You Asked, We Answered: Patrick Bergstrom, StormForge CTO, on why the time for optimization is now

Even the most experienced Kubernetes performance tuners can’t compete with automation and machine learning, and it’s unrealistic to expect that any business will be able to put enough human resources against the problem. It’s impractical to approach optimization manually, and it certainly isn’t cost-effective.

I’ve seen this scenario first-hand when leading site reliability engineering groups that support application teams. I’d ask myself if we’d rather have people innovating, supporting the business, and helping it get where it needs to go or did we want them to spend hundreds of hours trying to tune apps manually? In my entire career, I’ve never met a developer who gets excited about manually tuning their application. It’s often a no-win situation, and given a choice as to where to prioritize your engineering talent, the smart option is to focus on engineering-driven solutions and not potentially spend more on compute resources. Why? Because your app is perfectly tuned – automatically.

With the automation StormForge provides, we can run, analyze, and iterate on dozens, or even hundreds, of performance tests every night. I’ve found that most organizations run maybe one test a week, so at that rate, we’re talking about three-plus years worth of performance tests that StormForge can do in a single night. And we’re also able to leverage our machine learning capabilities to tweak settings that most humans don’t even consider changing – and observe how these changes affect the application’s performance. It’s not just optimization; it’s real-time decision support.

StormForge recognized early on that developers don’t know what they don’t know, and they didn’t have the tools to help them find out what they don’t know. We set out to solve that gap by looking at the Kubernetes challenge holistically – from optimization in non-production environments to continuous optimization in production using observability data.

Enterprises had been putting off making their applications more efficient, simply due to their inability to truly address the problem by hand. Fortunately, we now have tools that empower us to be efficient in an automated fashion. This is a critical step forward, as developers and platform engineers strive to make better decisions about application resources to deliver business value and reduce waste. The problem has grown to the point where there’s simply too much at stake to wait any longer. Whether it’s in support of your company’s cloud-efficiency pledge or simply to reduce the cost of compute overhead, the time to optimize is now.