The Prometheus project, an open-source monitoring solution for application containers, has just received a major update that brings greater stability and improved performance to the platform.

The Cloud Native Computing Foundation, which houses the project, said the main updates in Prometheus 2.0 include a more efficient time-series database storage format, better handling of “stale” data in containers, and support for full database snapshot backups.

Prometheus monitors services such as containers by collecting metrics at specific intervals, comparing and evaluating that data against established rules, and then triggering an alarm should something go awry. With regard to containers, which are used to abstract software away from the hardware so it can be run anywhere, Prometheus monitors their status, the internal metrics of the application running inside, and the requests that flow through them. It also uses a query language so it can transform the metrics it collects into insights for developers.

The CNCF said container monitoring is essential as more organizations come to rely on the technology, but the process isn’t easy. Fabian Reinartz, a software engineer at CoreOS Inc. and a key member of the CNCF, said in a blog post today that older versions of Prometheus were being “strangled” by the overwhelming number of containers in usage. He said that at a scale of hundreds to thousands, it’s not uncommon to see millions of time series being tracked across a cluster.

“Prometheus has a simple and robust operational model that our users quickly learn to love,” Reinartz said. “Yet, the infrastructure space did not stand still, and projects like Kubernetes and Mesos rapidly change how software is being deployed and managed. Monitored environments have become increasingly more dynamic.”

Prometheus is a key component of CoreOS’s Tectonic Kubernetes platform. Other companies using Prometheus include CoreOS’s main Docker Inc., which develops the rival Docker container engine, and music streaming service provider SoundCloud Ltd., which first built the platform before open sourcing it.

According to Reinartz, tests of Prometheus 2.0 demonstrate that CPU usage is reduced by around 40 percent, while disk space usage has fallen by 50 percent from the previous 1.8 release.

“A modern monitoring toolkit is table stakes for organizations as they move to the cloud native paradigm; otherwise they operate blindly,” said CNCF Chief Operating Officer Chris Aniszczyk.

Image: CNCF