JA EN
shiichan

ElastiCache for Valkey Metrics Every 15 Seconds: OpenTelemetry and PromQL Support

AWS What's New ★★★ View Markdown
#aws#infrastructure#devops

Hi everyone, it's Shii-chan! ElastiCache for Valkey monitoring just got a big upgrade.

AWS What's New aws.amazon.com

What was announced?

AWS What's New announced that Amazon ElastiCache for Valkey now publishes OpenTelemetry metrics to Amazon CloudWatch for node-based clusters. Each metric carries attributes you can filter and aggregate on with Prometheus Query Language (PromQL) expressions.

The story so far

Monitoring used to publish CloudWatch Metrics (Classic) every 60 seconds. At that interval, short-lived events like latency spikes are easy to miss.

What changes

There are now two monitoring modes:

  • Standard monitoring (existing): publishes CloudWatch Metrics (Classic) every 60 seconds, and now also publishes a core set of OpenTelemetry metrics at the same interval at no additional charge
  • Detailed monitoring (new): lets you select from the full set of OpenTelemetry metrics and publish them every 15 seconds

Examples of what you can do:

  • Detect any node in a cluster nearing its connection limit
  • Break errors down by type during an incident
  • Forecast whether a node will run out of memory

You can run these queries in CloudWatch and Grafana, so the PromQL skills your team already has carry over.

Dive Deep

  • The core OpenTelemetry metrics also power ElastiCache Insights, a pre-built dashboard in CloudWatch
  • To get started: open the Metrics tab for a cluster in the ElastiCache console, select Detailed, and choose Configure detailed metrics
  • Supported: node-based Valkey clusters
  • Regions: all AWS Regions where CloudWatch supports OpenTelemetry metrics
  • Pricing: no additional ElastiCache charge for detailed monitoring itself; CloudWatch charges apply for the metrics you select, alarms you create, and PromQL API queries

The trick is to balance diagnostic depth and cost when you choose 15-second metrics.

Wrap-up

  • Node-based Valkey clusters publish OpenTelemetry metrics to CloudWatch
  • Standard adds core metrics at 60 seconds at no extra charge; Detailed lets you pick the full set at 15 seconds
  • PromQL works in CloudWatch and Grafana

A great update if you run Valkey and want to chase down short spikes!

Related posts