PerfectScale Unfurls SaaS Edition of AI Platform to Control K8s Costs

PerfectScale this week launched a software-as-a-service (SaaS) edition of its namesake platform that employs machine learning algorithms to identify usage patterns that better enable IT teams to control Kubernetes costs.

Brendan Cooper, director of marketing for PerfectScale, says a SaaS edition makes the platform for optimizing Kubernetes environments much more accessible because organizations no longer need to dedicate a Kubernetes node to run the platform. Instead, IT teams can now log into a SaaS platform that surfaces optimization recommendations to better govern Kubernetes clusters in a way that reduces the total cost of deploying cloud-native applications.

In addition, the PerfectScale platform also enables IT teams to track optimization progress over an extended period of time, adds Cooper.

Kubernetes costs are spiraling because many organizations did not anticipate the impact constant updates to cloud-native applications would have on the consumption of Kubernetes resources, notes Cooper. As a result, it has become exceedingly difficult to predict CPU and memory consumption, he adds.

That lack of visibility is especially problematic in cloud computing environments where many organizations are finding the monthly consumption of Kubernetes infrastructure resources are far exceeding budget allocations, notes Cooper.

The overall goal is to leverage artificial intelligence (AI) in the form of machine learning algorithms to surface instances where, for example, Kubernetes clusters might be overprovisioned or are consistently running out of memory resources, he says.

There’s clearly a lot more sensitivity to Kubernetes costs during uncertain economic times, but Cooper notes that predictability is also crucial. Finance teams don’t want to be surprised by unexpected bills because there was an unexpected increase in consumption of cloud resources, he notes. The challenge is most IT teams don’t have access to a set of financial operations (FinOps) tools that make it easier to identify issues before those bills arrive, adds Cooper.

The PerfectScale platform also makes it easier to identify seasonal consumption of infrastructure resources that may be occurring because of a spike driven by, for example, a Black Friday sale occurring during a holiday season, he says.

Theoretically, of course, the ability to leverage Kubernetes to dynamically scale infrastructure resources up and down as needed is one of the primary reasons organizations are adopting the platform to deploy cloud-native applications. The issue they encounter is the rate of change in those environments. That can easily result in inefficiencies that conspire to increase costs despite Kubernetes’ orchestration capabilities. Machine learning algorithms promise to enable IT teams to identify and mitigate those issues in an IT environment that is otherwise too complex to manage.

It’s not clear whether IT teams will eventually depend more on machine learning algorithms and other forms of AI to manage cloud-native application environments. The one thing that is certain is that as the number of Kubernetes clusters deployed continues to increase, the level of expertise needed to manage and optimize those environments increases exponentially. The only thing left to determine going forward is how heavily IT teams will rely on AI tools to reduce the cognitive load currently required.