How far along are you in your automation journey? Have you dipped your toes into the river of scripting yet? Or did you dive in headfirst and hope that you could swim? There isn’t really a right answer here, as everyone’s automation journey is going to feel a bit different. But one thing that does matter is how you know whether or not your automation plans are succeeding.
Just like any good project, you have to have some kind of reporting. Reports require statistics. And statistics are data. Data comes from measurement. If you’re not measuring the results of your automation journey then you have no way of figuring out how successful it is. There is no way to know for sure if your scripts are doing what they’re supposed to do or if they’re just firing off into thin air. So you need to make sure you’re measuring and monitoring the right systems and data points.
Insights Into Information
During Networking Field Day 22 this past February, we had a great presentation from Juniper Networks all about how to monitor and measure your network using Contrail Insights. Take a look at this presentation from Harshit Chitalia:
The ideas Harshit talks about here are super important to automation. We always talk about building in stages. Sometimes, this is referred to as the “crawl, walk, run” methodology. The parallels are obvious. You have to be good at each stage before you’re ready to leap to the next. If you try to leap over walking and go right to running you’re going to fail. Even if you do manage to succeed at running full speed, what are you going to do when you have to fall back to something that’s between a full sprint and crawling around?
The 5-stage chart in this presentation perfectly illustrates the difficulty in an automation journey. If you want things to work automatically without your input, you have to know what you measure first. You can’t turn everything loose without knowing it’s working like you want. Could you imagine the chaos if you just turned an automation script free to run around the network and do things without you checking up on it to make sure it did what you asked it to do?
Everyone has some kind of horror story about making a typo in a script or doing something that causes an outage or unexpected incident. I’ve even talked in the past about rolling out changes in a staged way to prevent something from running rampant without proper testing. The important thing is not to make the changes happen piecemeal but instead test along the way to ensure the changes are happening the way you want them to occur.
Everyone wants an automation strategy that makes the network completely automatic. Policy is implemented which causes scripts to run and software to program things. But that’s the full-on “run” in the above methodology. And you can’t implement that run without making sure you have a solution like Contrail Insights in place first. You have to have reliable data to build a networking policy to fix issues. Think about something as simple as a traffic spike. Maybe it’s a website that is putting tickets on sale for a concert. Or Black Friday shoppers. Maybe it’s a website for signing up for something important related to a disaster. Whatever the reason you have to be able to figure out that something is happening and you have a need to allocate more resources.
With proper monitoring and data collection, your network can detect an influx of traffic and start automating resource allocation. Maybe you need to start a Kubernetes pod with more load balancers. Or change the way traffic is flowing into the edge to send it across a cluster of devices. Or, perhaps things are so crazy that you need to burst to the cloud or even multiple clouds to handle the load. All of this can happen in the background as you get the alert that resources are being allocated. But only if you have the right insights to make those decisions. And having the right insights also means that you know when the peak traffic has passed and you can return to a more normal level of resources.
Bringing It All Together
Contrail Insights may not be the most exciting part of the Juniper networking portfolio. It’s not the fastest or the biggest or the shiniest. But I will argue that it is the most important because it is the part that helps all the other pieces of the strategy work together. Proper networking monitoring and automation measurement form the bedrock you need to build your lofty goals from. If you don’t measure what you’re doing, you won’t have the insights you need to make it work the way you want it to work.
