Fairwinds, a provider of managed IT services for Kubernetes environments, has launched an open source project dubbed Astro that provides an easier application programming interface (API) for monitoring Kubernetes clusters using monitoring tools from Datadog as well as tools that make it easier to navigate Datadog.
Company CTO E.J. Etherington says Astro was created so organizations that have embraced Kubernetes can monitor these platforms using the same tool they already employ to monitor existing monolithic applications.
One of the challenges Datadog customers routinely encounter, Etherington says, is that each monitor typically is deployed manually. Astro automates the management of Datadog monitors using rules based on defined patterns in Kubernetes to uncover potential performance issues.
Specifically, Astro manages all monitors within a Kubernetes namespace and makes certain Datadog monitors are updated continuously to reflect any changes to a Kubernetes object. Any data from a managed Kubernetes object also can be inserted into a monitor managed by Astro to create more contextualized alerts to make it easier to triage issues. Etherington says the goal is to eliminate the need for a site reliability engineer (SRE) to manually manage the processes associated with integrating Kubernetes with the Datadog monitoring service.
There is, of course, no shortage of tools available when it comes to monitoring Kubernetes. One of the most widely employed is Prometheus, an open source project that, like Kubernetes, is being developed under the auspices of the Cloud Native Computing Foundation (CNCF). However, while developers tend to favor Prometheus when building cloud-native applications, many IT operations teams have already standardized on monitoring platforms such as Datadog for applications running in production environments. In fact, Datadog can consume a wide variety of metrics generated by a wide variety of open source and commercial software sources, Etherington notes.
Fairwinds itself is no stranger to open source software. It launched the Polaris open source configuration management tool for Kubernetes and the Goldilocks resource management project to optimize the consumption of memory by Kubernetes projects. All three projects were created to address issues Fairwinds encounters as a provider of a managed Kubernetes service.
It’s unclear whether the CNCF will address any of these issues within the context of an existing project or one the foundation might choose to incubate. However, as more instances of Kubernetes are deployed in production environments, it’s clear there is a growing need for tools that enable IT organizations to operationalize Kubernetes at scale.
Regardless of the platform chosen, the one thing that is certain is organizations that run applications based on microservices on Kubernetes will need to rely on monitoring tools to achieve observability in keeping with best DevOps practices. IT organizations, given the costs involved, have tended to limit the use of monitoring tools to their most mission-critical monolithic applications. However, all the dependencies that exist between the microservices that make up a cloud-native application make monitoring all but a necessity. The challenge now is determining how to provide those capabilities in a way that doesn’t give the IT operations team a case of whiplash while swiveling between screens.