Just about everybody in IT would agree that some degree of monitoring for containers will be required as more of them find their way into production environments. There’s much less agreement about how to go about doing that.
The Cloud Native Computing Foundation (CNCF) has been making a case an open-source Prometheus project that has been designed from the ground up to monitor containers. With the release of version 1.0 of Prometheus, the CNCF project now has reached a point where its application programming interface (API) is now stable.
Brian Brazil, founder of Robust Perception and core developer of Prometheus, says this is significant because organizations building container applications now can incorporate Prometheus without having to worry about whether some underling piece of functionality will change. Any and all changes to Prometheus in the future will be compatible with the existing API, he says.
Other additions provided within Prometheus 1.0 include the ability to explore events at scale and support for version 1.3 of Kubernetes orchestration framework.
At present, Prometheus consists of roughly 210.279 lines of code created by more than 700 contributors around the globe. Of course, there is no shortage of commercial IT monitoring tools already in use within many enterprises. But Brazil says contends that the scale at which containers soon will be deployed by enterprise IT organizations is well beyond anything involving virtual machines. As such, there’s a critical need for an IT monitoring tool specifically designed for containers that can be extended in almost any direction via a plug-in architecture.
Developed mainly using the Go programming language, a multidimensional data model is the heart of Prometheus that, when coupled with a dedicated query language, enables an IT organization to identify time series by both metric name and key/value pairs. That approach, Brazil said, makes it easier to identify what containers are associated with any give service level agreement (SLA) that has been put in place.
It will be interesting to see how much traction Prometheus ultimately gains in the enterprise. Contributors to the project include developers who work for SoundCloud, Digital Ocean, Ericsson, CoreOS, Weaveworks, Red Hat and Google. In that context, it’s clear that Prometheus is already gaining adherents among some of the leading IT vendors helping drive container adoption. The CNCF itself also counts IBM, Google, Docker, Cisco Systems, Red Hat and CoreOS among its governing board members.
In the meantime, IT organizations would be well-advised to revisit their DevOps processes before deploying containers in production environments en masse. Most IT organizations today already struggle with application release cycles in environments where it still takes days and weeks to fully provision a virtual machine. It’s clear that containers are helping to reduce the time it takes to not only provision IT infrastructure, but also to develop the application. As the rate at which container applications head into production environments is increasing, most existing DevOps processes more than likely will soon crack under the strain.