Using the machine learning capabilities of CrystalClear Time Series Analysis Engine, Federator.ai predicts application workload dynamics to scale containers/pods (replicas) and provides Just-in-Time Fitted allocation recommendations.
By leveraging application-aware insights into individual applications metrics and CPU/memory usage, Federator.ai improves performance (e.g., reducing latency in Kafka, lowering average response time & HTTP error rate in NGINX) while minimizing resource usage (e.g., reducing Kafka consumers, optimizing CPU & memory management).
Effective workload predictions
Focus on accurate indicators, like the message production rate of a Kafka topic, to generate precise workload predictions and ensure timely autoscaling for optimal performance.
Cost-effective application deployments
Integrate workload metrics, predictions, and application KPIs to determine the optimal number of replicas, enabling more cost-effective application deployments.
Achieving desired performance
Eliminate the need for manual threshold setting in Kubernetes native HPA by automatically optimizing resource usage to meet desired performance targets.
A presentation in Kafka Summit on why intelligent autoscaling is better than Kubernetes native HPA
A demo of how to configure an application on Federator.ai for autoscaling Kafka consumer
Philip Roberts
CEO of Cloudshape
