The reason why we use the cloud so much is the bottom line: Saving money. Well, that’s the theory anyway. StormForge, a start-up specializing in reducing cloud waste with machine learning (ML) and artificial intelligence (AI) has found in its recent survey that businesses waste over $17-billion a year on unused or idle cloud resources. That’s serious money.
Now, it’s not that companies have an unrealistic view of what they’re going to be spending. Ninety-four percent say they know, at least roughly, what their cloud spend will be each month. That’s the good news. The bad news is they also estimate that nearly half of their cloud spend is wasted on unused or idle resources. That’s no way to make friends and influence others in your company’s accounting department.
Unsurprisingly, survey respondents said that reducing cloud waste is a priority, with 33% saying it’s a very high priority and another 43% saying that, while not the highest priority, it is still important. It had better be or the CFO may end up recommending the CIO and IT staff find new jobs elsewhere.
The two most significant causes of cloud waste are cloud complexity, which makes it hard to estimate the resources that are actually needed and intentionally over-provisioning. The idea, of course, for the latter is that over-provisioning is a safety net to ensure application performance. In particular, the container orchestrator Kubernetes is a significant contributor to the cloud complexity issue, with 62% agreeing that it is a major or contributing factor.
In most organizations, 55%, IT Ops or Cloud Ops teams are responsible for deciding how Kubernetes is to be deployed while dev and engineering teams are responsible for 29% of companies. Regardless of who makes the call, no one seems to be particularly good at deploying or managing Kubernetes.
This is understandable. Mastering Kubernetes is in no way, shape, or form easy. When you deploy an app on Kubernetes, you must make many decisions on resource allocation including memory requests and limits, CPU requests and limits, and replicas. Add to that the app-specific parameter settings like Java Virtual Machine (JVM) heap size and garbage collection, and multiply that by the number of containers, and you quickly have a highly complex, multi-dimensional optimization problem. Failing to manage it correctly impacts both the cost of running the app and its performance and reliability.
StormForge, since it’s their business, of course, recommends its ML and AI business tools to set up and optimize your cloud-native Kubernetes-based applications. There are very few Kubernetes wizards available. If you’re having trouble with your Kubernetes setup, I’d give StormForge a call. It may be just what you need to cut your cloud costs and save your job at the same time.
