When you start down the path of infrastructure automation, one thing you’ll want to consider is what your key metrics will be. For most organizations, projects both in DevOps and automation are aimed at reducing the cycle time needed to make and go live with application changes. It’s logical to look at cycle time as a key metric for monitoring how those projects are faring and providing basic infrastructure automation testing.
Given the tendency of organizations to want to move closer to Continuous Integration and Continuous Deployment (CI/CD), it makes sense to look at cycle time not in terms of general product upgrades, but instead to focus on the time it takes a single code update to propagate through the system. This is different than a manufacturing cycle time metric, where the time measured is how long it takes to make a single part or product. Similarly, infrastructure automation should be analyzed in terms of how quickly a single modification to a virtual workload or configuration can be deployed by the automation system (that is, not by hand tweaking to just get the change done, but changing the state files that describe the desired system and letting the system carry out the update).
You can always try a hosted instance of SaltStack Enterprise for a first-hand, self-guided tutorial.
Looking at cycle time as an approach to infrastructure automation testing means taking into account elements of a change that are outside the scope of the infrastructure automation system. If manager approvals are needed as a part of production system changes, delays in those approvals are delays in the overall cycle time. It’s important to understand these parts of the overall equation not so that management can be blamed for the time involved, but because you want to understand gains in time from automation in terms of what was possible (that is, automation won’t inherently make the administrative overhead in an organization any better).
Automating Edge Computing and IoT
All of this becomes more important as your infrastructure becomes bigger and more far flung. If edge cloud architectures are part of your roadmap, you can see how a lack of infrastructure automation is likely to lead to chaos. Consider also the effects of mobile and IoT devices as part of your overall game plan. For full infrastructure automation, you’ll want to take advantage of SaltStack’s capabilities to provide both agent-based and agentless automation models, not to mention the incredible scalability of Salt-based systems.