How to get started with application performance monitoring

Rather than checking on your standalone software, modern application performance monitoring products are working with cloud services and microservices, and with artificial intelligence down the road.

Image: gorodenkoff, Getty Images/iStockphoto

The field of application performance monitoring (APM) software is changing, so now is a good time to learn the latest tips for how to use it in your company.

Infrastructure monitoring, which focuses on networks and servers has been around longer, and application monitoring started becoming more complicated earlier this decade with the rise in cloud-based software and microservices, explained Chris Hansen, director of product management at APM specialist New Relic, in San Francisco.

Originally, "APM would rely on angry customers calling in to you when there's a problem," Hansen joked, but modern software is inching toward an environment where artificial intelligence would be useful to predict problems and adjust settings on its own. Another trend is that APM products are now available for smaller companies, not exclusively for enterprises, he said.

"Generally you do not need to make changes to the applications in order to have them monitored. APM vendors rely primarily on instrumenting well-known libraries and frameworks," Hansen explained. Most APM companies also offer programming interfaces so that customers can integrate the monitoring input and output with other systems.

SEE: How system admins can stay relevant: 10 tips (free PDF) (TechRepublic)

What exactly is being monitored varies by company. "Some people assume that it means we're monitoring every bit of their application code, which would cause prohibitive overhead," Hansen said. Instead, APM products look for issues such as server latency, database errors, microservice errors, and capacity limits. Capacity applies to things like processors, storage, and containers, Hansen said.

Hansen suggested some important advice for newcomers to using application monitoring, such as not setting up too many alerts, and being sure to include your DevOps team to keep them in the loop. "With the switch to DevOps, we're seeing more and more teams where they're owning and running their own software, not just in preproduction but also in production," he added.

On the microservices front, it's important to use modern APM software where the product developers understand that the word "application" no longer just means files and libraries contained on one server. The microservices trend, and the cloud trend alike, is also leading toward an APM industry standard being developed.

"There's a lot of buzz about service mesh. Service mesh is a further abstraction of how to run your microservices," so you don't have to worry about scaling and security, Hansen noted.

"That's something that customers are starting to talk about and experiment with." Standards-wise, "I believe we're going to see some convergence around monitoring," he said.

SEE: Service mesh: What it is and why it matters so much now (ZDNet)

Current standards in the works from industry bodies include Trace Context, from the World Wide Web Consortium, and Open Tracing, from the Cloud Native Computing Foundation. Trace Context applies to how services share their context for monitoring, and Open Tracing is an API approach for instrumentation.

Also see

By Evan Koblentz

Evan became a technology reporter during the dot-com boom of the late 1990s. He published a book, "Abacus to smartphone: The evolution of mobile and portable computers" in 2015 and is executive director of Vintage Computer Federation, a 501(c)3 non-p...