SYSTEM Cited by 1 source
Amazon Managed Service for Prometheus¶
Amazon Managed Service for Prometheus (AMP) is AWS's managed Prometheus-compatible metrics backend — scales horizontally, 150-day retention by default, PromQL-compatible query API. Stub page.
Seen in¶
- sources/2026-03-18-aws-ai-powered-event-response-for-amazon-eks — named as the metrics data source for AWS DevOps Agent investigations. One of the four canonical Agent-Space data sources (with CloudWatch Logs / X-Ray / EKS topology).
Adobe Firefly: managed high-cardinality GPU metrics¶
Adobe Firefly supplies AMP's first substantial wiki production case beyond its use as an AI-agent metrics source. Its EKS training clusters moved 2 million critical time series to AMP while retaining self-managed Prometheus. AMP managed scrapers collect the selected metric tier; reported GPU-utilisation query speedups were 3.5× over four hours, 22.6× over 12 hours, and 28.8× over 24 hours. The source treats the migration as incremental and names AMP's built-in high availability and managed operations as the operational rationale. (Source: sources/2026-08-13-aws-adobe-firefly-simplified-observability-with-amazon-managed-prometheus)
Related¶
- systems/aws-devops-agent — consumes AMP workspace endpoint per Agent Space.
- concepts/observability — the concept this service operationalises for metrics specifically.