Developers love containers. They’re a great way to turn fresh ideas into software at speed and catapult that software into the cloud. And they love Kubernetes, the open source platform for automating the deployment, scaling and management of containerized applications, so that developers’ time is freed up to turn even fresher ideas into yet more software.
Without effective monitoring, however, highly containerized environments can be extremely tricky to manage. Containers can be numerous. They are spun up and destroyed on a regular basis and at short order. It’s hard enough to track them, let alone diagnose the exact nature and cause of any problems with them that impact the end-user experience. Kubernetes itself only goes so far to help, which is giving rise to a brisk and fiercely competitive market in tools for monitoring Kubernetes environments.
At its Perform Summit event in Barcelona, application performance management (APM) company Dynatrace announced it has expanded its support for Kubernetes. As Steve Tack, senior vice president of product management at the company explained, this builds on work already done by the company to help customers automatically instrument containers and container payloads in Kubernetes environments and use its artificial intelligence engine Davis to analyse these environments:
Today, we are making Dynatrace even smarter, by bringing Kubernetes cluster and node health and utilization metrics and dashboards into our open platform.
Extended support for Kubernetes
In essence, Dynatrace’s OneAgent technology, deployed on nodes in a Kubernetes cluster to discover components and dependencies, already enables companies to monitor container-level metrics on applications and transactions in that environment.
With this expansion of its capabilities, Dynatrace adds to the mix integration with the Kubernetes API. This runs on the Master node in a Kubernetes cluster, serving as its control plane, explains Alois Mayr, technical product manager at Dynatrace. In this way, data can be collected on overall cluster health, performance and utilization and added to the data from OneAgents in the environment.
This combined information, in turn, is fed to Dynatrace’s AI engine, Davis, which continually learns what normal performance looks like in the Kubernetes environment and uses this as the basis for detecting and flagging up anomalies, suggesting root cause answers to issues and providing insight on how user experience is being impacted by them. This additional information is particularly important, says Mayr, when you consider that a Kubernetes cluster is typically shared across a company:
You may have different users, from different business units running their applications on the same cluster and so it’s possible to run into the situation where one business unit is deploying lots of different applications to this cluster. Eventually, the cluster could run out of resources. With this release, we can now measure how much additional resources are available on the cluster, because this helps operators understand when they need to scale the environment to still be able to deploy additional workload as needed.
Kubernetes’ ‘observability’ issue
Basically, this is Dynatrace’s latest move in a wider industry trend to increase what some refer to as the ‘observability’ of Kubernetes environments - an umbrella term that includes monitoring, logging and tracing.
In essence, Kubernetes provides various mechanisms to monitor its own components, but IT teams typically need a more comprehensive view of underlying infrastructure and of software interdependencies. And, in the view of executives at Dynatrace, this multi-cloud Kubernetes performance monitoring and availability should be integrated with overall application performance monitoring, via a single pane of glass. AI is an important part of the picture, as an essential ingredient in automating the detection and fix of issues, to free up IT teams’ time. As Tack put it, it’s about:
...bringing AI to cloud infrastructure management in its entirety - not just from a server or host perspective, but for all the components in that ecosystem that make up the fabric that delivers applications. We’ve delivered AI capabilities for many different layers of the stack - for public cloud providers, such as AWZ, Azure and the Google Cloud platform; for technologies that you still rely on such as applications running on System Z that underpin modern application front-ends; and extending into things such as cloud container monitoring.
Others are thinking the same way - not least Dynatrace’s rival New Relic, with its New Relic One platform, as diginomica reported last week. And this week, Barcelona is also playing host to KubeCon + CloudNativeCon, a biannual event for cloud-native computing enthusiasts. Kubernetes monitoring is high on the agenda there, too, with Google Cloud announcing the release into general availability of Stackdriver for monitoring and logging data from the Google Kubernetes Engine (GKE).
In total, the Cloud-Native Computing Forum, the open source software foundation that oversees Kubernetes among other projects, has identified some 48 organizations providing tools for monitoring Kubernetes environments. Along with Dynatrace, they include other APM leaders including New Relic and AppDynamics, smaller players such as Datadog and Honeybadger, and the CNCF’s own open-source monitoring tool, Prometheus.
In short, this is a busy market - but one in which multiple players, including Dynatrace, can expect to see plenty of action, as Kubernetes increasingly consolidates its position as the container orchestration platform of choice for increasingly cloud-inclined organizations on their digital transformation journeys.
As Felix Gratz, head of application performance management and systems architecture at automotive company Daimler AG, puts it:
Dynatrace works seamlessly with our Kubernetes environment to provide precise answers that help us to innovate faster. We adopted Kubernetes because it would help us accelerate time-to-market, and Dynatrace helps us to do just that.