FLOW Programming in Network Automation: Recipes of Quick Development
With a rapidly changing post COVID scenario, automation is not only a mandatory phenomenon but the time it takes to complete full development lifecycle to production is also becoming critical. CIOs has a major task of democratizing the solutions which are independent of particular skills and dependency apart from finding new automation opportunities across entire organization.
Automation is not a new phenomenon and happening since decades, but it gained significant momentum in last 5 years partly due to multiple technologies/vendors solving the similar problem and advanced virtualization techniques have made things easier to do but difficult to manage. While for engineers an automation may be a set of scripts and its orchestration and deployment, leadership visions Automation as overall Value delivery System spanning multiple verticals of an enterprise. Since than the focus (investment) on automation has seen considerable jump and gradually Automation became as important planned milestone project rather than one of multiple initiatives.
And now with full automation lifecycle management, CI-CD deployments, leaders are exploring further ways on reducing the development cycle and increase reusability aspect. This means “centralized deployment” across enterprises and “full alignment” across multiple teams with focus n standardized solutions.
At Prodapt Labs, team have worked on providing integrating flow programming tool like node-red (https://nodered.org/) and using it for configuration, service provisioning, Device turnup and Onboarding scenarios.
How the Team started:
1. Once we understood node-red (which is pretty easy, Thanks to its community!) team looked at a use case which they already automated by using python. We found one BGP create neighbour instance to begin with.
2. Team started building a customized set catalogue of bots in automation platform. Here they used Netbots automation platform but it’s a choice to use any other framework as well. This was the only tricky part. How they team analyses and synthesize entire flow and create generic nodes.
3. Once we have nodes created, we just need to connect with other flows. This requires logical skills but no coding skills.
4. The time from analysis to production between python-based automation to using node-red was astounding. Few important data below.
Use Case Demo:
Use Case: Consider you are a network engineer who wants to do service provisioning for BGP in cisco router. In normal scenario, being a network expert, will learn few python programming basics, do some development, debug, test and ultimately deployment. Since you are still not expert programmer, your program will just solve “THIS” purpose (no external interfaces, reusability aspects). Overall, you can get going with basic working model within 7 days. With visual programming it will take ½ hour without needing programming skills!! What needs to be done is just identify right process/nodes from the node catalogue management, drag those nodes in studio and stitch them in one flow! You can change available parameters, parse output, put your own customization all in menu driven user interface. While the available nodes can be reused for other flow, the entire flow can be called from any of the external interfaces.
While the above program is quite basic operation, but its capability is boundless. With nodes as abstraction, Prodapt Lab team is visualizing a paradigm shift in the way automations are developed. Instead of talking about basic function as parameter, we have started talking from nodes stitching with each other to complete use case. The comprehensive list of nodes resides in catalogue of Netbots platform where user can filter, customize and use in studio to complete any use case.
Here is the sample of flow created for creating instance and neighbour.
Similarly, for deleting BGP instance:
So using the Netbots catalogue management, just drag your nodes to be used and link them to create the flow.
As you would have imagined, this is just a small example and its true potential can be thought as a set of complex use case. Consider a scenario in which you have an automation which is linked to your organization LDAP for authentication and you need to touch base with a data store for extracting some value. After extraction, you need to have a threshold mechanism, which when breaks generate alarms, trigger email and also send report. With built-in node-red nodes, we can reuse many of the existing functionality. Even if built-in nodes are not found, creating nodes are not difficult and can be reused across organization. This power of node-red can be used in multiple network automation, cloud automation, self-serve testing scenarios. The integration between third part interfaces have become much easier than ever.
We have started using this for many of scenarios and I hope this will create lot of interest in organization who are focused on automation aspects.
