Akamas Live Optimizations - Minimize Kubernetes cost while ensuring service quality

Akamas provides a single platform to optimize applications both in pre-production environment, for example to anticipate planned changes, cloud migration or chaos engineering scenarios, and directly live in production while applications are running under a dynamically varying workload.

In this video, we show how Akamas AI-powered, application-aware optimization can live-optimize a sample (Konacart) Java-based microservice running on Kubernetes.

The optimization scope includes both JVM and Kubernetes parameters, that is CPU and Memory limits as well as GC type, active processors and max heap size. The optimization goal is to minimize the cost of the container running the microservice, with the additional constraints that both response time and error rates need to stay within the defined SLOs.

Akamas decreases costs by applying different configurations over time, without any significant impact on the error rate. At the same time, Akamas takes into account the response time constraint when pursuing the cost reduction, automatically adjusting the container limits to ensure SLOs are matched.