Fair winds ahead, StormForge ‘bi-dimensional’ Kubernetes pod autoscaling

StormForge is a cloud-native application performance testing and resource optimization company.

The company has used the week of KubeCon and CloudNativeCon North America this year to detail its StormForge Optimize Live product.

This is said to be the industry’s first bi-dimensional (i.e. both vertical and horizontal) Kubernetes pod autoscaling tool. The enhanced capabilities are available now.

StormForge Optimize Live uses machine learning to automatically right-size pods while also setting a desired target utilization level for the horizontal pod autoscaler (HPA). 

This enables vertical and horizontal autoscaling to work together without contention, to minimise resource usage and cost without sacrificing application performance or reliability.

“The promise of Kubernetes is still beyond the reach of so many organisations, but bi-dimensional pod autoscaling, now made possible by StormForge Optimize Live, will change that for the vast majority of Kubernetes users,” said Matt Provo, CEO at StormForge. “We’re excited to offer yet another industry-first innovation from StormForge, which we expect will unleash innovation in enterprises across the globe through application performance improvements and cloud cost savings.”

Autoscaling said to be is both a significant challenge and opportunity for Kubernetes users. 

Kube-pod autoscalers

Kubernetes natively offers two primary ways of dynamically scaling applications – the HPA and the vertical pod autoscaler (VPA) – but it’s not possible to use both together without extensive customisation effort. Additionally, the HPA requires users to set a target utilisation that determines when to add or remove replicas, but it’s nearly impossible to arrive at an optimal target utilisation using manual methods.

While many organisations moved to containers and Kubernetes with the promise of improved efficiency and cost savings, the reality has been that costs have risen dramatically while scaling up for ‘day 2’ operations. 

Autoscaling holds great promise for improving efficiency, especially as application usage fluctuates, but organisations have been largely unable to benefit from autoscaling advantages until now.

“Despite using the HPA, Kubernetes over-provisioning is still a significant problem, driving an unsustainable growth in cloud costs,” said Chad Upton, VP of infrastructure engineering with Firstup. “Our work with the [StormForge] solution so far has shown promise for significantly reducing cloud resource costs while still ensuring application performance and reliability.”

With the new release of StormForge Optimize Live, users are no longer limited to using just horizontal scaling without proper tuning, a process which inevitably ends up leaving substantial cloud resource savings untapped.

Right-sizing Kubernetes apps

By enabling bi-dimensional Kubernetes pod autoscaling, StormForge Optimize Live ensures continuous right-sizing of Kubernetes applications; reduces Kubernetes application resource usage and costs while still ensuring performance and reliability; and reduces the risk of out-of-memory (OOM) errors and CPU throttling while minimising cloud costs.

StormForge Optimize Live was introduced in February of this year and delivers ML-powered Kubernetes resource optimisation through analysis of observability data. The ML focus goes beyond cost or performance alone, optimising both to enable intelligent business decisions with minimal trade-offs.

Purpose-built for Kubernetes, StormForge runs in any CNCF-certified distribution. 

StormForge: Overcome these tough Kubernetes tradeoff choices.

Data Center
Data Management