Randall Hogan

Developed a React front-end web portal with a Node.js back end to provide rapid centralized access to disparate network data sources. Each page on the front end loads in under a second and provides the user the capability to real-time search all fields as they type, combined with sorting and complex regex multi-column filtering, all with two seconds or less response time. The result sets can then be actioned for automated change creation, incident creation, and project creation.

The project was to develop a fully automated network device upgrade process capable of upgrading up to 15,000 devices in a 24-hour window without causing outages. This was a team project where I was both team lead and key contributor. My key contributions were developing the auto-execution engine, the REST APIs for access, and the front-end dashboard and management pages.

The execution engine starts the upgrades according to a predefined schedule and then monitors the task status and determines when to start new devices optimally, restart the devices that have issues, or remove the devices from maintenance mode when they are completed. The execution engine interfaces with the network automation tool to start and stop upgrade scripts as needed.

The Node.js REST APIs allowed performant access to the data collected near real-time upgrade process. This allowed users to update and action the data as needed, with updates taking effect immediately for other users to see and action accordingly.

Finally, I developed an interactive React dashboard to provide monitoring and near-realtime actionable data with drill-down management pages for more information and to trigger automated actions as needed.

One critical challenge for large networks is performing a complex Regex search of all network configurations quickly. Tools are even considered performant if they complete a config search in less than an hour. In operations, this can cause costly delays in accessing critical data.

I developed a solution for a full Regex search with multiple and/or conditions against 50,000+ network configurations to be completed in ten seconds or less on average. Leveraging Node.js on the back end. The service distributes the search load over several processors, so each worker searches less than 10,000 configurations at a time. Each worker retrieves and optimizes their own configs for rapid searching, and the service manager then maintains a compiled cache of all the search results.

The service leverages two load-balanced servers which keep their cache in sync. The REST APIs serving the front end also maintain a cache, keeping the load off the search engine while keeping their cache current and in sync. Finally, the intuitive and responsive React Web UI provides the users an interface to perform searches and leverage results into various automated actions such as creating projects, changes, or incidents.