SigNoz vs Datadog

Datadog is the observability platform that made "just install an agent and see everything" the industry default: infrastructure, APM, logs, and synthetics, all managed, all billed per host and per data type as you grow. SigNoz exists because that convenience has a recurring price tag, and it gets steeper the more you actually use the product. This is not a question of which tool has more features — Datadog does, by a wide margin. It is a question of who runs the observability stack, and who keeps the bill for running it.

What each one actually is

Datadog is a fully hosted SaaS. There is nothing to deploy beyond a lightweight agent on each host you want visibility into; the collection, storage, indexing, dashboards, and alerting all happen in Datadog's cloud. Coverage is genuinely broad — infrastructure monitoring, APM, log management, synthetic checks, and well over 800 integrations for databases, queues, cloud services, and language runtimes, most of it working within minutes of installing the agent. That breadth is the entire pitch: whatever you're running, Datadog probably already has a clean integration for it.

SigNoz is a self-hosted application you run yourself. It is OpenTelemetry-native — point OTLP exporters at it and metrics, logs, and traces show up correlated in one UI — backed by ClickHouse for storage and bundled with its own OpenTelemetry Collector. Its core is MIT licensed, though it is open-core: enterprise SSO and advanced RBAC sit behind a paid tier, while the observability functionality itself is fully open and unrestricted. It is explicitly built as the closest thing to a self-hosted Datadog equivalent — the same job, metrics plus logs plus traces correlated in one place, minus the vendor.

Pricing model: metered usage vs. a flat server bill

Datadog's pricing is per host and per data type, and it scales with how much you actually send it — more hosts, more log volume, more custom metrics, and more advanced modules like APM or synthetics each add their own line. Pricing starts from ~$15 per host per month and moves from there depending on data volume and which modules you turn on; we are not going to invent a specific bill size, because the honest answer is "it depends on your usage," and that unpredictability is itself part of the trade-off you're evaluating.

SigNoz replaces that metered model with a flat one: the cost of the VPS running it, full stop. Send it ten times the log volume and the bill does not move — only your disk and RAM headroom do. For a team whose Datadog invoice climbs every time they add instrumentation or retention, that flatness is the entire appeal, even before "own your data" enters the conversation.

Ops burden: managed vs. yours to run

This is where the trade reverses. With Datadog, the platform is Datadog's problem — no upgrades to schedule, no storage to size, no database to back up. Install the agent and the rest happens in their infrastructure.

With SigNoz, ClickHouse is a real dependency you now own: its own disk, its own memory headroom, its own backup story, separate from the SigNoz application containers themselves. SigNoz's documented 4 GB RAM floor is the honest cost of that ownership, and a fresh install measured ~501 MB idle on a GCP e2-standard-2 running Ubuntu 26.04 in August 2026 — a number that only grows once real traffic and retention are involved. Rating 4 out of 5 on install difficulty reflects a genuine multi-container stack, not a one-line agent install.

Data ownership and lock-in

Every metric, log line, and trace you send to Datadog lives in Datadog's cloud, under Datadog's retention rules and Datadog's export tooling. That is fine until it isn't — a pricing change, a retention policy you didn't choose, or simply wanting your historical data on infrastructure you control are all reasons self-hosters end up here. SigNoz keeps everything on infrastructure you run: the ClickHouse database backing it is yours to query, export, or migrate off of on your own terms, with no vendor sitting between you and your own telemetry.

Feature depth: where Datadog still wins outright

It would be dishonest to frame this as an even fight on features. Datadog's 800-plus integrations, its synthetic monitoring, its security and compliance tooling, and years of polish on alerting and incident workflows are not something a self-hosted OTel-native app matches out of the box. SigNoz covers the core APM workload well — metrics, logs, and traces correlated in one UI, with service maps and RED metrics — but it is one product built by one project, not a platform with a marketplace of integrations behind it. If your stack depends on deep, first-party visibility into a long tail of managed services, that is a real reason to stay on Datadog regardless of cost.

Switching costs, in both directions

Moving off Datadog is not free, and it is worth naming the friction rather than pretending otherwise. Dashboards, alert rules, and monitors built up over months on Datadog's UI don't port over automatically — instrumentation that leans on Datadog-specific agent integrations needs to be re-pointed at OTLP exporters instead, and any team habits built around Datadog's particular alerting or incident workflow have to be relearned somewhere else. That migration tax is real, and it's a big part of why teams that are already deep into Datadog stay there even when the bill stings.

The direction that's easier is starting fresh. If you haven't built years of Datadog tooling yet, standing up SigNoz from day one with OpenTelemetry instrumentation avoids the migration question entirely — OTLP is a vendor- neutral wire format, so the same instrumented services could, in principle, point at either backend later without a rewrite. That portability is one of OpenTelemetry's actual selling points, independent of which backend you choose today.

Which to choose

Choose Datadog if you want observability handled as a managed service — broad integration coverage, nothing to operate, and a bill you're willing to let scale with usage in exchange for zero ops. It is the stronger product on raw feature count and the safer default for a team that does not want to own another piece of infrastructure.

Choose SigNoz if the recurring per-host bill is the specific problem you're trying to solve, and your core need is metrics, logs, and traces correlated in one app rather than Datadog's full integration marketplace. You trade a metered invoice for a flat VPS bill and the responsibility of running ClickHouse yourself — a fair trade if your usage has grown to the point where the SaaS math no longer works, and a bad one if you need the breadth Datadog offers and don't yet have the appetite to operate a database.

Running SigNoz on a VPS

Self-hosting SigNoz means budgeting a real box, not a $5 instance — 4 GB of RAM is the documented floor, and comfortable retention pushes that higher. It installs via SigNoz's Foundry installer, which brings up the application, its bundled OpenTelemetry Collector, and ClickHouse together in one deploy step. Once it's running, treat ClickHouse like the database it is: give it disk headroom, watch it as retention grows, and back it up like you would any other production data store. If you're weighing SigNoz against other self-hosted options rather than against Datadog directly, our SigNoz vs. OpenObserve and SigNoz vs. Grafana comparisons cover the lighter and the more assembled alternatives, respectively.