This content is part of the Essential Guide: Use these DevOps examples to reimagine an IT organization

Sharper DevOps tools hack microservices monitoring

Microservices monitoring technology is only part of what enterprise IT shops say they need, as pricing changes are also necessary to accommodate the new architecture.

As IT pros adapt their apps to microservices, the tools that serve them must follow.

Customers are demanding vendors adjust products to accommodate new app development techniques, such as microservices, and vendors are responding with adaptations of products, such as AppDynamics Inc.'s Microservices iQ, which was released this week.

The AppDynamics portfolio already helped with root-cause analysis in a Java-based environment run by iJET International, but the risk management company in Annapolis, Md., needed better support -- and pricing -- for microservices monitoring.

The AppDynamics licensing model didn't really work for microservices, because it was based on an agent running in one monolithic server, with many apps deployed to it, said Nuno Pereira, CTO of iJET.

"I reached out to AppDynamics and said, 'We're moving into the microservices world, and this is just not going to work for us,'" Pereira said. "It'll be prohibitively expensive, and we'll have to make some choices about what we monitor and don't monitor. And, inevitably, we're going to have to go in a different direction if we don't resolve this problem."

AppDynamics' Microservices iQ introduces customized pricing for monitoring microservices specifically when apps are running on Java. The vendor considers a microservice for Java to be a Java virtual machine (JVM), with a heap size of less than 1 GB. For each of those JVMs, the license cost is one-fifth of the standard price of $3,300 per unit, per year. For other runtimes, when used in the context of microservices monitoring, AppDynamics is willing to work with the organization to make sure the pricing makes sense. 

"Within a couple of months, they came back and said they were working on the microservices pricing model," Pereira said. When it came to renewal time, "they had their whole act together around microservices."

Microservices iQ also offers some technical feature updates to accommodate microservices, such as a discovery engine that automatically builds a visual map of service endpoints, which can then be tracked by the monitoring process. Individual microservices can also be seen in isolation, so ops teams or other groups can manage them without having to look at the whole service architecture. The product will adapt with an elastic infrastructure to ensure no data is orphaned.

AppDynamics' thread-contention analysis can also trace whether individual threads of a Java app compete with one another for access to a database, and help resolve the contention so apps perform smoothly.

A lot of vendors are adding [service maps], but as you bring up microservices, you get queues; and having the ability to understand those queues is incredibly important.
Donnie Berkholzanalyst, 451 Research

This functionality differentiates Microservices iQ, according to Donnie Berkholz, analyst at 451 Research, from competitors such as New Relic, which also added microservices mapping in May and customizable pricing for various cloud instance sizes this month.*

"Service maps are a function a lot of vendors are adding at this point, but as you bring up microservices, you get queues; and having the ability to understand those queues is incredibly important," Berkholz said.

Pereira said his team hasn't gotten the chance to try out the thread-contention analysis. He said he would like to see AppDynamics continue to develop its product to support the Go programming language, which iJET is in the process of rolling out.

AppDynamics could also work to correlate and perform monitoring analysis across multiple data services, according to Berkholz, although AppDynamics has more than 140 connections to partners that do this kind of analysis already.

Still, Berkholz argued this kind of correlation is uncommon in tools with a data science bent that perform anomaly detection.

"Without it, you miss a lot of things that are problematic, but may be hiding in the noise," Berkholz said.

* Information added after publication.

Beth Pariseau is senior news writer for TechTarget's Data Center and Virtualization Media Group. Write to her at [email protected] or follow @PariseauTT on Twitter.

Next Steps

Monitoring tech isn't the problem: It's metrics

The right tool for container monitoring

The concepts underlying microservices management

AppDynamics' APM future in the Cisco portfolio

Dig Deeper on Real-Time Performance Monitoring and Management