I've been building out the ossfind.com OSS alternatives directory, and observability tools come up constantly as a category where the SaaS-to-OSS migration question is genuinely interesting. Datadog is the canonical expensive monitoring platform; replacing it with open source isn't a simple one-for-one swap. I spent time researching three projects that together — or individually — cover a meaningful slice of what Datadog does.

These aren't toy tools. All three have serious GitHub star counts and production deployments. The question for an indie project isn't whether they're capable enough, it's whether they're operable enough given a small team and a tight budget.

Netdata: fastest time to value on this list

GitHub stars: ~79k. License: GPL-3.0.

Netdata installs with one command and immediately starts collecting ~800 pre-built metrics across your host, containers, and services. No config file required to get a working dashboard. That's a legitimately different experience from standing up Prometheus from scratch.

GPL-3.0 is clean for self-hosting. The copyleft only matters if you build a commercial product on top of Netdata and sell it. Running it internally doesn't trigger the requirement. The scope limit to understand: Netdata is infrastructure metrics and alerting. If distributed traces or centralized log search are the Datadog features you actually depend on, Netdata alone doesn't close that gap.

For an indie project running a few Vercel deployments and a VPS, Netdata on the VPS handles host-level observability without any setup cost. That's its actual sweet spot.

SigNoz: the all-in-one APM stack

GitHub stars: ~27k. License: see LICENSE (dual — a non-AGPL community edition and a paid enterprise tier).

SigNoz is the most ambitious of the three. It bundles infrastructure metrics, distributed traces (via OpenTelemetry), and log management in a single product with a single login. That's the Datadog feature set as a self-hosted stack.

The OpenTelemetry integration is worth noting: your instrumentation code sends standard OTLP data, and you could swap SigNoz for any compatible backend later. That's a better position than adopting a proprietary agent format.

The tradeoff is operational weight. SigNoz runs multiple containers — ClickHouse, query service, frontend. A basic deployment consumes meaningful memory. That's fine for a team with dedicated infrastructure, but for a solo project where the "server" is a $6 Hetzner box also running a database and three Astro sites, it's a lot to manage.

The license situation needs attention before any commercial deployment. The community edition is open-source. The enterprise tier is not. Check the LICENSE file rather than assuming one way or the other.

OpenObserve: the log-first approach

GitHub stars: ~19k. License: AGPL-3.0.

OpenObserve optimizes for storage efficiency on logs. The project claims significant storage savings compared to Elasticsearch-based setups for log data — that's its differentiator. It also supports traces and metrics, but log aggregation is where it's strongest.

AGPL-3.0 means the same copyleft as SigNoz's community license: use it internally without restriction, but if you expose it as a SaaS product, you need to open-source that service. For a directory site or a personal project, that's a non-issue.

The use case I'd reach for it: a project where log volume is high and storage cost is a real constraint. If you're paying for a logging SaaS and the bill is mostly driven by log data volume, OpenObserve is where I'd look first.

How I'd pick between them

For most indie projects in 2026, the honest answer is that full observability — metrics, traces, and logs — isn't the first priority. Getting something running cheaply is. Netdata solves the "is my server on fire" question immediately; that's usually enough to start.

The moment distributed tracing becomes useful is when you have multiple services talking to each other and you can't tell which one is slow. A single Astro site served from Vercel edge functions with a Turso database doesn't really need distributed tracing. A microservices setup does.

What I actually run

For my three sites, I use none of these currently. The sites are static Astro deployments on Vercel; Vercel's built-in analytics and the GitHub Actions run logs cover 90% of what I need right now. I'll revisit when I have a persistent service that generates actual log volume worth searching.

The research fed directly into the ossfind.com Datadog alternatives page, which lists these three plus Grafana and HyperDX with more detail on each.

---

Part of an ongoing 6-month experiment running three AI-curated directory sites. The technical claims here are real; this article was AI-assisted.