As IT organizations embrace Kubernetes to deploy containerized applications, they are discovering a need for tools to manage a new type of cluster that is proliferating rapidly across the enterprise. To address that issue, Scalyr has added support for Kubernetes clusters to its hosted log management service.
Scalyr CEO Steve Newman says IT teams can now monitor and troubleshoot multiple instances of Kubernetes cluster by centralizing logs. Regardless of the source, Scalyr is designed to ingest and parse logs to provide an application view of the overall IT environment. That capability is especially critical in IT environments where an application is now made up of large numbers of distributed microservices running on multiple Kubernetes clusters, he says.
A recent Scalyr survey found one-third of IT organizations are already delivering most of their applications as microservices. A full 71 percent of the software engineers surveyed are pushing code into production at least weekly and nearly one-third are doing so at least once per day.
Another new capability includes a stack trace linking that enables developers to jump from their log search directly to the reference source code residing in repositories such as GitHub and GitLab. That capability enables organizations to streamline debugging processes.
Finally, Scalyr has added chart annotations that can be employed to focus attention on specific logs or time periods that can then be shared using collaboration tools such as Slack, along with native integration with the CloudWatch log management service provided by Amazon Web Services (AWS). That latter capability is becoming more important as Kubernetes clusters increasingly are residing in public clouds alongside on-premises IT environments, notes Newman.
In fact, Kubernetes represents something of a double-edged sword for IT organizations. On the one side, Kubernetes enables organizations to achieve a level of hybrid cloud computing that previously was too difficult to embrace practically. For example, it will become easier for an on-premises application running on Kubernetes to invoke a serverless computing framework running on a public cloud to process workloads that need to consume compute resources for only a short period of time. On the other side, however, Kubernetes cluster sprawl soon may prove to be too much of a good thing.
Newman says IT operations teams still will need to monitor and manage legacy IT environments as well. That requirement necessitates the adoption of a hosted log management service capable of support both legacy platforms alongside emerging platforms such as Kubernetes, he says.
It’s already clear more mission-critical applications are headed for Kubernetes as the number of stateful applications built using containers increases. It may take a little longer for those applications to achieve a critical mass in production environments. But new approaches to log management will be needed as Kubernetes clusters are deployed not just on virtual machines but also bare-metal servers.
In the meantime, IT operations teams would be well-advised to start standing up Kubernetes clusters on their own just to learn how they behave. After all, provisioning something once is relatively easy. The hard part is dynamically provisioning Kubernetes hundreds of times over and making sure those Kubernetes clusters continue to run as intended.