Monitoring is an interesting subject. You need to have it but it is rarely true that it represents core business. For many organizations, it is just another cost of doing business. As a result, it doesn’t always get the focus that it should have. This is despite the level of inefficiency that can result if you’re doing it wrong. A scant few years ago, this used to be not such a big issue. Things were simpler, more straight forward, and, let’s be honest, a whole lot slower.
But the world is changing and it’s time to consider whether your monitoring strategy and implementation can cope with it.
Here are 4 reasons you should have a assess your monitoring strategy, and 1 reason you shouldn’t.
Reason to #1 – You have a lot more “things” to monitor.
The way that organizations deliver services has changed. The move from monolithic to microservices provides understandable and maintainable services. It also cranks up the number of moving parts to levels not seen before. The Internet of Things (IoT) is characterized by unprecedented control over and interaction with what used to be dumb endpoints. You also get equally unprecedented growth in the number of end points you need to manage. These factors, among others, are driving up the number of “things” you need to monitor. You often measure these increases in orders of magnitude. For operations groups already drowning in a sea of red, the challenge of scale is only going up.
Reason to #2 – You’re producing a lot more monitoring data.
Instrumentation of the monitored end points is the base of any monitoring strategy. Greater access to instrumentation options, both commercial and open source, combined with an “Instrument Everything” mentality means we’re producing an ever increasing number of metrics and events from each monitored element. The concept is that if anything happens, you’ll have captured everything about it. Unfortunately, most of this extra data is noise that drowns out the useful information. This makes it impossible to reasonable track what is going on. It seems a bit ironic but adding more instrumentation to a legacy monitoring strategy can actually cause monitoring efficacy to drop rather than improve.
Reason to #3 – The nature of modern service delivery is ephemeral.
Everyone understands the benefits of a “Just in Time” approach to resources supporting your services. There’s no point in running a server 7/24 just to run an occasionally intensive activity like end of month reporting. With the advent of cloud, container, and most recently serverless models, the ability to allocate resources on an as needed basis is becoming a reality. The move to the “Software Defined Data Center” means infrastructure is only a logical construct defined at run-time. Even traditional infrastructure delivery around things like network and storage moves into the realm of the transitory. Unfortunately, many traditional monitoring solutions assume that components under management are long lived and once in the system, will continue to be available to attach monitoring data to. In this new world of the ephemeral, these assumptions break at a fundamental level.
Reason to #4 – Adoption of CI/CD
There’s no question that implementing CI/CD improves your ability to deliver functionality to users in a more timely manner while, at the same time, improving the quality of your product. But a similarly increased rate of change is the corollary to this increased rate of deployment. Older monitoring strategies base their assumptions of relatively static services and require time to retool and adjust after a change. That needed time has now disappeared. When something goes wrong, operational personnel are forced into ad hoc manual processes and “Superhero” activities to try and save the day.
Reason to Not #1 – Hats, Pens, Shirts, and Other Vendor Swag
As we’ve seen, there are a lot of good reasons to assess your monitoring strategy but there is also one very good reason not to: vendors trying to entice you to “add-on” additional software. That is not to say that there aren’t opportunities to leverage your existing investments as a base to improve monitoring in your environment. However, a monitoring assessment to determine a clearly defined need and strategic response should be your driver, not simply a new software feature that a current vendor has tacked on. This is particularly true of vendors that provide legacy solutions. All of the reasons to assess listed here require fundamental changes to existing monitoring paradigms, tweaking the old ones just won’t cut it.
The world of IT is changing or, more realistically, has already changed. As with many things, the time to assess your monitoring strategy was actually some time ago. But, as the ancient Chinese proverb points out: “The best time to plant a tree was twenty years ago, the second best time is now”. By honestly evaluating where you are today, and where you want to go tomorrow, you can start to understand your organization’s needs and put the right monitoring in place not just for now but for the future.