProxy, etc.).


Let’s get back to the application level. Another example where desktop software dominates, but could be easily replaced with a real-time web solution is brokers’ trading software, used to track stocks prices, perform calculations/technical analysis, and create graphs/charts.

Switching to a real-time web-based solution would allow brokers to easily switch workstations or working places. Soon, we might start seeing them on the beach in Florida.. or Ibiza.. or Bali.


Another common use-case in which Node-with-web-sockets fits perfectly: tracking website visitors and visualizing their interactions in real-time.

You could be gathering real-time stats from your user, or even moving it to the next level by introducing targeted interactions with your visitors by opening a communication channel when they reach a specific point in your funnel. (If you’re interested, this idea is already being productized by CANDDi.)

Imagine how you could improve your business if you knew what your visitors were doing in real-time—if you could visualize their interactions. With the real-time, two-way sockets of Node.js, now you can.


Now, let’s visit the infrastructure side of things. Imagine, for example, an SaaS provider that wants to offer its users a service-monitoring page (e.g., GitHub’s status page). With the Node.js event-loop, we can create a powerful web-based dashboard that checks the services’ statuses in an asynchronous manner and pushes data to clients using websockets.

Both internal (intra-company) and public services’ statuses can be reported live and in real-time using this technology. Push that idea a little further and try to imagine a Network Operations Center (NOC) monitoring applications in a telecommunications operator, cloud/network/hosting provider, or some financial institution, all run on the open web stack backed by Node.js and websockets instead of Java and/or Java Applets.

Note: Don't try to build hard real-time systems in Node (i.e., systems requiring consistent response times). Erlang is probably a better choice for that class of application.

Where Node.js Can Be Used


Node.js with Express.js can also be used to create classic web applications on the server-side. However, while possible, this request-response paradigm in which Node.js would be carrying around rendered HTML is not the most typical use-case. There are arguments to be made for and against this approach. Here are some facts to consider:


  • If your application doesn’t have any CPU intensive computation, you can build it in Javascript top-to-bottom, even down to the database level if you use JSON storage Object DB like MongoDB. This eases development (including hiring) significantly.
  • Crawlers receive a fully-rendered HTML response, which is far more SEO-friendly than, say, a Single Page Application or a websockets app run on top of Node.js.


  • Any CPU intensive computation will block Node.js responsiveness, so a threaded platform is a better approach. Alternatively, you could try scaling out the computation [*].
  • Using Node.js with a relational database is still quite a pain (see below for more detail). Do yourself a favour and pick up any other environment like Rails, Django, or ASP.Net MVC if you’re trying to perform relational operations.
[*] An alternative to these CPU intensive computations is to create a highly scalable MQ-backed environment with back-end processing to keep Node as a front-facing ‘clerk’ to handle client requests asynchronously.

Where Node.js Shouldn’t Be Used


Comparing Node.js with Express.js against Ruby on Rails, for example, there is a clean decision in favour of the latter when it comes to relational data access.

Relational DB tools for Node.js are still in their early stages; they’re rather immature and not as pleasant to work with. On the other hand, Rails automagically provides data access setup right out of the box together with DB schema migrations support tools and other Gems (pun intended). Rails and its peer frameworks have mature and proven Active Record or Data Mapper data access layer implementations, which you’ll sorely miss if you try to replicate them in pure JavaScript.[*]

Still, if you’re really inclined to remain JS all-the-way (and ready to pull out some of your hair), keep an eye on Sequelize and Node ORM2—both are still immature, but they may eventually catch up.

[*] It’s possible and not uncommon to use Node solely as a front-end, while keeping your Rails back-end and its easy-access to a relational DB.


When it comes to heavy computation, Node.js is not the best platform around. No, you definitely don’t want to build a Fibonacci computation server in Node.js. In general, any CPU intensive operation annuls all the throughput benefits Node offers with its event-driven, non-blocking I/O model because any incoming requests will be blocked while the thread is occupied with your number-crunching.

As stated previously, Node.js is single-threaded and uses only a single CPU core. When it comes to adding concurrency on a multi-core server, there is some work being done by the Node core team in the form of a cluster module [ref:]. You can also run several Node.js server instances pretty easily behind a reverse proxy via nginx.

With clustering, you should still offload all heavy computation to background processes written in a more appropriate environment for that, and having them communicate via a message queue server like RabbitMQ.

Even though your background processing might be run on the same server initially, such an approach has the potential for very high scalability. Those background processing services could be easily distributed out to separate worker servers without the need to configure the loads of front-facing web servers.

Of course, you’d use the same approach on other platforms too, but with Node.js you get that high reqs/sec throughput we’ve talked about, as each request is a small task handled very quickly and efficiently.


We’ve discussed Node.js from theory to practice, beginning with its goals and ambitions, and ending with its sweet spots and pitfalls. When people run into problems with Node, it almost always boils down to the fact that blocking operations are the root of all evil—99% of Node misuses come as a direct consequence.

In Node, blocking operations are the root of all evil—99% of Node misuses come as a direct consequence.

Remember: Node.js was never created to solve the compute scaling problem. It was created to solve the I/O scaling problem, which it does really well.

Why use Node.js? If your use case does not contain CPU intensive operations nor access any blocking resources, you can exploit the benefits of Node.js and enjoy fast and scalable network applications. Welcome to the real-time web.

About the author

Tomislav Capan, Croatia
member since February 12, 2013
Tomislav is a software engineer, technical consultant, and architect with over 10 years of experience. He specializes in full-stack, highly scalable, real-time JavaScript and Node.js applications, with past experience in C#, Java, and Ruby. He is an agile Kanban practitioner who loves to collaborate on development projects. [click to continue...]
Hiring? Meet the Top 10 Freelance Node.js Developers for Hire in March 2017


Anony Mouse
For relational DBs on Node, I like
Adin Scannell
Awesome article! I definitely agree that node.js has some really perfectly suited use cases. However, I do want to comment on something that is a bit of a pet peeve of mine -- I wish you wouldn't contrast it with a non-existent straw-man "traditional" system in "how it works". 1) No server spawns a thread per request (they use thread pools or process pools). 2) You say "cost of context switching" as if it only applies to OS threads. Userspace frames need to be saved and loaded in the same way. Plus the OS does it with a few instructions, leveraging specialized support from the hardware -- which userspace can't do. 3) Similarly, userspace threads (I.e. frames or closures if you don't want to talk threads) take memory resources in the same way kernel threads do. It's not as if each system thread has it's full stack limit allocated, so that's an unfair analysis. Normal systems regularly have more than 4000 without coming anywhere near where you've pegged it. 4) The BIG problem with single-thread concurrency is the lack of parallelism. Sure, you can handle thousands of requests per second, but only one CPU on your 40 core server is going to doing ANY work. Anyways, all the above is in regard to a pretty minor paragraph in your excellent post. Node isn't really guilty of this, it's just that there's a lot of FUD out there around threads and processes which people often use to justify insane designs (and avoid threads when they are completely the right approach for the majority of situations).
Eric Elliott
All your advice about computation heavy apps could not be more wrong. It's certainly true that attempting heavy computation inline with the request-response cycle, is a bad idea, but the same could be said of threaded environments. If you have CPU bound operations, it's a good idea to handle them with worker processes. JavaScript, and Node in particular are actually very well suited to handle distributed computation - especially with the good support for functional style programming. If you write your algorithms using pure functions and distribute workload to workers, you can easily distribute your workload over networked clusters. Node's great support for networking makes it an ideal environment both for computing and orchestration tasks, and it's orders of magnitude faster than Ruby at both.
jim thomas
Very helpful. Thanks a lot!
You certainly don't want to block any Node.js process that is handling server requests, but that doesn't mean you shouldn't do heavy computations behind a Node.js server. So long as the process doing the heavy computations is spawned asynchronously from your server process. Still you could end up blocking your Node.js process if you overdue it, but that is also true with the traditional threaded server.
Danny Machal
This is going to make me out to be a giant noob but can you give me an example of a "blocking operation" ?
Irné Barnard
Any such hybrid system? I'm thinking: Use Node.js to process a request by simply placing it onto a to-process queue, return a message to the client stating something like "calculating...". Then Node.js is free to continue with the next request. The to-process queue can then be run through using a difference thread (or even multiple threads). As and when these complete, they send their results to Node.js's queue which will then relay it back to the original client? Of course this means some ID key needs to accompany each item in this queue to ensure the correct data is returned to the correct client. Anything like this possible? Or even already implemented?
Great article, thanks :)
Gerd Jungbluth
Tomislav, thanks for this very well written and concise article!!! We 've been using MongoDB and Node.js (in combination with AngularJS for the user facing part) for > 2 years now and couldn't imagine to ever, ever, ever switch back to Flash (after 10 years of experience) or JEE / RDMS. So it boils down to just one programming language (JS), one data format (JSON) and one programming paradigma (Async), wow.
That's by far the best explanation out there about Node.js. I finally understood that it is very useful (in certain scenarios) and not just a hype. Thanks a lot this was very informative!
Node works well with relational databases, just don't use an ORM. SQL isn't that hard to learn. =)
Tomislav Capan
True that, but I thought we as an industry got over the idea of writing all the SQL manually. Tools are good for common operations, tools that just get out of the way when needed to write some specifics manually, otherwise reducing the possibility of errors and security issues (that happens with more junior developers whether we want it or not).
Tomislav Capan
Hi Adin, let me comment back 1) same scenario happens, there's a limitid amount of threads serving limited amount of clients. 2/3) Referenced presentation shows some measurements and numbers, you may be quite right on the internals but the common rough overview still stands as a general comparison of how things work between those two worlds. 4) parallelization options are also discussed within the article - as background worker processes, or several node processes behind a reverse proxy, or with Node clustering API (which is still in Experimental, but will be there eventually). Thanks for the great comments and the quality feedback on the article with that additional info, I appreciate it.
Tomislav Capan
Yes, you can have multiple worker processes, even communicating through Message Queue (MQ). Those workers can be separate Node processes (as node is single-threaded, unless you experiment with clustering API - I haven't tried yet as the API is very early and probably immmature), but can be any other language. I've worked on such a system which ran C# on Mono for background processing in a distributed CQRS architecture.
Tomislav Capan
Any calculation that keeps the CPU busy until the calculation is finished. Imagine some operation that requires 2 seconds to perform the calculation. Hit that with 100 clients - you get a 200-sec delay. Note the article I have referenced, which explains the blocking of the event loop:
Tomislav Capan
Yes, that falls under the idea of having 'backend worker processes' in a distributed system. As the system is distributed, those workers can use any language/platform, including Node.
Tomislav Capan
Thanks for your feedback. I agree on that completely, for worker processes you could use JS when it fits, and that implies Node.js as that's what runs JS on the server, but you can also use other languages that do the particular work at hand fast.
For ORMs in general, I just see them as a tool that can get something done quick for a newbie, but end up really gumming up the works later on. The closest thing I use for an ORM is a PDO wrapper (which I borrowed and rewrote from an old co-worker), that helps with writing PDO statements.
Adin Scannell
Apologies for the wall of text. Excellent discussion! 1) Agreed. But threads and processes are powerful tools. That's why a hybrid of threads/processes and event systems typically does best in the real world. Like the much beloved nginx :) 2/3) Sorry, I may not have been clear. When I said FUD regarding threads and processes, I *meant the referenced presentation*. It's just a bunch of specifically tailored microbenchmarks designed to prove a certain point. (Which is fair, given that it's a lightning fair and can be a bit polemic. In fact, it's a great talk. But these synthetic micro-benchmarks are not the basis for a fair and through comparison.) To say the difference between nginx and apache benchmarks is purely because of context switching is an extreme oversimplification. Nginx is specifically designed to serve HTTP requests really, really quickly in common circumstances (IMO generally by a tight coupling with the latest OS event systems, etc.). You could specifically measure the overhead of context switching, and I would wager it's trivial. 4) But then aren't you at the mercy of the horrible "process overhead" the referenced presentation talks about -- can't have it both ways :) (To be clear, my position using threads/processes or whatever is not in and of itself a problem. There are way more important design factors there, the OS overhead of those entities is pretty trivial. Hence I hate it when people adopt a silly design based on the idea thread-are-bad or processes-are-bad or some other such nonesense).
Tomislav Capan
Everything always needs to be put in the right context. I have presented possible situations, to analyze each one deeply I'd need a book :-) Still, I really appreciate your comments and insights, they are valuable addition to the article. Thank you for those.
Aaron Wang
When use node.js as api server, and the back-end db is the bottleneck, if the clients are the other applications, not user interface, can we just let these client requests hang there waiting for db operation complete(sine node.js can handle massive concurrent connections easily)? Is a MQ necessary in this scenario? In my opinion, node.js is right the queue.
ORM does not exist to make query languages "easier", it's a tool used to encapsulate database concerns, isolating them from the application. There are several benefits, but ultimately, it makes applications easier to test and maintain years down the road. Excuse the tangent, nothing to do with node.
Eric Elliott
Yes, you could, but Ruby would be a poor choice if your aim is performance. =)
thank you for a great article. By using the node.js child process - would you be able to overcome the high computational blocking Fibonacci issue?
nene odonkor
any real life examples?
nene odonkor
What of the Facebook example you used. When a user clicks on the like button there is an immediate acknowledgement but the data is written later. Cant that be an example of node used with relational db? Anyway what makes up the message queue?
Moch Lutfi
Maybe go-lang is alternative choice. :)
Anthony Hildoer
This article is great, except for the part where it says don't use NodeJS for computation because it doesn't have threads. Since when do we need threads? Run child processes. The only advantage to running threads over child processes is shared memory. Last time I checked, any system large enough for this entire debate to be relevant anyway is going to span multiple servers anyway, their by nullifying any benefit of threads. So, get it out of your head that NodeJS can't do CPU intensive work. And, if you can't, contact, and we can fix all the stuff you built wrong with NodeJS.
One application that few people use, but could be really fanastic, is using NodeJS to build a desktop application. There's plenty of packages out there - personally, I favour Node-Webkit
Hi, nice article. However I would like to point one thing, NodeJS is not running in a single thread. The programmer doesn't have to spawn new threads, they are handled by node itself on event basis. NodeJS is evented, each function call per event will run in a separate thread. That approach encourages writing lighter functions. If your function does a lot of computation, reactor it into smaller ones and they all will run in separate threads. Think about the example where you process file while streaming. That is possible thanks to threads.
Juanchi Viotti
Great article! BTW, what software did you use to make the images/wireframes?
Matthew Keas
This is very well written. Thank you for this!
Matti Schneider
> The technique used to avoid exceptions bubbling up to the surface is passing errors back to the caller as callback parameters Excuse me, but that seems misled. In my understanding, [“Node-style callbacks”]( (i.e. the pattern of passing errors as the first param to callbacks) are a side-effect of the event queue (returning control as soon as possible to allow for “concurrency”) itself rather than a design to avoid interrupting flow. The fact that exceptions do not bubble are actually quite often a source of errors, especially to newcomers.
Great Article. Wondering who made the beautiful diagrams :)
Great article, thanks :) BTW, which tool do you use to draw the images?
Matthew Keas
+1 for that
Matthew Keas
For those that are curious, it seems the images were created with Adobe Photoshop CC. I checked this by looking at the EXIF data of one of the images: File Size – 61 kB File Type – PNG MIME Type – image/png Image Width – 624 Image Height – 600 X Resolution – 72 Y Resolution – 72 Color Space – sRGB Color Mode – 3 Compression – Deflate/Inflate Orientation – Horizontal (normal) XMP Toolkit – Adobe XMP Core 5.5-c014 79.151481, 2013/03/13-12:09:15 Creator Tool – Adobe Photoshop CC (Macintosh)
Juan G. Nuño
Also, the fact that javascript is not able to check type compliance introduces dificulty in expontaneous organization of huge number of coders updating the same codebase simultaneously.
Juan G. Nuño
mmmm... disagree, if you use correctly a good ORM (take a look at Mature ORMs) you get also a distributed cache of your Relational Database that allows you more performance for the same bucks and more scalability of your sistem... ORM is not just for easing the life to newies... It is non sense to use a non-blockin sistem such as node.js, if at the end you get blocked at your Database. But using a ORM that way, is not for newbies.
Glad you mention, most people don't mention this.
Richard Ayotte
I'd rather keep Node decoupled from the browser and keep the flexibility of running the frontend on platforms that normally don't run Node like smartphones.
For that matter, since you can do an async shell to a console application, you could easily write your worker in, for instance golang, and then use a generic pool to limit your cpu workers... from there, you can shell out to a more efficient worker. You can also do that for CPU intensive JS as well, I did this for my scrypt-js module (there are binary modules that are more performant, but I wanted one without compiled dependencies). It's not that hard to queue work to other systems, and no reason node can't be used to orchestrate said work.
Could *could* use an intermediate system such as TypeScript, you can also use a linter (jshint) and even require a level of test coverage in order for release. Getting 100% test coverage is generally *very* easy in scripted environments. Would suggest looking into Mocha, Chai, and Proxyquire. If you aren't writing unit tests, type safety really doesn't give you much.
Steve Naidamast
I am beginning to find Node.js quite interesting. However, the paradigm it appears to be promoting is hardly new to IT. In the mainframe communications world we called such capabilities, "re-entrant", where a single process could handle a high level of calls to it. Microsoft implemented similar capabilities with its Singleton object infrastructure and I imagine the Java world has done similar implementations. Oddly enough, the Microsoft recommendation for enhancing scalability across the wires to back-end services was to promote the "Single Call" object structure or one object instance per calling request. Thus, the argument made for Node.js is actually contrary to the Microsoft recommendation; BTW, a recommendation I never quite understood. In any event, as a business ASP.NET developer, I am not sure if I would find any use for a Node.js implementation, though our web designer may. On another note, I would like to add my own opinion on the use of ORMs. ORMs are great tools when faced with an existing database structure against the requirements of a new application as the ORM can handle a lot of the mundane, repetitive coding that is usually found with any database application. However, because ORMs are high-level layers, they are usually not the most efficient options to use against databases whereas direct access through native providers are. In many respects, better to do the repetitive coding for efficiency over applying a heavy-weight interim layer such as an ORM...
Very interesting observation. I'd like someone to elaborate a bit more on that. I would like a more concrete example, like some piece of concrete code calling an async heavy-computational function, and someone explaining what happens if that is run in a multicore equipment. Does the thread that was sercing http requests gets blocked? Or another thread is used for that?
Derrick Simpson
Node is terrible as a web server. Even the creator suggests that this is not what Node is intended for, and that it should be leveraged for it's strengths. JavaScript "Everything", just like worker processes, is ludicrous.
Eric Elliott
This is ridiculous. Node was built with the web as first class, and it excels as a web server. It is rapidly replacing Ruby and PHP in many enterprise organizations because it has demonstrably boosted both application performance (reduced page-load times, etc...) and developer productivity.
You can in fact create multi-threaded servers using Node.js to offer more stable performance but it does take a lot of extra work (see the built-in cluster module and the child_process module). I have created a framework which runs on multiple CPU cores and solves much of the issues discussed in this post. Check it out:
Connor James Leech
does something like mongoose.js solve the node relational database issue?
Rob Tweed
Yes - see the queue/pre-forked worker pool approach adopted by EWD.js: Summarised here:
keith lumanog
yeah, I was wondering about it too. great visual + awesome post.
Erick Ruiz de Chavez
Just to complement what @Tracker1:disqus said on his comment, JavaScript is usually criticized about the type checking (and many other misunderstood features, like "this"), but the truth is that JavaScript is just very different to what most of the developers are used to do, and most of the developers who sadly have bad comments about it are just those who haven't had the interest or the time to understand how it really works. JavaScript development is not only about the language itself but also about the tools you use to work with it, to mention some more you should really take a look at Grunt, Uglify, Yeoman, CoffeScript, etc.
Erick Ruiz de Chavez
I might as well make myself a giant noob, but to put a quick example, any disk I/O operation is considered a blocking operation as it has to perform a physical operation to read some data from the disk. Another blocking operation example is querying a DB, where you basically have to wait for the response before doing anything else. I'd say, any sort of computational operation will be as blocking as the time it takes to complete. In any of the above examples, you usually have your blocking request on line A and line B wont be executed until line A is completed. Node.js instead is asynchronous, so line A is requested and then queued, and line B is executed no matter if line A is completed or not; when line A is complete, you are notified either with a callback or with an event and then you can continue doing whatever you need to do with such result.
Carlos Ballena
Very helpful article!
Jesus Nuñez
npm install felixge/node-mysql. That might be the most useful repository I've seen in ages.
Erick Ruiz de Chavez
I have a small example of mysql and node.js at
How about: "Why the hell would I read an article without an obvious date and time stamp?"
Matthias Lienau
Nice article, indeed. But, @hfuti:disqus, you're wrong in saying "NodeJS is not running in a single thread ... each function call per event will run in a separate thread." As far as I understood, the main NodeJS event loop consuming your JS code in fact runs in a single thread. All potentially blocking and/or long-running tasks like disk I/O is delegated to and executed by libuv as part of the node runtime engine which spawns a *limited* number of threads made available as thread pool. Said this, it's obvious that the main event loop thread is not affected or blocked by other (yet again asynchronous) code execution. So, for example, file I/O with the fs module itself is implemented in a non-blocking manner - in the end as good as the OS/kernel allows non-blocking operations. Of course foreign processes or threads could always be spawned by custom modules or 3rd party services (worker processes). And yes, there are runtime env projects handling more than one node process and their threads on the same machine or cpu cluster. But that's another story. Feel free to correct me if I'm wrong!
This is fantastic.
Jacopo Chiapparino
Nice article, thank you!
I think Django still kicks Node's ass.
Hey Tomislav, I'd love if you could add the above post to the list :)
Victor Lava
Omg, this is amazing.. I have always wanted to learn node.js!
Thanks for providing examples, illustrations and use cases. The detailed explanation and steps in some cases with relevant resources makes it a great read. Where nodejs *should* be used is very useful.
All Node code is thread-bound. It may not run on the same thread at all times (which has nothing to do with Node itself, and instead has to do with the operating system process scheduler), but it is thread bound. In order to take advantage of multiple threads, you need to use something like Cluster:
Agreed! I use Postgres for all sorts of GIS operations in an application I've been working on for about a year, and it works great, and benefits from the same non-blocking I/O that any other database access does. It's just that you have to drop down directly to the SQL. There are a couple ORMs, but nothing that is real mature.
Awesome, thanks a lot for this!
Minor comment: don't confuse Object DBs with Mongo (Document DB). There are 2 completely different things.
i agree, good post.
yes, i agree
Very well written article. You put a lot of thought into it. Quick node/express question I am I am adding rapid data entry, not batch, data entry to my system I have a form (mfntapes) that works great. I have a test form (joe) the prompts for the number of times I want to call the mfntapes form The gist contains the code, fails to work, the after prompting for cnt attempts to call for mfntapes cnt times. The question is how to call form mfntapes x number of times where x is a vairable? See lines 45-55 of
Jesus Bejarano
ORM for real projects are useless hone.
Oncle Tom
Nice article :-) > SERVER-SIDE WEB APPLICATION W/ A RELATIONAL DB BEHIND I wonder why Node is not good for that? You have promise based *SQL libraries (like Knex) which enables non-blocking queries. Node has no opposition with relational databases: it has no interest with blocking I/O drivers only. Relational or not relational.
I wish NodeJS was not (almost) systematicallty presented as THE de facto alternative to the traditional threaded webserver. Async IO is not exclusive to NodeJS! Most major languages out there - including Java and PHP - have async IO frameworks similar to NodeJS. This wiki page lists most of them: I'm not saying NodeJS doesn't have its merit, but I'm sick and tired of seeing people flocking to it as if it's the *only* async IO stack available, and the *only* alternative to the good ol' PHP or Rails stack. To be honest, we may be stuck with Javascript on the browser, and will be for a very long time (except if Javascript becomes just another target platform/language thanks to asm.js), but I can't come up with any good reason to use it on the server side in favor to other available languages. Most of the alternatives are just way better as a language, have better standard libraries, tools and resources. Please stop spreading Javascript on the server, it's creepy :p
Jarle Leopold Moe
How would you upload files? What about Directory integration? (Active Directory, Open Directory) What about communication with external services? All request depending on a response from any of these types of services would block the calls... Node.js is _single-threaded_ it's only applicable for certain types of tasks.
Eric Elliott
These re great examples of why Node is so much better than the competition at web services. None of thee are blocking operations in Node. They're asynchronous. While other servers waste resources spinning off separate threads, Node fires off an event-driven asynchronous operation and keeps taking more requests in the meantime. The practical upshot is that porting web services from PHP or Ruby can deliver between 2x and 10x improvements in simultaneous connections and typically 30% - 60% improvements in average response times. Lots of big companies are doing web Node projects just for these reasons, including Adobe, Paypal, eBay, Walmart, Yahoo!, Groupon, Under, etc...
jenit shah
very nice article and very clear thought regarding node.js where to use and where not
Every major application development platform has already had support for HTML5 for the past several years, including websockets for bi-directional communication. Add to this the strong types, multi-threading and the tons of other stuff that languages such as Java have to offer and I seriously question why anyone would waste their time using this.
Just want to mention that node is not the only server side option for javascript especially since oracle put the nashorn javascript engine on the jdk. Now you can also use all the jvm goodness from javascript we are using vert.x as the async server and nashorn to use javascript. You can see some code samples in this module the project site is
This is an amazing overview ..Thank you
Vishnu Tekale
Very well written. Thanks for sharing.
I'm with you. For trivial application an ORM is fine. You have to remember that most devs really only have experience with trivial applications. So they will argue for an ORM because they can't learn SQL or don't have the time or staff to right stored procs. And they have never heard of Memcached and so think the ORM is a cache...what ever. And the weirdest excuse I hear is about decoupling. The ORMs I have had the misfortune of running across do the opposite. They very much tightly couple the UI to the database. ORM layers are large and buggy. So how do you test it? Yeah, good luck unit testing it (which you have to do because they are buggy).
If you are integrating with AD then you are probably in a Microsoft shop. And if you are uploading files to a web server then your environment isn't very sophisticated. So for you, you probably should just stick with IIS and VB.Net.
Microsoft public recommendations for architecture are a joke. Microsoft is a marketing and licensing company. They are not a tech company primarily
If you have a "huge number of coders updating the same codebase simultaneously" you have larger problems. Dev teams don't scale well. I would run. With too many cooks in the kitchen the broth will spoil. There is going to be no accountability and no responsibility with that many people contributing.
Gleb Bahmutov
Why Node is different - using bar as analogy
<a href="">شركة تنظيف شقق بالمدينة المنورة</a> <a href="">شركة تنظيف فلل بالمدينة المنورة</a> <a href="">مكافحة حشرات بالمدينة المنورة</a> <a href="">شركة تنظيف شقق بالدمام</a> <a href="">تسليك مجاري بالدمام</a> <a href="">شركة تنظيف منازل بالدمام</a> <a href="">نقل اثاث بالمدينة المنورة</a> <a href="">شركة تنظيف شقق بالمدينة المنورة</a> <a href="">شركة كشف تسربات المياه بالمدينة المنورة</a> <a href="">شركة نقل اثاث بالدمام</a> <a href="">شركة مكافحة حشرات بالخبر</a> <a href="">شركة تنظيف منازل بالدمام</a> <a href="">شركة مكافحة حشرات بالدمام</a> <a href="">شركة كشف تسربات المياه بالدمام</a> <a href="">شركة تسليك مجارى بالدمام</a> <a href="">شركة رش مبيدات بالدمام</a> <a href="">شركة تنظيف منازل بالدمام</a> <a href="">شركة تسليك مجارى بالدمام</a> <a href="">شركة تنظيف خزانات بالمدينة المنورة</a> <a href="">شركة كشف تسربات المياه</a> <a href="">مكافحة حشرات بالمدينة المنورة</a> <a href="">شركة تنظيف مسابح بجدة</a> <a href="">شركة تنظيف فلل بجدة</a> <a href="">كشف تسربات المياة بجدة</a> <a href="">شركة تنظيف بجدة</a> <a href="">شركة مكافحة حشرات ورش مبيدات بالقطيف</a> <a href="">شركة تخزين أثاث بالجبيل </a> <a href="">شركة عزل أسطح بالدمام</a> <a href="">هنا</a> <a href="">here</a> <a href="">here</a> <a href="">هنا</a> <a href="">هنا</a>
Just to let you know you have "support support" in How it works under-the-hood is pretty interesting. Compared to traditional web-serving techniques where each connection (request) spawns a new thread, taking up system RAM and eventually maxing-out at the amount of RAM available, Node.js operates on a single-thread, using non-blocking I/O calls, allowing it to support support tens of thousands of concurrent connections (held in the event loop).
Node.js will never replace Rails as the number one web framework.
Alifa Nurani Putri
Hello, Is node a best choice for video streaming and uploading? Is any different performance w/ PHP for that?
Nick Mok
How come PHP isn't mentioned for interacting with relational databases? It's as if PHP isn't used anymore? Why is it being neglected?
Clain Dsilva
Thanks for taking time to explain Node.js. I really had hard time understanding it from their website as well as Wikipedia. Well written post..!! Keep it up.
Cezar Luiz
What about this Nice article! Thanks!
Mona Ali
<a href=""> شركة نقل اثاث بالرياض </a> <a href=""> شركة كشف تسربات المياه بالرياض </a> <a href=""> شركة عزل مائى بالرياض </a> <a href=""> نقل عفش مكة </a> <a href=""> شركة نقل عفش مكة </a> <a href=""> شركة تنظيف بجدة </a> <a href=""> شركات نقل عفش مكة </a> <a href=""> شركة عزل خزانات بجدة </a> <a href=""> شركات مكافحة حشرات بجدة </a> <a href=""> عزل خزانات بالرياض </a> <a href=""> شركات العزل الحراري بالرياض </a> <a href=""> عزل مائي في الرياض </a> <a href=""> دكتور عماد جراح تصغير المعدة بالمنصورة </a> <a href=""> بالون المعدة </a> <a href=""> تدبيس المعدة </a> <a href=""> شركة القمة لابادة الحشرات بالرياض </a> <a href=""> شركة مكافحة البق بالرياض </a> <a href=""> شركة مكافحة النمل الابيض بالرياض </a> <a href=""> شركات نقل اثاث الرياض </a> <a href=""> شركة كشف تسربات بالرياض </a> <a href=""> نقل اثاث </a> <a href=""> نقل اثاث بالمدينة </a> <a href=""> شركة رش مبيدات بالمدينة المنورة </a> <a href=""> مكافحة حشرات بالمدينة المنورة </a> <a href=""> نقل اثاث بجدة </a> <a href=""> كشف تسربات المياة بجدة </a> <a href=""> شركة تنظيف خزانات بجدة </a> <a href=""> تنظيف فلل بالدمام </a> <a href=""> شركة تنظيف بالدمام </a> <a href=""> شركة تنظيف شقق بالدمام </a> <a href=""> نقل عفش بالرياض </a> <a href=""> شركة كشف تسربات بالرياض </a> <a href=""> تنظيف فلل بالرياض </a> <a href=""> شركه عزل خزانات بالمدينه المنوره </a> <a href=""> شركة تنظيف بالقطيف </a> <a href=""> شركة تنظيف بيارات بالأحساء </a> <a href=" /"> nile7seo</a> <a href="">nile7</a> <a href=" /">nile7</a> <a href="">nile7</a>
Marcelo Lima
Thank you for this article! Cleared a lot of questions I had about Node internals.
Awesome Article !
Insane IT Skills 1.Lifetime access to 24 lectures 2.Over 120 mins of high quality content! 2.Join to a community of over 2687 students learning together. 4.Course information and regular updates 5.Learn node.js and mongoDB from scratch by examples. 6.Course information and regular updates. 7.Discussion forum, which we encourage you to make use of -- both to pose questions about the material to the course instructor and to exchange ideas with your classmates. 8.High quality course support.
rahul garg
As per my understanding with article, if you send any request to node server it will actually process on some background thread (libuv.dll). Now I want to post some file through node server, why it block the node server while it will process on background thread ?
Super Article..!!
Isn't it possible to get the date of posting of your article? or all your posts for that matter. It is really disturbing to read technical posts without dates... thanks
You mention Node.js is good for a chat application because it's not CPU intensive. I can't think of too many use cases that don't require much CPU. For example, could Twitter be built with Node? It seems there's a lot of computation required just to generate your feed. Node.js uses certainly can't be so narrow.
Ian Kaplan
Great article. Thanks. There's an old joke: you're not a paranoid if they're really out to get you. That's the attitude that should be taken when it comes to Web application security. One thing that bothers me about Node.js is that I have no insight into how secure it is. When an environment like Grails, where I have Java/Groovy running on Tomcat I have some confidence in Tomcat security, if for no other reason than the fact that Tomcat has been around a long time and has evolved. This isn't true with Node.js. I was also looking at the Node.js libraries. At least for the file access, it appears to be nothing more than a thin layer over the POSIX file system calls. I like the Java or even the C++ file abstractions more. The main argument that I see for Node.js (made in this post) is that it has really lightweight threads and that you can have JavaScrip everywhere. This last issue is an attraction for those who only know JavaScript. As far as the light weight threads, this is not so much a language issue as a platform support issue. It should be possible to support this kind of thread model in another language.
Huge Web
Toptal designer did the design for them.
Great article! Thank you for all the information.
I agree. I guessed it was around "a year ago" because of most of Disqus comments. This is taken from their meta tags anyway: 2013-08-13.
Adnan King
Ethan Garofolo
Node.js isn't a web framework. It's a packed V8 runtime with some core libraries attached. Node is more like mri in the Ruby world than Rails.
yes <a href=""> ارشفة المواقع </a> <a href=""> كشف تسربات </a> <a href=""> عازل حرارى </a> <a href=""> عزل مائى </a> <a href=""> ترميم منازل </a> <a href=""> تنظيف خزانات </a> <a href=""> تنظيف خزانات </a> <a href=""> ترميم منازل </a> <a href=""> كشف تسربات المياه </a> <a href=" %D8%A7%D8%AB-%D8%B1%D8%B4-%D9%85%D8%A8%D9%8A%D8%AF%D8%A7%D8%AA-%D9%86%D8%B8%D8%A7%D9%81%D8%A9-%D8%B9%D8%A7%D9%85%D8%A9-0555126103_creation"> نظافة عامة </a> <a href=""> عزل مائى </a> <a href=""> عازل حرارى بالرياض </a> <a href=""> شركة نقل عفش بالرياض </a> <a href=""> غسيل خزانات بالرياض </a>. <a href=""> نقل اثاث بالرياض </a> <a href=""> ترميم منازل بالرياض </a> <a href=""> كشف تسربات المياه </a> <a href=""> عزل مائى بالرياض </a> <a href=""> نقل اثاث بالرياض </a> <a href=""> تخزين اثاث بالرياض </a> <a href=""> تخزين عفش </a> <a href=""> نقل اثاث بالرياض </a> <a href=""> تخزين اثاث بالرياض </a> <a href=""> نفل عفش بالرياض </a> <a href=""> تخزين اثاث بالرياض </a> <a href=""> كشف تسربات المياه </a> <a href=""> عزل مائى </a> <a href=""> ترميم منازل </a> <a href=""> مكافحة الحشرات </a> <a href=""> نقل عفش بالرياض </a> <a href=""> غسيل خزانات بالرياض </a> <a href=""> رش مبيدات بالرياض </a>
Node.js is like Rack. MRI is just a language implementation.
Lambda Pool
the problem is that mongodb, its too risky using a database who can't be migrated never...
Lambda Pool
well its true, ORM is a bad practice after all.
Lambda Pool
yes but it does in a very non efficiently way.
Lambda Pool
the database engine itself will be ever more sophisticated than any ORM library home made by someone else.
Yaser Elshafey
Thank you Will allow me to put some links to my website
John Bailo
It's ameezing how asynchronous the web still is.
Jayaraj Poroor
I agree - I'm not a fan of ORMs either. ORM is a classic example of leaky abstraction.
Jayaraj Poroor
Regarding using Node.js with relational databases: Shelloid ( is an open source application server for Node.js that we developed to simplify lot of programmer tasks. e.g., You can declare named SQL queries as annotations and a function by that name automatically gets added to the DB object for you to use.
I think you do not understand the reason for or characteristics of ORM's very well. ORM's not only isolate data concerns, they provide testability and greater flexibility than traditional SQL clients. ORM's also provide in memory access in place of inefficient joins, so they can actually be far more efficient than SQL in real world scenarios. Plus, given the choice of performing complex set operations in a limited procedural set language like SQL or complex in memory operations using a far richer, more expressive and maintainable language, the richer language wins if you ever expect to maintain or extend the application.
You don't know what you are talking about. If you are writing a non-trivial applications and not using an ORM than I'm sure there will be many developers both today and in the future that will be cursing your name as they wade through ridiculous rafts of SQL. Testability? I guess that's just for trivial applications. When I see ridiculous comments like this coming from node.js fanboi's it makes me suspect that they'd be singing a different tune if node.js had decent RDBMS capabilities.
Lambda Pool
you just overate ORM such as most OO fanboys does.
Lambda Pool
your comment is ridiculous too, ORM is not a everyone's accepted pattern.
The irony is that lambda's (aka, anonymous methods) are the basis of both javascript and the preferred technique for C#'s multiple implementation of the same callback pattern, albeit with full type safety. In fact node's major characteristics (2-way, lightweight, asynchronous has effectively been implemented in C# as "SignalR", and there are several language features which are both asynchronous and natively multithreaded, (unlike node.js). Now I could understand if your experience with ORM's was limited to Hibernate (absolutely awful, one of the worst implementations of a major concept I've ever had the displeasure to work with) or even EF, which at least is fairly powerful and expressive due to LINQ, if data nanny overkill. But a micro-ORM like Dapper gives you all the flexibility and power of direct SQL with the benefits of rich data manipulation via objects, and is fast, very fast. And I'm sorry, but OOP is a well proven, very effective design strategy if you know what you are doing, i.e., you are a pro. I see three major problems with the node.js approach that no one seems to have a good answer for: So you start with the premise that there are such inefficiencies in waiting for IO that there is a huge potential there. So yeah, you've got great concurrency, right until you get to the first blocking library, a database, a file server, etc, which is practically every major piece of functionality written. So it's just hurry up and wait for 10,000 connections instead of 100, ok that's something, but where does it get you? So then you start rewriting all those services asynchronously, but that's just a snake eating its own tail, because that's hurry up and wait too: Because in the end every database on earth ultimately comes to a magnetic arm skipping across a disk or reading solid state memory, and that's assuming you're on the same machine, as opposed to distributing data across a pool of far slower network connections. Then you have the single threaded issue: There is probably no more misused language than Javascript because the barriers of entry are so low. And while early adopters might understand a bad practice when they see one, most javascript programmers simply have no idea what they are doing. Thirdly, Javascript is not exactly a particularly fast language, so the advice seems to be don't do compute intensive tasks. Where does that get you? So you end up with an elegant solution to one category of issues, but you're trapped there, with no way out of the non-compute intensive, hurry up and wait ghetto.
Wow, that is one seriously ignorant statement. Uploading files to webservers is *THE* stated rational of node.js's inventor for why he wanted a non-blocking library. So when you fanboi's break out of the single threaded, slow dynamic world of cut and paste code with zero type safety, multithreading, and only one trick pony (lambdas) to match .Net's multiple asynchronous capablities, (all so that you can get 10x more users to hurry up and wait for IO), then you can lecture us about "sophistication". The .Net framework has a vastly more advanced feature set than any language you can name, period, and IIS, properly configured, can beat the piss out of Apache and can readily match nginx as a reverse proxy or for static content. In the end you've got several problems which are intractable: 1. Javascript is slow, and your technology of choice is single-threaded, so intensive compute tasks are off the table. 2. Javascript debugging and testing blows in comparison with any typed language. Pro's use extensive testing, not fly by the pants 'hey it works'. 3. The 'everything is modular' approach is not an architecture, it's an invitation to chaos, like Perl or PHP. 4. Concurrency doesn't buy you crap if all you are doing is waiting for some other IO bound process, which is pretty much everything worth doing with computers. 5. The quality and experience of many javascript programmers (not node.js programmers necessarily, but it's popularity will bring them) is remarkably poor, as demonstrated by comments in which fanboi's prove that they have no idea how other systems that actually don't ignore any computing problem they can't solve work.
Lambda Pool
ORM is just a work around for the problem of impedance mismatch, nothing else. Its not big deal, about OO its a cross paradigm like AOP and its not really based in nothing else beyond encapsulation. Without procedural paradigm OO does nothing, every real skilled programmer already knows that.
Impedance mismatch involves all sorts of ramifications, especially in regards to maintainability, portability, extensibility and performance. To quote Ethan above (since my explanation obviously didn't take): "'s a tool used to encapsulate database concerns, isolating them from the application. There are several benefits, but ultimately, it makes applications easier to test and maintain years down the road." "about OO its a cross paradigm like AOP and its not really based in nothing else beyond encapsulation. Without procedural paradigm OO does nothing" Ok, I'll be sure to spread the word that we should stop fooling around with all this esoteric OOP stuff and get back to a language that lets us do everything vis-a-vis encapsulation: VB6. You'd think a Javascript programmer might want to mention inheritance, if not polymorphism, since javascript has a relatively unusual mechanism for it: prototypes.
A singleton is an instantiable (non-static) object with a private constructor so that only a single instance can be created. (Imagine an instance class with a static property exposing an instance of itself that depends upon a private instance constructor). It was actually implemented in Java before .Net existed. Perhaps what you are thinking of is a non-blocking Callback mechanism, which is handled in C# via delegates, events (multicast delegates) and the new async and await keywords that transform standard C# into a lambda to be executed at compile time. It also handles multi-threaded asynchronous operations via it's Parallel extensions, which permit spreading load across multiple threads/cores as long as that work is not serial in nature. Also, I would encourage you to check out a micro-ORM like Dapper. It's far less comprehensive than EF, but it's far more flexible and gets out of the way, plus it's significantly more efficient than Hibernate and EF (about 5x faster for reads)
Yes this article rocked tits. Esp good for a Croat with ESL! ;) #nodeboy
Ty Turner
We are searching for a Node.js developer. Please see for details! Thank you
Edward Roberts
So after reading your article I still havent figured out why Node js want to reinvent the wheel... I still think node js and the likes are crap...
A great introduction to find out what it is JavaScript and once the answer to from <a href="">bagaimana tips mengatasi wanita frigid</a>
Noob questions: Um, isn't the ease with which you can move request data (i.e. untrusted data) into database (where data is assumed to be trusted) a big hazard? How do you ensure that you never forget to inspect every piece of incoming data when it arrives, before you start trusting it? Generally, I would assume something as popular as Node.js would have thought of this, but I remember back when Rails had blanket model update. That changed real quick when Github's use of this "feature" was exploited (fortunately, by a whitehat). Also, of course, just because you add a conversion speed bump does not mean that people won't make mistakes, but at least they're more likely to give it some thought, which probably means they're going to make less mistakes.
I don't understand the angst against using an ORM. Were you in a proper environment where concerns were separated? The ones bashing ORM just sound like they don't know what they're doing or how to engineer proper software. Why on earth would I want to go from writing software in Java/C#/_whatever_ to drop into SQL where it is hard to version, properly test, and can apparently cause severe brain damage? Everything is a double edged sword - an implementation or a convention like using an ORM over raw SQL really doesn't matter. Depending on the situation raw SQL might be best... it might be better to use a NoSQL store... maybe an ORM is fine! Usually, from my experience, I can tell you that an ORM is better for a lot of reasons, and they have been relayed by M. I spent the time to write my own libraries to abstract vendor specific implementations and you need to create your own mappings. You can easily spawn from a certain state or use existing data structures. It took time to write my libraries and it was not easy at all to do it but it was well worth my time to do it since I can now reuse my libraries. Is it the BEST? I don't know... I like it but I certainly won't go around to arbitrary technical articles that have nothing to do about SQL and post something like "Yes this technology is good but stay away from raw SQL!" I just don't see a need to be bashing anything here, especially an ORM when the article is exclusively about JS and NodeJS.
Probably one of the main points of this article, that gives Node is supposed scalability, is the offloading to a Queue or Service Bus that leads to asynchronous processing. That is a well proven architectural pattern, available in many languages, is especially used in CQRS (Command Query Responsibility Segregation) with Event Sourcing, is very well suited to be used by technologies such .Net Reactive Extensions that provide considerably greater functionality and flexibility than Node. Asynchronous programming and handling its pitfalls has been around without Node for years, if you had knowledge of enterprise development. As for the hate against ORMs... you guys crticising it seem to be moving from front-end development into an area where you have no knowledge or expertise in OO, BDD, TDD or any other proven Enterprise level methodology. No concept of integration other than Twitter feeds. No experience of complex Workflow or scalable caching. This is one of the dangers - you know JavaScript, and a bit of SQL. So everything else is superfluous - until you need it, such as the attempts to bring an element of type checking to JavaScript. Seriously, each technology has its place, but there is no one size fits all approach. Appreciate the strengths of each technology, and use them where appropriate.
Great introduction to node.js! Thanks, i samo naprjed!
Richard Bellantoni
Because they suck for a complex application. Sure if your scenario where you spent all this time trying to make the ORM work the way you need it to, you could have just put that same effort into writing the SQL properly and making sure it's organized etc and you would be further along in the project and have more control over the application. ORM's are great at saving time on a small to medium scale project, but once you delve into more complex and larger applications, you're going to spend either A.) A lot of time coding to make the ORM work the way you need it or B.) Just decide to write the SQL yourself because the time it takes to make the tool work how you need it just isn't worth it.
Very great post you add
sinta maharani
JavaScript is better than some of the other programming languages. therefore <a href="">cara menghilangkan gatal keputihan</a> better fit with JavaScript
Great, informative article. Thanks for sharing this. <a href="">Unlockpwd</a>
Its very good post thanks
Eric Elliott
Pot, meet kettle. Those may be intractable problems if any of them were true. None of them are: 1. a) JavaScript is all JIT these days and delivers 1-2x native code performance (faster than any other dynamic language I'm aware of). b) Non-blocking by default can deliver orders-of-magnitude improvements in code efficiency - transparently. 2. Almost all modern editors support type inference for JS. ESLint, Closure Compiler, and a number of other options offer sophisticated static analysis capabilities. TypeScript even offers a nice structural type system. 3. The opposite of modular is a tightly coupled monolith. That's a bad idea in ANY language. 4. "if all you are doing is waiting for some other IO bound process" - Non-blocking by default means you're NEVER waiting for some other IO bound process. That's why Node delivers such huge improvements on resource utilization. 5. 2006 called. They want their language-snob attitude back. The days when serious engineers considered JS to be a toy are long since over. JS powers sophisticated enterprise applications at just about every fortune 500 today. Additional point: JS is the only language with fully native support for isomorphic code (meaning you reuse most of your application on both servers and clients). You can write JS ONCE, and it will power the server, the web browser, and mobile devices including iOS and Android. See React Native.
1) 1-2x native code performance? Do the electrons run faster on node code? Not a good start. 2) Type inference is mapping variable declarations to types without explicit syntax, i.e., a) it requires actual types, and b) compilers enforce type safety, not editors. And you can dress it up any way you want, but there is no way to enforce sophisticated state analysis with slop-tastic dictionaries of whatever stored in strings & functional delegates. Also, Typescript is not Javascript, it is Javascript with a half-assed type system pasted on top. Even Google is abandoning Js for Typescript in Angular 2.0 Why? Because Google has decided that an untyped system is insufficient for serious work. But is that type system anywhere near as sophisticated as a compiled language? Nope, not even close. 3) You are misunderstanding what I was characterizing as "modular design". The alternative is not monolithic code, but encapsulation, specialization via inheritance (or prototyping), polymorphism, and externalizing dependencies via IoC. The alternative is SOLID, i.e., modern Object Orientation. 4) Your process may not be waiting, but your customers are waiting for the callback. My point is that being able to serve more requests doesn't do you any good if every served request then has to wait on yet another operation. In the end, somewhere somehow, you will eventually be going to a disk or waiting for IO, because that's where the information the customer wants lives. 5) I didn't say that JavaScript was a toy language, I said that most JavaScript developers are well-meaning, cut and paste amateurs, and they will invade the ranks of node.js developers as it becomes more popular. I'll take a strong type system over "full native support" (that's not native) for "isometric" code that can run on clients or servers [you say that like that's a good thing] any day. Wow, Js on android and iOS! I guess the days of Apple adding another strongly typed, native language for iOS are over</sarc> Obviously you've never heard of Mono, the cross platform .Net that compiles and runs on every major OS, produces native runtimes for all three major mobile platforms, and runs on everything from beowulf clusters to wearable devices. This is why we don't respect you. You don't know or understand anything that came before, or anything outside of your javascript bubble. You don't solve any problems that weren't solved before, but yet you're convinced you have all the answers. That seems to be a common theme with anyone whose education system stressed self-esteem over critical thinking.
Eric Elliott
This reply just strikes me as willful ignorance. 1. Really? Rather than investigate and learn the truth, you just want to ridicule the answer? I'm not taking the bait. Watch Brendan Eich's Fluent talks if you're interested in actually learning something. 2. I know what types and type inference are, and I know the benefits of static types. I've been in this game since before JavaScript was invented. TypeScript is a superset of JavaScript that compiles to JS & allows for sophisticated static analysis. It's structural type system is better than the type system Java had back when I was coding in Java, and it's much better (as in more reliable and flexible) than C and C++. 3. "encapsulation, specialization via inheritance (or prototyping), polymorphism, and externalizing dependencies via IoC" - these are all forms of modularity, and JavaScript modules provide viable alternatives to all of them - and in the case of class inheritance, it's far superior. See 4. "your customers are waiting for the callback." Fortunately, that's where the performance I talked about in point 1 shines. JS provides efficient utilization of I/O resources. In fact, dealing with I/O bottlenecks is the entire reason that Node was invented. I've ported several large production projects from PHP and Ruby to Node, and seen dramatic reductions in response times, both average response times and response time ranges - and since a typical Node app utilizes a small fraction of the memory required for C# applications, your customer I/O competes less with the RAM paging you might experience with a compiled C# application. 5. "you say that like that's a good thing" I've seen objectively measurable increases in team velocity ranging from 40% - 60% improvements. Believe it or not, it's a fact, and being more able to adapt to changing needs and experiment more (particularly in the UI layer) delivers very real business value. Why do you think so many enterprise organizations are adopting Node? It's not because some dev prefers JS. It's because they ran the tests themselves and figured out it's a huge win. "Obviously you've never heard of Mono, the cross platform .Net that compiles and runs on every major OS" Yeah, I have - what I haven't heard of is Mono delivering anywhere near the value that Node delivers in enterprise production. Got a good article on that? 'Cause a quick Google search isn't turning up much... Check out this awesome result in the top 3 of the SERP: ...but a quick Google search for Enterprise Node.js delivers quite a bit. Here are the top 3 search results I see: There's really no contest here.
1, yes, I'm mocking your evidently accidental claim that JavaScript makes electrons run faster, or even as fast as native, because it's patent nonsense. Not only is that impossible, it ignores one of node.js' acknowledged weaknesses: It sucks on compute intensive operations, because it's single threaded, which means compute intensive operations block execution... duh! 2. Your knowledge of Java does not qualify you to understand what a competent type system is. Java's generics are a largely useless, johnny come lately me-too feature when compared against to C# generics because they suffer from run-time erasure, in other words, the generic type safety and reflection only works at compile time, because at run time, everything is cast to an object. So when you are going on about static analysis, you are effectively trying to claim it's as good as compile time + run time type reflection, which is very far from the truth. 3. Meh. Chopping up everything into discrete functions is a form of modularity too, but it's vastly inferior to SOLID, which was my point in the first place. And while prototype based inheritance is interesting, it's hardly better than real inheritance, which permits far more flexible arrangements. 4. I don't see how it's impressive to speed up a Ruby app, or refactor some craptastic PHP into something faster. Your memory overhead claims are equally baseless: I can run micro.Net on a watch or on an arduino device. I can write .Net that runs very well on an under-powered phone. Look at the memory chrome consumes for an SPA, or try to run a complex javascript app on a tablet, and then tell me how "lightweight" JavaScript is. 5. The lack of a proper separation of concerns (which is the cause of most maintainability problems) is the number one issue I encounter at enterprise scaled customers, and an impressive team velocity is always how they got there. Why do I think a number of organizations are choosing node? Because typically a mediocre, over-sized team of moderate competence f'd up the previously shiny new thing that was supposed to solve all their problems, so they want to believe the hype that the problem is not them, but their previous technology choices.
Eric Elliott
1. I think you misunderstood my meaning. JS runs 1-2x SLOWER than native -- much better perf than any other dynamic language I know of. It's fast enough to run AAA game engines like Unreal Engine and Unity in stunning quality at 60+ fps. 2. I actually think that a good native type tool would be a good addition to JavaScript, but only if they're user-definable structural types. That said, JavaScript does support static analysis via type inference, and there are a number of ways to provide type hints for dev tools. In addition, JavaScript also has an impressive array of runtime analysis tools. 3. "hardly better than real inheritance, which permits far more flexible arrangements..." Wrong. 4. My apologies. I was not aware of micro.Net. JavaScript also runs as-is on low-powered devices including Arduino, Tessel, and a number of others. Node works great on most of them. If that's not small enough, you can create a custom Node compile, drop features, even swap out the V8 engine for a different JS engine if you need to. You can also restrict yourself to using tiny JS libraries (of which there are many on npm) to keep your codebase compact. 5. "...they want to believe the hype that the problem is not them, but their previous technology choices." That might explain an experiment or two, but the Node takeover is much more than that. We're rapidly replacing apps in a variety of languages with Node. Having worked at a fortune 500 during the transition to Node, I can tell you our justifications: We did experimental ports to Node, found: 1. that the app was faster and more reliable, delivering huge wins in both average response time, and the number of requests we could serve with the same machine, and 2. The developers were more productive for a variety of reasons, including the fact that JavaScript specialists could more easily work on both sides of the stack without context switching, and a lot of code could be shared in both the server and the client. Those advantages have real, measurable influences on the company's bottom line. That's why Node is taking over at both startups and enterprise companies.
Kesava Velugubantla
especially with edge it is really handy !
pawan kumar
Thanks Tamisalv I was looking for quick reference read for understanding node.js and why projects might use it. Reading this gave me a better understanding of advantages of this this technology and also when one might use it. Cheers!
"After over 20 years of stateless-web based on the stateless request-response paradigm, we finally have web applications with real-time, two-way connections." As a result we have webpages which take longer to load nowadays when we have Internet speeds in the order of megabytes per second, than back in the day when our speeds were in the order of bytes per second but our webpages were plain simple HTML. Nowadays we have webpages which load halfway and then stop, which crash at the slightest network error i.e. dynamic IP address reassigning, or a momentary lapse in the WiFi signal forcing you to reload the whole page, whereas browsers are designed to gracefully handle those errors or resume once the network connection is restored, buggy Javascript scripts don't handle errors as well.
I only have 1 question here with your statement... "It's structural type system is better than the type system Java had back when I was coding in Java, and it's much better (as in more reliable and flexible) than C and C++." I can not take that. How? can you examplify? Some Facts (with obviously answer in no) - Why did Node inventor used V8 engine which was made in C++ to power node, if the js type system was so flexible and much better? - Can nodejs decode vp8 codec video by itself as efficiently as C/C++? - Could you build nodejs on top of pure javascript instead of C++? - Is there any such thing as pure javascript? You really have to understand my friend that it is a type system that can take advantage of underlying hardware and makes your program efficient at CPU and memory. Strongly typed C++ can solve any problem, even build node js. Node shines at non-blocking i/o and that is about it, it can not do anything else. Yes you can chop-off node code and make it work on micro devices but will it be efficient and make sense? Can you make product like node where you have to code in C++, yes you can but it would not make sense as power of C/C++ in not in web, its optimizations and hardware. It not like non-blocking I/O is something new, we have had that in many technologies including java, .Net, basic, python and perl, this is very old. The only reason why this thing is in limelight is it has enabled millions of frontend javascript developers, who do not understand pointers and bash about C++, to write server code, which is simply overwhelming, that's why the buzz. And about "Node in particular are actually very well suited to handle distributed computation", why on earth one would write such a statement? Node is not made for computation. it can not compute as efficiently as C/C++/Java. Period. With all due respect, lets not bring C++ in picture here, there is simply no practical comparison at all, or it will be very touchy.
Eric Elliott
You may be interested in this:
Ken Kopelson
Actually, Javascript in the V8 server engine is QUITE fast. The statement that Javascript is not fast is outdated. You combine Node with Chrome and you get a very fast environment. If you understand how Node works, it has an event loop that processes all code that is ready to run. So, you call a function that has a blocking database call inside and a callback when finished, it allows the the function to be called, which returns immediately allowing you to get on with other things while the database call is being processed. So you aren't "hurry-up and waiting" as you suppose. You get on with other things, and once the database call finishes, execution will continue to the callback which is invoked after the database call returns. So, you get the same basic abilities as a multi-threaded environment without all the extra overhead incurred by thread-state swapping. Because you don't get "time slicing" you have to be careful that no piece of code takes too long to run, and if it does, you just break it up using packages like "async". I come from a long C, C++, Delphi, Java background, having mastered the multi-threaded paradigm, and I can state assuredly that the Node single-threaded async paradigm is super cool, fast, and highly scalable...IF YOU KNOW WHAT YOU'RE DOING. But that's the same for any of the other technologies as well. None of these technologies are for neophytes.
A complex application is precisely where well tailored Domain objects and properly decoupled subsystems are most necessary. If you are spending all your time trying to get the ORM to work you should either learn that ORM better or get another ORM. I would assume the former goes without saying so if you are still fighting your ORM it's time to choose one that just works.
Excepting your last comment that these technologies are not for neophytes, I think you are ignoring the last two paragraphs you were responding to. Even the founder of node.js says to avoid compute intensive operations. Yes, V8 speeds up javascript, but that doesn't solve the problem of a compute intensive operation or IO dependent operations blocking a single thread. You then respond that node.js is non-blocking and therefore it can be awaited. Great, I got that, sure, but what are you waiting on? Some code that isn't written in node.js apparently, (so much for the claimed benefit of "one true language to rule them all"), because your single threaded node.js process is happily moved on to something else until the callback. So, node.js works great just as long as you can rely on other processes handling the workload that is blocking (like waiting for IO), or another instance of single threaded node.js, that will (if written "correctly") undoubtable kick the can down the road to yet another process. And that's the issue: there are a limited subset of stuff done in programs that can be executed in parallel/out of order, and/or doesn't rely on blocking operations. Multi-threading is more complicated and has some overhead every time there is a context switch, but it means that more cores can be thrown at parallelizable units of work, and even discrete operations can be handled by the different threads. In a "node.js solves everything" world, you are either relying on something not written in node.js or waiting for your single threaded node.js application to finish up all the stuff that must be done, one way or another.
Ken Kopelson
You know what? You seem to just like spouting on about something you are clearly not qualified to speak about. You obviously have never built anything serious in Node.js. Well, I have, and I can tell you that it works great if you actually program it correctly. That means you use things like job queues, you use the clustering that Node provides, you make sure you do everything with callbacks and promises, you use "async" and "setImmediate" to properly share the processor between code that is waiting and code can now execute, you make sure UI code has priority over CPU intensive code, etc. For example, I wrote an "async" heap sort algorithm that works great, sorting massive lists while not blocking for any appreciable amount of time. I also have a 5000 line heuristic algorithm that is quite complex that I split up so that the main loops are executed using async constructs. I then have these executed from a job queue called "Kue" that allows for efficient use of all cores, no threads, great UI response time, and complex calculation jobs being executed in the background using all available processor power. This is ALL done in javascript with excellent performance in both CPU intensive tasks and response to front-end data requests. In other words mate, the UI is super responsive, and the background processing (complex heuristic calculation) performed quickly and very responsively. This is all done with a single language for both backend and front end, which is a huge deal when it comes to system architecture.
Jacob Ross
Why are they not for "neophytes"? Is it alcoholic?
Jacob Ross
I agree. Most times that I think an ORM is inadequate for complex queries, I write out raw SQL, later to find out the ORM has an "app for that". I like using an ORM as much as possible, but I won't spend too much time making it work for me, otherwise, as M said, I'll find another ORM.
Sounds like the words from a "Michael" I know. ;) I agree with your last statement.. "each technology has it place". But why do you assume all Node developers are front-end developers with no back-end knowledge? Why is it NOT possible for these developers to make a scalable solution? Like or not Node is being used in Enterprise today and will continue to be adopted.
Javascript is a poor choice of language for enterprise/complex development. It is messy, difficult to read, difficult to organise, doesn't support an entire raft of OO paradigms that save a lot of code repetition and provide readable abstraction, predominantly gives run time errors, no AOP, no by convention, no reflection, no generics/templates, no precision of scope, no rich low level processing, and it is not type safe. We are stuck with it in the browser, and frankly it is only inertia and legacy support that means it is still used there. In all rights it should have gone the way of the dinosaurs 10 years ago. The main driver to move away from scripting languages was that they were a maintenance nightmare and led to ball of mud applications that had constant bugs that couldn't be tracked. It has only been a short 10 years since we breathed a huge sigh of relief when serious development moved off of scripting languages and here we are doing it again unwilling to learn from the past, convinced that this is "new and cutting edge", rather than just old, tired and regurgitated. It will end up the same way as it did last time, being talked about with disgust like classic asp and perl cgi. I can only conclude the developers championing it now were just not around to see the fallout of this last time around. Every new generation of developer is convinced they have discovered "the truth", those of us who have seen this cycle of pain just have to sit back and shake our heads in disbelief. Unfortunately you can't teach experience, it is something you have to learn the hard way. Sure if you are an amateur and know nothing else then by all means, but anyone trying to do a professional job needs to leave this alone and stop making populist technology choices without considering the outcomes. If you can't evaluate the long term limitations of a technology for yourself you shouldn't be working in development. Developers need to stop being so childish, acting like a bunch of deluded fan boys, this is a serious business, not a game.
OMG. Eric ! Are you for real? Do you even know how Nodejs works? Did you even try to look at backbone of Nodejs? Just download yourself a source-code and checkout. IT IS IN PLAIN C/C++. http_parcer is C++ lib. libuv is another and single most important C++ library in nodejs backbone which makes nodejs async, event driven and non blocking. Javascript by itself is body without a soul and life. JAVASCRIPT is just a script that you use to script your logic. One day if smeoene ports lua with this libs, he will not need Javascript to code. same for python etc. But the basic fact remains the same. IT IS C++ code that makes what nodejs is, not that javascript is too fast. in fact javascript is slowest in all of the scripting language. So don't flatter yourself in believing that just java-script is great and other things are shit. And do not insult C/C++ if you have no knowledge of it. So, again, i urge you to take back your words: "Javascript type system is better then C++". (Javascript has no type system. and if you still believe that it has one, you are in a very wrong field, go build some scaffolding) Also, these companies are not choosing nodejs for the reason you mentioned. This is possible in almost all languages. The reason why companies are choosing nodejs is that, they get ready made javascript coders, which are in millions and can not do c++/java code, to do server side because it is cheaper. Another reason -> its ecosystem. Yet another reason -> Its lot easier to conduct load tests in nodejs than other scripts. The link you mentioned is work that is intended to be started. Why do not you suggest them that please write this web-assembly compiler in nodejs rather than c/c++/assembly because according to you that is superior. C'Mon man, how can you compare Nodejs (A technology) with a C++ (language) they are not in the same league. C++ makes node possible, its not visa versa.
"The main driver to move away from scripting languages was that they were a maintenance nightmare and led to ball of mud applications" -- could you clarify what you mean by scripting languages, and what replacements "serious developers" migrated to over the past ten years? I was pleased to see a reference to Django in the article and have not run across maintainability issues caused by the use of Python, SQL, or ORMs in general. On the contrary, Python is my go-to language precisely for its maintainability. Your criticisms of js are spot on, but I can't see how they apply across the entire universe of scripting languages.
Well I define scripting languages as runtime compilation languages, but there is a lot of overlap these days. I prefer the reassurance of compile time verification of at least coding accuracy but that is not the only factor. The depth of invasion into the inner workings of the compiler that tend to be exposed by compiled languages these days allow for a whole range of design constructs and patterns that allow programming to be more "intelligent", I just don't find this level of sophistication in the scripting languages I have used. It is a severe limitation for serious enterprise development. ORMs are an ugly approach to data access on relational databases, but you would probably have to be a database developer to realise why. Data design and Program design have different constraints, ORMs do either an injustice or have to be modified so much that they provide no productivity. There are many issues such as security, isolation, atomic operations that ORMs break, and remember a database is a living system and may require changes in between code releases as a matter of course. ORMs are a blunt tool if you want real performance from your database and want high concurrency without locking. Its a detailed subject I could probably write a book on it, so sorry if this isn't conclusive enough for you. Can't say much about Python other than I have heard good things in general. I'm the other end of the market on .Net, I crucify the open source guys next door in productivity and my defect level is about 1% of theirs. I think you need a large system before it makes significant differences, as you need to invest in framework and substrate to get the main benefits back, its "mass coding" that is the enemy here. When you have over a 100,000 code files you need a higher level of maintainability as it is simply beyond human capability to do it file by file (and certainly beyond maintenance budgets). By making core services that consume code as content you can achieve a high level of quality while keeping everything granular and ensuring release are small and targeted rather than entire system drops. Each to their own, but if you are an IT pro you must have seen the millions of script based systems festering away in businesses because no one can find anything or understand how it works. It's such a common complaint I should think it doesn't need justifying.
P.S. I think "serious developers" have migrated to either Java, C# or back to C++ (along with their associated web techs etc). I wasn't really intending to be derogatory but these three probably account for 90% of all commercial development atm. C# wins out on the commercial front for me solely on Microsoft's considerable ongoing investment and new found modernist approach. C++ is not very productive and Java is really starting to look a little dated. Still I work in all three and they get the job done, each have their place.
P.P.S. "Python is my go-to language precisely for its maintainability"... What do you consider maintainability? It is often not what people think it is (or is not as simple as they think). It encompasses the cost of change and that is the primary cost on the business for a living project. Example: I have a service with 1000 public methods and the business asks me to de-prioritise all calls that take over 2 seconds. If I have to modify any of the code in those 1000 calls then my code has seriously poor maintainability. What I should be making is one code change in my service substrate pipeline. I should not even be modifying the substrate I should be probably writing a statistics module and a de-prioritise module for that substrate loaded in a separate dll that can be loaded dynamically. Now my testing is isolated to just this dll (reverse harness testing) and when ready for release I can add this dll and maybe make one small config change, that's it, no regression testing and no risk to existing code, so no production bugs in the service methods. So many typical code bases would require all 1000 methods to be altered or at least marked for an AOP operation. Enterprise design requires upfront anticipation of future "crazy" business requests. I find with most scripting languages and even with Java that finding insertion angles later on is nearly impossible. Even if I have a complete mare in C# I can emit the code directly into the methods using reflection, I have never seen this level of access on a scripting language and even if it was there it would be dangerous code to emit into runtime compiled operations (because I am literally changing the operations content so I would need to test the result of each). This is just one example I could probably come up with 100s... I'm a technical architect (framework and substrate) so it is my place to "save" my devs from backing themselves into a corner. If I do a good job I can reduce the coding and testing effort to 1% of a "mass coded" system. There is a whole other level of development that most devs will never see or appreciate, this means they are never equipped to make the most appropriate technology choices.
Josh Morgan
Interesting article, I've got quite a bit of experience in other realms but I'm somewhat new to Node.js. There's a few things I'd like to clear up: Flash was always async as well, it merely emulated threads much like it sounds like node does by using an event queue. However, I believe it is technical ignorance to claim that trusting thread management to a 3rd or 4th generation language would be better than trusting a well-tuned JRE or operating system optimized for it's multi-core chipset. How exactly do you think threads work in the lower levels anyway? It isn't some "magic", the only way to get true simultaneous code execution is via multiple processors, something you simply can't accomplish with a single thread. It's also a mistake to say that a new "event" does not add memory or clock cycles taken to a stack just because said events are managed by an interpreted scripting language rather than optimized, compiled C++. I'd bet my lunch that a well-written multi-threaded web application written in C or C++ will blow away any node.js app performance-wise, and that's even without getting into servers and their current multi-core processor architecture. If you've got a quad or 8-core server running a single node thread... you're only firing on one piston (quite ironic that Google calls their engine "V8" when considering such a fact). Another thing to realize is that while Flash (or even Java applets) ran in their own runtimes, so does node -- it's just hidden to the user. That is nothing more than "good" (perhaps hostile?) business moves on Google's part. Lets be honest here, if all browsers came with Flash automatically installed on them, and Apple actually supported Flash on their mobile devices, node probably wouldn't even exist today. I have other concerns about security. What kind of protection does it have against cross-site scripting and other attacks? Earlier today I stumbled across a TOR/Bittorrent client that ran in my browser window, and after opening it my computer wouldn't shut down nicely (it scares me to think what it could have been seeding!). No warnings about security or what types of connections my browser window was opening up, just went along with it's P2P business... the hacker side of me could have a real hay day with those kind of "features". I doubt that kind of stuff has been tested much either which means there's a lot of room for bugs, and where there's a lot of room for bugs there's a lot of room for vulnerabilities. But hey, at least your entire stack is all in the same language! Means you can hire less experienced developers for less money, right? ;)
Good Article for Node JS, you can learn Node JS online in or send a email
Avinash Shah
You could remove all the pitfalls of JS by using the its superset aka TypeScript.
Tl;dr Use node for IO-heavy processing and delegating CPU-intensive processing to a cluster of specialized worker nodes (ex database, media processing, etc). This isn't exactly new information. I covered this topic back in '12: Ideally, the HTTP and API servers should be mostly stateless (excl session management) and disposable. They're just a functional pipeline that translates the raw data into consumable representations. That way, the servers are easy to provision/destroy dynamically to meet the spikey nature of demand. I'm not sure why so many of the commenters are vehemently arguing in favor of multi-purpose vertically-scalable server architectures. By nature, vertical scaling will always have an upper limit predicated by hardware capacity. No matter how efficient the code is. The writing is on the wall... You can spend a fortune on hardware and lose sleep questioning the validity of your risk assessment (aka WAG). At the end of the day, bare metal is a fixed asset. Best case, it meets expected demand and justifies the cost. Worst case, it either costs more than it's worth or lacks the capacity to meet demand. Alternatively, you can embrace distributive computing and automate the infrastructure to grow/contract relative to demand. For the people fighting religious wars over which language is best, node | C# | java. Who cares. All 3 allow 'functional-style' programming. All 3, support async processing (natively or through extensions). All 3 can be managed via provisioning. All 3 are perfectly valid for building distributive infrastructure. Choosing which one to use depends on the quality of the tools, whether or not it will be used to extend existing infrastructure, and the perception of the client. Build whatever you're good at building. If you're really good; build whatever is easiest to implement, support, and generate the most profit (or save the most on cost). BTW, kudos to the author. It's nice to see somebody do a comprehensive (and mostly objective) writeup on this topic.
Yes, both languages support horizontal scaling with asynchronous message management infrastructure. CQRS is nothing but an API implementation pattern. CRUD is the typical use case (as it should be) but Node doesn't automagically scaffold 1:1 mappings between DB and CRUD (see rails/laravel/django for that). Node isn't a framework at all, it's just a HTTP server. You can leverage frameworks (ex Express) to more easily leverage the full power of Node but you still have to fine-tune your routes to define a fully functional API. .Net Reactive Extensions have been ported to JS: In fact, even LINQ has been ported to JS (yes, seriously). ORMs are only an issue because they require an additional layer of abstraction from the underlying data. If (read when) the data models need to change to adapt to business demands, both the ORM and the database schema will need to be updated and tested to reflect the changes. Which is not really a big deal if there's a good update strategy in place. As for the rest of your comment, you'd do well if you stepped out of your comfort zone once in a while to see how JS development really works. 1. JS classes are currently supported now via ES6 (also, available client-side via polyfills). Prototypes really aren't much different than classes in terms of encapsulation (except they're a lot more flexible). Compile-time static type checking is even supported via TypeScript/Dart if that's what floats your boat, it's just not the default. 2. TDD/BDD isn't a feature exclusive to statically typed languages. There are a lot of great testing frameworks available in JS (both server/client-side). Choose your flavor, unit testing (Mocha), behavior driven unit testing (Chai), api testing (SuperTest), and continuous integration testing (TravisCI, and many others) are all used extensively throughout the community. JSUnit (the JS equivalent of JUnit/NUnit) is even available if you miss unit testing in Java/.NET. If anything, testing is a basic requirement of any non-trivial JS app that goes into production because you don't have a compile-time type checker to hold your hand. 3. Complex workflow? Seriously? So, you've never heard of NPM scripting, grunt, gulp? Automating any-and-everything in JS is pretty easy. Style enforcement, linting, documentation generation, scaffolding, one-click-deployment, language transpilation, bundling, distribution building, package/dependency management, release management, etc... 4. :cringe: if you rely on the compile-time static type checking system alone to validate user input, you're doing it wrong... Building a data layer in any language requires constraints above-and-beyond what the default types provide. So, either way you'll have to extend your data models with custom validation checks. The cool part about handling validation in JS is you can use the same routines to check user input on both the client/server-side. Less duplication of effort FTW. Contrary to what you think. Javascript really is a 'one size fits all' approach if you prefer to use it as such. Seriously, you can even compile C/C++ directly to javascript using asm.js. Does that mean you have to use it? Of course not. Any developer with a speck of sense wouldn't fault you for choosing C#, it's a great language. I have experience writing code in many languages, including building non-trivial desktop applications in C#. Given a choice, I'd prefer to use Javascript. The mix of, looser constraints and functional/imperative/prototype styles allow for a level of creativity I haven't experienced in any other language. The tools are great, the module system is amazing, and the language itself is getting substantially better with each update.
Upload Files: "All the I/O operations is handled by Node.js is using multiple threads internally; it's the programming interface to that I/O functionality that's single threaded, event-based, and asynchronous." Libuv uses a thread-pool to handle I/O operations (files, sockets, etc) in an asynchronous manner. Where most languages block by default during CPU-heavy I/O operations, Node doesn't. It simply fires an event when the I/O operation completes on the worker thread. Active Directory:
The distinction is that Node is async by default So, the number of developers doing async programming in other languages are the minority so they're not as well represented. "I can't come up with any good reason to use it on the server side in favor to other available languages." Not gonna lie, using Node at first was... Challenging to say the least. Getting used to async-by-default is not an easy transition. The nice part Node is, the primary focus of the platform is building servers/clients so the ecosystem has a lot of powerful tools to do anything dealing with web development. "...have better standard libraries, tools and resources." I'm not sure what gave you that impression. It doesn't use the monolithic-everything-and-the-kitchen-sink base class library approach. The core itself is very small but that's a benefit as it's much lighter to deploy. It also includes a very powerful, full-featured package manager by default so you're expected to add the dependencies your project needs. NPM (Node Package Manager) has over 200K packages and counting. Since the majority of the modules are developed independently from the core, they iterate and improve much faster than the equivalent core libraries in other languages. Dependencies are managed locally on a per-project basis in the package.json file. Typically, it's bad form for a module author to require that their package be installed globally. Installing packages locally prevents version conflicts at the global level and guarantees that -- when you install a package -- everything required to use the module is included. It may seem inefficient at first glance because many dependencies may have copies of the same sub-dependencies (or sub-subdeps, etc) but compared to the cost of including a massive standard library, the storage space is insignificant. The workflow to setup a project is: - clone the source - run 'npm install' NPM will automatically download and install all of the dependencies (incl sub-deps, sub-sub-deps, etc). Since the dependencies (and their specific versions) are explicitly defined in the config, you don't need to check them into source control. In addition, with ES6 (incl the new ES6-module-loader) about to be released, a new JSPM (Javascript Package Manager) has been created to manage client-side javascript dependencies. Module imports in the browser have finally been formalized in the language spec, so Bower and the variety of module-loading pseudo-standards (ex AMD, CommonJS, UMD) will go away.
As said above modern OO languages have a vast array of options to formalize and control your code and solutions that are lacking in scripting languages. That's just the plain truth, no amount of griping is going to change that. My point is there are a lot of developers choosing technology by popularity rather than suitability, that's what makes them fan boys. Right tool for the right job, applies in every trade apart from software development apparently. But that is probably because most developers aren't true "Tradesman", more glorified "DIY'ers". The industry is full of amateurs who don't even know enough to know that they know nothing! They think because they can write an if statement and a while loop they are pros. Techno-weenies...
System level I/O operations (such as files, sockets) in Node are handled by libuv which does use a background thread pool. The difference is, the main thread can fire and forget the task to a background thread and the background thread will notify the main thread (via firing an event) when the operation is completed. Even with background thread processing, doing lots of I/O operations doesn't scale very well. For long-running CPU-heavy tasks (ex image/movie encoding) offloading the tasks to worker nodes is still preferable. In most languages, I/O operations are handled in a synchronous manner so if they requests are made on the main thread, they'll block execution until completed. The reason you don't see a noticeable pause in the UI when this happens is because the UI is asynchronous/event-based and runs on a thread separate from the main context.
There's nothing stopping you from exposing the API as a microservice. WebKit just allows you to run a native JS client. I may be wrong but from what I understand, unlike the browser a webkit client isn't as strictly sandboxed so you can make system calls (ex to open/save files without user input).
I think you made my point for me, frothing unthinking, emotional nonsense, with very little in the way of fact, from a mind so fanatic about one thing it can't even see its failings. You seem to have made a fair few assumptions about what I do and don't know, I have been doing Javascript for 20 years, I know its short comings, I can work effectively in probably 30+ languages, I use what is appropriate, I never indicated otherwise. You need to grow up or find a new industry to work in. People like you are the problem with Software Development, no nothing nobodies who can't even make a case for a technology, let alone use one. Please stay away from the keyboard and do the rest of us a favour.
Do you use version control with a standard workflow (ex Gitflow workflow) where developers make changes on feature branches and code is peer reviewed before being merged? I've been messing around with trying to build an Angular2 website lately and the Angular2 project is still in early alpha so breaking changes are a regular (albeit unfortunate) occurrence. All of the available examples online are pretty-much broken so I've been monitoring the project development on Github. The rate that the core developers are thrashing on the codebase is truly remarkable. What's even more amazing is that every PR is unit tested and continuous integration tested well enough that every release is guaranteed to be fully functional (as far as they've implemented it so far).
Following what Tracker1 is saying. Linting is the equivalent of compile-time checking in JS. I even use a Sublime extension that shows linter errors directly in the gutter of the editor as I'm writing code. If you want stricter checking you can add a style checker such as 'semistandard' which guarantees code styling on a project-wide basis. That means, spaces-not-tabs, indent 2 spaces, consistent functions, curly braces, etc... Type checking is good for superficial bugs (ex uninitialized variables, dead branches, invalid values) but eventually you'll have to verify the code doesn't have logic bugs through unit testing, continuous integration testing, api testing.
Node uses async event-based I/O via libuv (incl a thread pool reserved for I/O requests). The main thread absolutely does not block during I/O operations. It works the same way cluster would except it's built into Node. Check out one of the presentations on libuv for more details.
Performance wise, PayPal seems to think good things about Node For security, the 'cors' module is pluggable to Express and can be used for all the usual CORS control stuff. The 'helmet' module -- also pluggable to Express -- exposes a small suite of features to protect against malicious users including additional cross site scripting protection. I'm not sure if I'd call a Fullstack JS dev 'inexperienced'. Having a solid understanding of multiple domains in a development ecosystem that is constantly evolving is maddening... ya know, "10 years of experience vs 1 year of experience 10 times".
Josh Morgan
Node.js hasn't even been around for 10 years (which is funny considering I have seen job postings actually asking for 10 years experience with it). I understand that following evolving technology is a challenge, and after 20 years I can tell you that by the time you get completely comfortable with any "full stack" it will be less relevant because technology is always advancing. Nothing will change that, it's just how things work. However, you can't really have your cake and eat it too there. Newer technology is less tested and therefore less secure, but older technology does not have as many features. There's always a trade-off there. Anybody who claims otherwise is selling you something.
Anil Verma
I am sold on Node.JS (if your application is into building high scalability networking applications), Node.JS is the way to go in 2015. No wonder why so many startups and large corporations are adopting it. C++, Java, Ruby and Python havew their place in their respective domains. New companies and products will be getting built on wide variety of languages. I predict ROR adoption would still be high in coming years for building web applications (simply because ROR developers are easily available and time to market is so small). Excellent article though Tomislav.
Daniel Jawna
So true! My job is the maintainance of legacy apps usually SQL Server dB's + ms access or pop fronteds. These were usually made by "the grandson who is good with computers" Kind of guy. No foreign keys, but custom date + time functions. Your post is my sentiment exactly!
I find Java very unproductive and very expensive for a lot of low level tedious work. C++ is for programmers who can handle memory access/pointers. Most corporate programmers cannot. Back in the day you had PL/1 and C for software engineers and COBOL for information technology programmers.
joselie castañeda
thanks!!!.. this is so helpful as i am going to make a heavy-computation enterprise application.. i think i'll try node.js on other application.. for now, i'll use ruby on rails
Túlio Spuri
When this article was written ?
Great article
Thanks for this article I think than the argument of same langage for front and dev is the worst I can liste or read. Organized à good coding with js is the more horrible things append in a team. I work since last year on backend project and where this should be coded in some times with mature framework like django it took too much time understand hundreds bugs. Where mongodb feels cool ? Never. I realy thinks than nodejs is a big joke and not So cool. The package manager give us some cool package to patch and hide the bad side of node but there is nothing to do about the callback hell. Finally the code looks like a big sand box than i hatte to open a file for debug or add some lines of code. So my conclusion is than for small app why not but for big and evolutive project not use nor nodejs and mongodb. Regards
This is a fairly ill-informed comment as ES6 was alive and well as at a year ago. Besides the OO from ES6, there is also TypeScript which adds even more enterprise-level OO to JavaScript. Like .NET is compiled, TypeScript may also be "transpiled" to Javascript. NodeJS allows for pretty much of all this now with an even still better utilization of server resources and no OS lock-in. Think cutting your infrastructure costs by 2000% because you really shouldn't need to scale too vertically or pay for that .NET license per node. Fintech companies like Paypal are happily doing great things with node as well so I highly doubt this comment of yours comes from a place of non-ignorance about what NodeJS eco-system truly offers the enterprising product.
I really wanna be able to try out Node.js but it's such a pain in the effing ass to get something running. Installing a "simple" app always turns in to a list of things you need to do, install shit globally (which is NOT always possible), edit files, try to figure out what the damn devs mean in their scant instructions. It quickly goes to shit if you do the slightest thing wrong. If it's so great why hasn't anyone figured out a way to make simple installers with this thing?
Alexis Menest
Ghost blogging platform
David M Tromholt
Yeah it would be nice with a date. I assume they don't show it because they know how biased people are towards new information — but it doesn't make much sense to hide it in an article about a fast growing technology.
Hi Avinash, Can you please eloborate little more on your statement. I am trying to understand more on this.
Would node be a bad idea for something like then? Would you need to do any image manipulation on a different server to prevent blocking?
Hey Tomislav, Great article! I wanted to know that if suppose I want to get external hardware output in my app, e.g, scanner or e-signature (if user is doing e-signature or getting scanned copy from scanner), then can i get directly in my app? it will be like an api call as we do in Java.
Misha R
TypeScript is syntactic sugar wrapper on top of standard JavaScript and compiles into standard JavaScript. It is invented to make the code maintainable and allows it to be used more like a real programming language. Given that you have to use JS where appropriate it makes writing it more familiar to real programmers and makes it maintainable. That said, anything JS and Node is appropriate for thin, lightweight web apps that need to be put together quickly and efficiently, for which things like ASP.NET, JSP, Ruby etc are a bit of overkill. Believing that one can not use the new magic Node for everything just because one guy can write front and back end is amateurish.
Encapsulating a language with another language, which you have to translate will add overhead. Also there are things that don't translate, just in real life translating between language, you loose something. In this case you loose speed (having to translate from one to the other), and optimization, because of reason one. ORM's don't do it all. In the beginning, I was for ORMs, then I dove deep into it and you end up with "How do I do this?", "Oh you can't easily, you have to do 10 other queries", or you end up with stuff that just seems like it should work well, but you end up wrecking the performance. ORMs are for simple "Select blah blah from table where x = y"...
Nils Orbat
Hey, is there a library/tool you used to creste the graphics ?
Praveen kumar Pamani
Thank you for this article , I would suggest this article if people want to know what is node and what can we do with nodejs.
Great article, thanks for the overview!
thanks for great article really very helpful to understand the node.js
Tomislav Capan
Yes, this is true, article was authored and published in August 2013. (sorry for such a late confirmation, haven't seen this earlier)
In what universe can Mongo DB be referred to as an Object DB? It doesn't support inheritance. It's a Document Oriented DB not an Object Oriented DB. I just don't want any of the kiddies to get confused.
We switched to one language for backend and UI but we decided to keep the one that runs lightning bolt fast at runtime and has type safety and supports standard ORMs with lazy loading etc., We didn't want to throw all those things away, which is what you have to do when you go with a JS solution on your backend. The key to using Java for the UI is use a Java framework that does all the JS for you - so you can live in a world devoid of having to deal with JS and stick with compiled, typesafe, enterprise suitable code. This solution is also scalable - J2SE has supported clustering for nearly two decades now. One such Java UI framework that covers all our JS needs and gives us all the 'partial updates' goodness with an AJAX event driven model with websockets option is Wicket.
Rumana Amin
Hi Tomislav, your article is really nice and informative. It helped me a lot to understand Node.js and it's uses.
The simple answer is do not use Node.js for that. What part of that are you missing? You want to serve GEO data coordinates to 500K connected clients?...easy. Do a massive JOIN statement across a million rows on a RDBS? You're screwed. Surgeons do not work with chainsaws. Neither should programmers.
You underate ORM as most of the people who have never built an OO system with a decent highly performant ORM that rocks do (and by 'decent' ORM I'm NOT referring to the ORM most people think of using, Hibernate!)
I agree - this article and associated comments by the JS fan boys confirms my fears that most JavaScript apps must have anemic domain models ( that are quite trivial and virtually expression less - all the work will be done in disassociated (not encapsulated and definitely not polymorphic) business logic.
I'm guessing your experience with 'ORMs' is limited to Hibernate. I had a similar experience until I thought there must be better ORMs out there - look around - others exist.
Good points Adin - I was about to make the same ones. In regard to the article's claim that traditional frameworks "spawn a new thread for each connection (request)": So, so wrong. Must have made more than a few people livid to see such an error which either a) the author knows is clearly wrong and persists with the view or b) has never used any other backend framework and so doesn't realize that he is so wrong. The conflation of connection with request is also interesting. Does his knowledge not extend to understand the concept of connection "keep alives". A connection does not have a one to one relationship to requests as he suggests. This kind of skewed, twisted view of how intelligent, mature, highly evolved frameworks (eg., Java/Tomcat and I'm sure .net/ASP) handle requests makes me feel like this article is tainted and not representing the truth about alternatives. In regard to moving away from the Request/Response mentality - many existing frameworks have done this also and created a component oriented architecture that supports asynchronous model based updates. eg., Java Wicket or Angular JS. It's almost like running JS (a type unsafe, non OO language - if you're a believer that manual, self assembly 'prototype' inheritance is in the true OO spirit) on the server is not a concept that has enough merit without twisting the truth about how traditional server side frameworks work. It's like that or maybe it is that.
Not every client who has an active session needs to be allocated their own thread - threads are shared and each user only needs a thread to service requests. Requests come in in short bursts and don't need to tie up a thread for very long if the back end code is executing in a highly optimized, compiled language and a fast database. For this reason you can't say "System A can support 10,000 threads therefore it can only support 10,000 clients". The number of clients supported is orders of magnitude greater than the number of threads available because of diversity. Most UI's, if well written, only 'phone home' to the server when absolutely necessary - not "all the time".
It sounds like the node.js people think that JS is the only language that supports concurrency (although they promote a single threaded web server solution -WTF?) and non blocking I/O - of course it's not that case but you seem so excited about such things that I can only assume you have just discovered concepts that have existed in other languages for 2 decades or more
Tomislav Capan
Of course, you understand this is an illustration. In practice, it's a thread pool, which is limited by memory available, and is magnitudes of order less than what an event look can support. Thanks for the comment, though
Dean Radcliffe
Damn, this is an oldie but goodie. I give talks just like this, and may borrow your slides. In light of these original intentions of node to do awesome things like queued db writes, and server-push, it's surprising how most node apps today are still built on the request/response model and synchronously (but non-blockingly) waiting on db writes.
Great insides, thanks for the article. Node.js do has plenty of advantages: it's lightweight, efficient, and gives an ability to use Javascript on both front-end and backend opens new possibilities. However, it also has drawbacks you should keep in mind. We have tried to describe some in the article Your opinion would be appreciated a lot.
comments powered by Disqus
The #1 Blog for Engineers
Get the latest content first.
No spam. Just great engineering posts.
The #1 Blog for Engineers
Get the latest content first.
Thank you for subscribing!
You can edit your subscription preferences here.
Trending articles
Relevant Technologies
About the author
Tomislav Capan
JavaScript Developer
Tomislav is a software engineer, technical consultant, and architect with over 10 years of experience. He specializes in full-stack, highly scalable, real-time JavaScript and Node.js applications, with past experience in C#, Java, and Ruby. He is an agile Kanban practitioner who loves to collaborate on development projects.