Observability is a necessary side of operating cloud infrastructure at scale. It’s essential know that your sources are wholesome and performing as anticipated, and that your system is delivering the specified stage of efficiency to your clients.
A whole lot of challenges come up when monitoring container-based functions. First, as a result of container sources are transient and there are many metrics to look at, the monitoring information has strikingly excessive cardinality. In plain language which means that there are many distinctive values, which may make it tougher to outline a space-efficient storage mannequin and to create queries that return significant outcomes. Second, as a result of a well-architected container-based system consists utilizing numerous shifting components, ingesting, processing, and storing the monitoring information can grow to be an infrastructure problem of its personal.
Prometheus is a number one open-source monitoring answer with an lively developer and consumer group. It has a multi-dimensional information mannequin that could be a nice match for time sequence information collected from containers.
Introducing Amazon Managed Service for Prometheus (AMP)
As we speak we’re launching a preview of Amazon Managed Service for Prometheus (AMP). This fully-managed service is 100% suitable with Prometheus. It helps the identical metrics, the identical PromQL queries, and may make use of the 150+ Prometheus exporters. AMP runs throughout a number of Availability Zones for prime availability, and is powered by CNCF Cortex for horizontal scalability. AMP will simply scale to ingest, retailer, and question thousands and thousands of time sequence metrics.
The preview consists of assist for Amazon Elastic Kubernetes Service (EKS) and Amazon Elastic Container Service (ECS). It can be used to observe your self-managed Kubernetes clusters which might be operating within the cloud or on-premises.
Getting Began with Amazon Managed Service for Prometheus (AMP)
After becoming a member of the preview, I open the AMP Console, enter a reputation for my AMP workspace, and click on Create to get began (API and CLI assist can be out there):
My workspace is lively inside a minute or so. The console offers me with the endpoints that I can use to put in writing information to my workspace, and to challenge queries:
It additionally offers steerage on find out how to configure an current Prometheus server to ship metrics to the AMP workspace:
I may use AWS Distro for OpenTelemetry to scrape Prometheus metrics and ship them to my AMP workspace.
As soon as I’ve saved some metrics in my workspace, I can run PromQL queries and I can use Grafana to create dashboards and different visualizations. Right here’s a pattern Grafana dashboard:
Be part of the Preview
As famous earlier, we’re launching Amazon Managed Service for Prometheus (AMP) in preview kind and you’re welcome to strive it out at the moment.
We’ll have extra data (and a extra detailed weblog put up) at launch time.