Monitoring

Grafana72.8k★

Grafana is the open-source visualization layer that makes your metrics, logs, and traces actually readable. 72k stars, 200+ data source plugins, and dashboards so good that even Datadog users sometimes pipe data into Grafana instead. It's the "build your own observability" foundation. Datadog is the all-in-one SaaS alternative — install an agent, get dashboards in minutes, pay $15-23/host/month and watch costs scale linearly with your infrastructure. For a solo founder, Grafana + Prometheus + Loki is free but requires assembly. For a funded startup, Datadog's simplicity might be worth the cost. Use Grafana if you want full control over your monitoring stack and aren't afraid of running Prometheus, Loki, and Tempo alongside it. The Grafana Cloud free tier is generous enough for small projects. The catch: AGPLv3 licensing may conflict with some enterprise policies. Grafana visualizes data but doesn't collect it — you need separate tools for metrics, logs, and traces. And "assembly required" means real operational overhead: configuring alerting rules, managing retention, and debugging query performance across multiple backends.

Prometheus63.3k★

Prometheus is the de facto standard for metrics collection in cloud-native infrastructure. Pull-based model, PromQL query language, and tight Kubernetes integration make it the monitoring backbone for most modern stacks. At 63k stars, it's what Grafana dashboards are usually wired to. VictoriaMetrics is the drop-in replacement that outperforms it — 20x better compression, lower RAM usage, and deploys as a single binary instead of Prometheus's growing component list. Grafana Mimir scales horizontally for multi-tenant enterprise setups. Datadog does it all as SaaS if you're willing to pay. Use Prometheus if you're running Kubernetes and want the ecosystem that everything else integrates with — alertmanager, exporters, and service discovery all assume Prometheus. The catch: Prometheus is designed for short-term metrics on a single node. Long-term storage, high availability, and horizontal scaling all require bolting on Thanos, Cortex, or VictoriaMetrics. PromQL has a learning curve. And at high cardinality (millions of unique label combinations), performance degrades fast. If you're hitting these limits, VictoriaMetrics is the pragmatic upgrade.

SigNoz26.3k★

SigNoz is the open-source Datadog. Logs, traces, and metrics in a single application, built natively on OpenTelemetry with ClickHouse (the same storage Uber and Cloudflare use) under the hood. No vendor lock-in, no surprise bills, no sending your data to someone else's cloud. Having all three observability pillars in one tool makes correlation trivial — click a slow trace, see the related logs, check the metrics. Compared to Datadog (powerful but expensive), SigNoz gives you 80% of the features at 0% of the cost if self-hosted. Compared to Grafana + Loki + Tempo (more flexible, more moving parts), SigNoz is simpler to operate. Compared to Jaeger (traces only), SigNoz does everything. Use this when you need full observability on a budget and want to own your data. Skip this if you need Datadog's APM depth or have a team that already knows the Grafana stack. The catch: self-hosting ClickHouse at scale requires real ops expertise. The managed offering exists but the open-source version is the draw. And OpenTelemetry instrumentation still has rough edges in some languages.

Highlight9.2k★

Highlight was the open-source Sentry alternative that bundled error monitoring, session replay, logging, and tracing in one platform. The full-stack observability story was compelling — see the user's session, the error, and the server logs in one view. Was. Highlight was acquired by LaunchDarkly and shut down its standalone service in February 2026. If you're looking at it today, you're looking at a dead product. Sentry is the dominant error monitoring platform. PostHog combines session replay with analytics. OpenTelemetry plus Grafana is the DIY full-stack monitoring stack. Don't adopt Highlight for new projects. The code is open-source, but without active development and hosting, you'd be maintaining a complex observability platform solo. The catch: the catch is that Highlight no longer exists as a standalone product. It lives on inside LaunchDarkly's observability features. For self-hosted open-source monitoring, look at SigNoz or Grafana's stack instead.

Uptrace4.1k★

Uptrace is the self-hosted observability platform built entirely on OpenTelemetry. Traces, metrics, and logs in one UI, stored in ClickHouse for fast queries. Think Datadog-lite that you own and control. If you want OpenTelemetry-native observability without sending data to a SaaS vendor, Uptrace is one of the few complete options. Jaeger handles tracing only. Grafana + Tempo + Loki is more flexible but requires assembling multiple components. SigNoz is the direct competitor — similar scope, more community traction. Datadog is the commercial benchmark but costs a fortune. Best for teams already using OpenTelemetry who want a unified, self-hosted dashboard. The ClickHouse backend means queries stay fast even with high cardinality data. The catch: it's AGPL-3.0, so you can't build a competing service on it. The community is small (~4K stars) compared to SigNoz or Grafana stack. Documentation has gaps. And self-hosting means you're operating ClickHouse + Uptrace + an OTel collector — that's real infrastructure to maintain.

pyrra1.5k★

Pyrra turns SLO management with Prometheus from a YAML nightmare into something humans can actually use. Write a high-level SLO definition, and Pyrra auto-generates the Prometheus recording rules, alerting rules, and error budget dashboards. One binary, one web UI, and your SLOs are managed without building custom Grafana dashboards from scratch. If you're running Prometheus and need SLOs without hand-crafting recording rules, Pyrra is the fastest path. Sloth is the similar alternative that generates Prometheus rules from specs but with less active development. Google's SLO Generator works for multi-backend setups. Datadog and Nobl9 are the commercial SLO platforms. The catch: Pyrra is tightly coupled to Prometheus — if your observability stack uses Datadog or Grafana Mimir natively, Pyrra doesn't help. The project has 1,400 stars and a small community, so edge cases may require reading source code. And SLOs only matter if your team actually acts on error budget burns — Pyrra gives you the data, but the organizational discipline is on you.