Based on the application-aware insight that digs into the metrics of individual applications and CPU/memory usage, Federator.ai helps users achieve much better performance (Kafka: reduce latency; NGINX: reduce average response time & HTTP response error rate ) with much fewer resources (Kafka: reduced number of Kafka consumers; Generic: CPU & memory management).
By using machine learning technologies of CrystalClear Time Series Analysis Engine, Federator.ai scales the number of containers/pods (replicas) based on predictions capturing the dynamics of application workloads to meet the resource demands by providing Just-in-Time Fitted allocation recommendations.
Federator.ai focuses on more accurate indications to reflect real application workloads, such as message production rate of a Kafka topic, so that accurate predictions can be produced and desired performance can be accomplished by autoscaling in time.
Federator.ai integrates the workload metrics, workload predictions, and application KPI in deciding the right number of replicas and achieves more cost-effective application deployments.
Without guessing or experimenting on what metric threshold to set in Kubernetes native HPA, Federator.ai achieves better use of resources for desired performance automatically.