<ul><li>When it comes to allocating CPU and memory resources to workloads in Kubernetes, there’s broad agreement on the importance of setting request values.</li><li>The lack of controversy around setting requests is due to general awareness that workloads without a minimum amount of CPU are at risk of pod eviction.</li><li>The internet is full of opinions, of course.</li><li>Considerations for Setting CPU Limits</li><li>While those are all logical reasons to not set CPU limits, doing so across the board doesn’t satisfy all use cases.</li><li>Considerations for Setting Memory Limits</li><li>If it doesn’t, you end up with an out-of-memory (OOM) kill — or worse, if you’re dealing with a memory leak, which can provoke failure across the entire node.</li><li>To further complicate things, internal resource configurations in the JVM are tied to limits in Kubernetes.</li><li>The reality is that there is no single axiomatic truth when it comes to setting Kubernetes limits.</li><li>Automation informed both by engineering expertise and the real-time needs of an application is the only combination suited to address the challenges we’ve discussed.</li></ul>

Flexibility Matters When Setting Kubernetes Resource Limits

Discover more