Prometheus unbound: Open source cloud monitoring

Prometheus 1.0 is part of the Cloud Native Computing Foundation's effort to assemble a product portfolio for a container-based, open source cloud

Prometheus unbound: Open source cloud monitoring

Prometheus, an open source system for monitoring and alerting a wide spectrum of enterprise IT events, including containers, released its 1.0 revision this week.

It's also the second product in what amounts to a portfolio assembled by the CNCF (Cloud Native Computing Foundation) for realizing the promise of a container-powered cloud built entirely on open source and open standards.

Originally developed at music hosting site SoundCloud, Prometheus has been a work in progress over the past four years, but during that time it's enjoyed attention and uptake from major names. CoreOS, for instance, found Prometheus useful for improving the performance of Google's container scheduling system Kubernetes, and Kubernetes now integrates natively with Prometheus without needing a plugin.

Prometheus can monitor a broad range of resources: containers, application frameworks like Hadoop, language runtimes and application stacks like Python or Go. Data collected from such sources is stored in a time-series database developed specifically for the project, and metrics can in turn be exported to a host of targets for further analysis.

The 1.0 label mainly implies that the project is no longer built on shifting sands -- that the API set, the way an installation is configured, and the querying methodologies can now be built on top of it safely. It also means third parties can start making a concerted effort to distribute or support Prometheus as a product and add it to existing enterprise container ecosystems.

Also important about Prometheus is who's supporting it and to what end. Earlier this year, Prometheus was made a hosted project of the CNCF, right alongside Kubernetes. Together, both projects are a reflection of the CNCF's overall ambitions to foster an open source platform for running containers at scale in cloud environments.

Kubernetes has been a success in that respect, since most every major cloud platform for containers now uses it. Prometheus already has a following, in the sense that there's a collection of companies using and expanding on it.

If Prometheus is to complement Kubernetes, the likely next step is for it to enjoy commercial support in the broadest way: as an ingredient in commodity public clouds. Prometheus developers claim Google is already using the software internally for monitoring container loads. The next stage would be for, say, Google Cloud Platform customers to be able to put it to work -- not only on container loads, but everything else that might also be running within and alongside them.

Copyright © 2016 IDG Communications, Inc.