The open-source Docker application container virtualization technology is becoming increasingly popular, spawning a need for distributed monitoring capabilities. To help organizations understand what tools are available for container monitoring, Docker last week announced a new Ecosystem Technology Partner (ETP) program that is starting off with six vendors that have integrated with Docker for monitoring. Those vendors include AppDynamics, Datadog, New Relic, Scout, SignalFx and Sysdig.
The vendors did not need to pay Docker Inc., the lead commercial vendor behind the container virtualization technology with the same name, to become part of the ETP either.
The reason the monitoring vendors decided to integrate with Docker? It all has to do with demand.
“We’re users of Docker, and so are many of our customers,” Karthik Rau, CEO of SignalFx told eWEEK. “It was natural to provide a way for everyone to use SignalFx to monitor, run streaming analytics with, and alert against metrics from the Docker Stats API.”
Abner Germanow, senior director, solutions marketing at New Relic, said his company sees significant value in its ongoing collaboration with Docker through ETP, which helps New Relic add Docker monitoring capabilities to meet the needs of the Docker community.
The Docker 1.5 release, which debuted in February, introduced the Docker Stats API, which enables vendors to plug into the Docker engine to gain insight into container activities. The API alone, however, isn’t a full solution, which is where the partner technologies come into the picture.
“The Docker API is a great starting point, but Sysdig Cloud integrates with Docker in a very unique way,” Loris Degioanni, CEO of Sysdig, told eWEEK.
Degioanni explained that by using open-source Sysdig technology, the Sysdig Cloud agent is able to get full visibility into all activity inside every container, without having to instrument the containers themselves in any way. He added that the Sysdig Cloud technology can monitor basic system resource use for each container (CPU, memory, etc.), as well as advanced information like application data, network traffic, topology and custom metrics.
New Relic is using Linux server monitor and application agent data in addition to the data coming from the Docker API, Germanow said.
“If your app is running slowly, we make it super-easy to determine if the root cause of that problem is container performance,” Germanow said. “This is essential because companies can create hundreds of Docker images every day. By putting Docker performance into the context of an app, we’ve made Docker monitoring an actionable app-centric data view.”
SignalFx’s Rau said that in Docker environments, it’s important to look at container-specific metrics such as CPU or memory used by an individual container, rather than looking at the metrics of the host as a whole. “With the new Docker Stats API, we can now provide this information to our customers,” Rau said. “But more importantly, we enable our customers to correlate those container-specific metrics with other metrics.”
The other metrics can be used to help answer questions about memory and host utilization as well as application performance latency.
“Docker provided a complete set of metrics with the Stats API,” Rau said. “Some of the things we’re focused on are adding more analytics capabilities as well as providing more means of correlating container metrics versus infrastructure metrics versus application metrics and business metrics.”
What’s driving the demand for monitoring tools is the fact that Docker is being used today in production. Degioanni said he is seeing Docker used in production though he noted that he’s hearing from many organizations that are using Docker in test or development environments, but are unable to deploy containers in production because of limited visibility.
Germanow said the need and demand for Docker container monitoring is real today. “I can’t think of another technology that has gone from unknown to proof-of-concept to production faster in IT infrastructure in a diverse set of customer sizes, industries and use cases,” Germanow said.
Yet there is a lot of confusion around Docker monitoring, and people want to understand the behavior of their applications and the impact those apps have on their infrastructure, Germanow said. The abstraction layer containers provide is great for portability and creating new application architectures, but it also affects the visibility that software teams are used to having from the server through the app, he said.
“Metrics on individual containers aren’t useful unless they help shed light on the behavior of the application,” Germanow said. “However, metrics on how container types behave when matched with the apps they support help people make infrastructure and app decisions.”
Sean Michael Kerner is a senior editor at eWEEK and InternetNews.com. Follow him on Twitter @TechJournalist.
