#72
A
Rank #16
Prometheus
The de facto standard pull-based metrics collection and alerting system for cloud-native infrastructure. Standout trait is its pull model and PromQL query language, which decouple monitored services from the monitoring pipeline.
82.3
/ 100
⚙ Full Stack
Go binary with its own local time-series database (TSDB) on disk, a pull-based scrape engine reading target lists from static config or service discovery, and PromQL for querying; scales further via federation or remote-write to systems like VictoriaMetrics/Thanos.
📈 Scaling Analysis
Scales very well horizontally via federation and remote-write to long-term stores, and the pull model means adding new services never requires reconfiguring a central push endpoint. The main native limitation is that a single Prometheus server's local TSDB isn't clustered — true multi-node HA needs an external system like Thanos or VictoriaMetrics.
🚀 Running on Nexlayer
Single pod per scrape domain, mounting a volume for the TSDB; scrape targets inside the same namespace should be addressed as <podName>.pod:<port> in the scrape config rather than hardcoded IPs. No auth built in, so front it with the Nexlayer ingress/auth layer if exposed externally.