We are in the process of researching the approach of auto-scaling kubernetes pods with metric data stored in NewRelic. Our application code is reporting custom metrics with the NewRelic Java agent, and we would like to make use of it for scaling the pods in Kubernetes. I did go through the NewRelic documentation and user forums about Kubernetes integration, NewRelic Infrastructure Agent, and Prometheus integration. We could use the NewRelic Alert workflow to trigger a pipeline which scales the pods. The other method is using Kubernetes Metric Server to scale pods via Horizontal Pod Scaler. There exist a rudimentary/unofficial integration GitHub - FlexShopper/newrelic-custom-metrics. Other solutions like Prometheus, StackDriver, DataDog have an official supported integration with Kubernetes Metric Server, which allows defining scale conditions for Horizontal Pod AutoScaler using their metrics data.
What is your recommendation to auto-scale pods with NewRelic custom metrics?