Before we get into the thick of why we love APM (Application Performance Management), let’s briefly go over what APM is all about. In reverse. We’ll start with what having a solid APM solution solves.
The APM Goal
Any IT infrastructure is comprised of at least one application and can run into the hundreds, where each one is responsible for various business functions. That’s a lot of responsibility! And one of the many reasons we love… Oh… We have more… We’ll get to that later. To ensure that all these applications are working as they should individually and working as the unit they were set up to, they need to be monitored. Yes. Like small children. Making sure they are all getting along on their own and with each other. APM solutions ensure they are acting responsibly and playing nicely together. It does this by monitoring and managing the availability of these applications and going deeper. APM monitors the performance. Ya know. How it’s doing. Processing time. Number of tasks that failed. Stuff like that.
APM wasn’t always the rock star solution it is today. It’s gone through an evolution as the IT landscape got more complex, with standalone applications making way for applications in a client-server setup. That migrated to distributed application environments and now. Yes. The cloud. Elastic applications. That’s where APM solutions are now. All caught up with a lot more players on the field.
When Things Get Out of Whack
The primary gist of APM’s performance monitoring is to detect anomalies, in other words, be able to alert those responsible when there is a change in performance. Via alerting and timely investigation, IT can prevent failures and costly downtime by pinpointing the root cause of any incident and fixing the problem. The costs can spread like wildfire throughout an organization, but the real damage occurs when it affects systems that deal with customer service, pricing, shipping. Just think if Amazon had just one of its thousands of servers go down. Yeah. You get the picture.
From Detection to Presentation
Since this is a core value of APM, let’s go deeper into this whole anomaly detection. Say you’re a family owned department store chain in the middle of holiday season. Your Point of Sale systems are processing up to 20 sales transaction every 30 seconds, but with a slight increase it can take up to a minute. Why is the question.
The APM solution needs to investigate the path of those sales transactions, including any external dependencies as well as the infrastructure where all this is happening, to pinpoint where your systems are deviating from the norm. Now, this is the part we love. One of anyways. It takes all this information and quickly presents it in a digestible format to all those responsible. It greatly empowers IT teams to pinpoint the root cause of what’s wrong and get to the bottom of things and fix.
Just a few of the things you could monitor with an APM solution are:
It doesn’t matter if it’s a physical or virtual machine, Linux, Windows or more exotic operating system (e.g. Solaris, AIX, HPUX) – after installing the agent you can get all the information about the transactions going through the applications that you’re focusing on.
The Real User Monitoring enables you to see the information not only about the application behavior in the data center, but also on the user side – including browser rendering and users’ network issues.
Monitoring can be set at the level of a specific method in the executable code, and every time the method is invoked you’ll get the full picture of this method’s execution performance.
All these layers can be monitored even without installing a specific agent on the server – only by monitoring other servers, that send outgoing requests to the DB, Mainframe or Message Broker.
Why We Love APM
All the things mentioned above are why we love APM. It just helps everything go smoother. We have explained why APM is super important, but what about justifying the costs to your CEO or CFO. What’s the alternative to having an APM solution in place and not? Reduced up-time, increased down-time, costs to manually detect anomalies and fix them. You’re talking days and lots of man and woman power.
We think we made a pretty good case for APM. What’s yours? What’s holding you back from implementing an APM solution? Feel free to chime in and share some of your insights in the comments section below.
About the author: Boris Krasniansky is a Solution Architect at Correlsense.