Monitoring containers can be something of a challenge for developers, particularly when the containers are part of a complex microservice architecture.
On this new episode of The New Stack Makers podcast, Cisco Systems Chief Technology Officer of Cloud Native Platforms Ken Owens discussed with TNS editor Benjamin Ball the microservice landscape as a whole, and the overall complexity that a microservice-based infrastructure introduces. The interview was undertaken for our fifth EBook on Monitoring and Management with Docker and Containers.
The interview can also be heard on YouTube.
When automating and orchestrating containers at scale, it becomes difficult to identify and isolate problems within individual microservices. “To me, the level of complexity goes up when you talk about microservices versus individual containers because whether or not a container is up and easy to determine. What’s the impact of that container not being up is the microservices problem that’s very difficult to determine,” said Owens.
Collecting monitoring data from containers running microservices is also vastly different from collecting data from a traditional application. “I think the learning curve is pretty high in order to sort of understand,” said Owens, adding, “On one hand, it’s great that you can sort of automate and orchestrate this set of services in very unique, interesting, and high-performing ways. It sort of becomes a very complex problem to solve when something breaks, because it’s very difficult to identify and isolate those problems.”
“Definitely include your Ops team in the discussion, because they’re going to have a very specific point of view on what they need to be able to see and how it needs to fit into the existing dashboard,” Owens explained. “The last thing you want to do is take somebody’s job that’s already a difficult job, and make it even more complex and difficult, right?”
Ultimately, choosing whether to build or buy a commercial monitoring solution, or go with the many open source packages available, is still something many enterprises are discussing today. Owens offered a sage piece of advice for those on both sides of the fence to consider.
“Just because you can do something, it doesn’t necessarily mean that you should do something. I think that’s the advice that we should all consider very carefully when we make these choices because these are not single decision metrics. I think when you make these decisions, go into them knowing that if you’re going to do this yourself, you’re going to have to update this on a pretty regular basis.”
Cisco sponsored this podcast.
The New Stack is a wholly owned subsidiary of Insight Partners, an investor in the following companies mentioned in this article: Docker.