Praveen Kashimsetty, director of product management for Rafay Systems, says that capability eliminates the need to deploy a separate tool to manage costs at a time when more organizations than ever are sensitive to IT costs during an economic downturn.
The Kubernetes Operations Platform (KOP) enables IT teams to manage Kubernetes clusters running in the cloud or in an on-premises IT environment. The Cost Management Service makes it possible to either “showback” costs or implement actual chargebacks to business units in real-time for both dedicated and shared clusters.
While Kubernetes provides a more efficient way to run applications, there is often a tendency to dedicate a cluster to each application to ensure availability. On a shared cluster, there may also be one application that is consuming more resources than other applications sharing the same cluster.
In addition, IT teams often create development environments that will still consume infrastructure resources when no longer in use.
Finally, in some environments, there may simply be no need to run Kubernetes clusters—on weekends, for example—when no one is accessing applications, notes Kashimsetty.
Regardless of the cause, it’s apparent there is plenty of opportunity to reduce costs by more efficiently using existing clusters or, in some cases, consolidating clusters, he adds. The challenge is that IT teams need more application performance context, provided via KOP, to make a truly informed decision about how to best allocate infrastructure resources, said Kashimsetty. Many of those IT teams now manage IT infrastructure much like an internal cloud service provider that is constantly seeking to optimize consumption of infrastructure resources, he adds.
In the longer term, Rafay Systems is also working toward adding predictive analytics capabilities to the Cost Management Service that would enable IT teams to forecast costs, he says. At the same time, Rafay Systems is working toward adding machine learning algorithms that would enable IT teams to apply artificial intelligence (AI) to IT operations (AIOps), notes Kashimsetty.
It’s not clear to what degree IT organizations might be already wrestling with Kubernetes sprawl as more cloud-native applications are deployed across what are becoming fleets of clusters running everywhere from the cloud to the network edge. The one thing that is certain is finance teams are requiring IT organizations to contain costs by finding ways to manage Kubernetes clusters more efficiently.
Of course, not every Kubernetes cluster is managed by a centralized IT team. There is no shortage of individual development teams in business units that have been spinning up multiple Kubernetes clusters on cloud platforms for several years now with little regard to cost. As more organizations embrace financial operations (FinOps) concepts to rein in those costs, it’s probable that centralized IT teams will be exerting more control over those Kubernetes clusters.
In the meantime, the pace at which Kubernetes is being adopted in the enterprise continues to accelerate. The issue now is striking a balance between the desire to have unfettered access to compute resources and the need for more adult supervision of the IT budget.