Datadog vs. SigNoz: OpenTelemetry, operations, and migration costs

A rising Datadog bill often sends a team looking at SigNoz. The first question is usually how much an open-source alternative could save. The more consequential question is whether the investigation that resolved the last incident will still work after the move.

SigNoz deserves a pilot when the main requirements are application traces, metrics, and logs, the team already uses OpenTelemetry, and it can manage collection and data conventions. If existing investigations depend on several Datadog products—APM, user experience, synthetic testing, and security, for example—establish the replacement scope before deciding how much to migrate. SigNoz has a managed service, and Datadog supports OpenTelemetry. This is not simply self-hosted open source versus a proprietary Agent.

Guance Editorial Team · Reviewed October 21, 2026. This documentation-based comparison covers Datadog SaaS, SigNoz Cloud Teams, and self-hosted Community, with enterprise options identified separately. We did not run a performance or cost benchmark.

Compare the actual editions

Consult the SigNoz edition matrix, enterprise operating models, and Datadog OTel compatibility matrix. A capability listed here is not necessarily included in every plan.

What OpenTelemetry makes easier—and what it leaves behind

SigNoz's collection architecture uses OpenTelemetry, with telemetry flowing through a Collector into ClickHouse. OTLP is not its only possible input: the architecture documentation also lists Jaeger, Zipkin, and Kafka, among others. Actual support depends on the receivers and configuration deployed. ClickHouse is the telemetry store; a unified backend does not mean a production installation consists of one database process. SigNoz architecture

Datadog accepts OpenTelemetry data too. DDOT, upstream Collector, and direct OTLP paths differ in supported features. Check the matrix against your language, signals, and products before choosing Datadog SDKs, OTel, or a staged combination. Seeing traces arrive does not establish parity for profiling, application security, or service metrics.

OTel can reduce the degree to which application instrumentation and export are tied to one backend. It does not automatically migrate dashboard queries, alert rules, authorization, historical data, or on-call habits. Less reinstrumentation is a reasonable expectation. A cost-free switch with no vendor dependencies is not.

Follow one failed order through both platforms

Imagine an order endpoint that occasionally times out. The database query is quick, but a downstream call retries repeatedly. Do not stop the pilot when both products display a service name. Select the same time window, environment, and service version, find a slow request, inspect the retry spans, and follow them to logs and resource context.

SigNoz correlates logs using trace_id and span_id. Supported OTel logging integrations can attach those identifiers within an active span; file-based or Kubernetes logs may still need parsing and field mapping. A request ID printed somewhere in a message is not automatically a recognized trace field. Trace/log correlation and field parsing

Datadog needs correct log/trace correlation as well. Check service and environment tags, trace identifiers, and sampling. Different request counts can reflect sampling or aggregation rather than missing ingestion. Latency percentiles are only comparable when the time windows, request populations, and filters agree. Datadog correlation

Then hand the investigation to someone who actually takes the pager. Can they find the entry point, reach the relevant logs, and understand the service context in an alert? Those answers are more useful than counting default dashboard charts.

Self-hosting changes responsibility, not just the recipient of the bill

SigNoz Community can be self-hosted. The main repository's license distinguishes ordinary code from enterprise code under ee/, while third-party components retain their own licenses. Do not describe every part of the product as covered by one open-source license. Repository license

Operating it means planning for ingestion peaks, concurrent queries, disks and retention, database backups, upgrades, and recovery. A successful single-machine installation proves the collection path works; it does not establish production capacity. Test your log volume, span sizes, metric cardinality, and query load rather than assuming a vaguely defined medium-sized workload fits one machine. Self-hosted installation

SigNoz Cloud is an option for teams that do not want to run the backend. Enterprise BYOC puts a vendor-managed deployment in the customer's cloud account; enterprise self-hosted deployments are customer-operated with the contracted support. Those responsibilities are different.

Check authentication, roles, and audit requirements individually. “All SSO requires enterprise” is too broad to substitute for checking the actual identity method and edition. Features marked as coming soon on a pricing page should not be treated as available today.

Calculate with your data

At the review date, SigNoz Cloud Teams starts at $49 per month, including $49 of usage. The default pricing display lists logs and traces at $0.30 per ingested GB with 15-day retention, and metrics at $0.10 per million samples with one-month retention. Longer retention changes the calculation. Enterprise and self-hosted deployments do not use this same managed-service bill. SigNoz pricing

No host-based charge does not mean cardinality is free. Each active time series emits samples at its export interval. With 10,000 series reporting every 30 seconds over 30 days, the total is 864 million samples: $86.40 of monthly metric usage at that rate, before logs and traces. This is not a full $86.40 added on top of the $49 included usage. Calculate aggregate usage, then apply the minimum and allowance correctly. Metric billing

Datadog estimates should include the products actually required. Host-based APM starts at $31 per host per month on annual billing when paired with Infrastructure Monitoring; standalone APM starts at $36 per host per month. Logs, RUM, other modules, and usage beyond included allowances may add charges. Comparing one SigNoz ingestion line with a complete Datadog bill is not a like-for-like estimate. Datadog pricing

Self-hosted estimates need compute, disks, backups, networking, and operations. Reducing noisy telemetry, controlling cardinality, and setting appropriate sampling and retention are useful before changing platforms. This article has not measured savings and does not adopt a vendor's advertised savings percentage.

Migrate one service first

Choose a real service with a manageable scope. Keep the existing path while sending the necessary telemetry to the candidate. Verify metric units and aggregation, resource attributes, log timestamps, and trace continuity before rebuilding the most important alerts. Watch Collector queues, retries, and network overhead during dual delivery; an open protocol does not make collection free.

Next, inventory the special workflows: browser replay, synthetic tests, security detection, profiling, and third-party integrations. SigNoz currently documents frontend Web Vitals and mobile monitoring, so “no RUM” is not a useful blanket claim. Those features also do not establish parity with every Datadog replay, synthetic, or security product. Accept the migration against the functions your team uses, not a full-stack label.

Validate historical-query requirements, permissions, and alert routing before stopping the old path. Moving a subset of workloads may be the right outcome; architectural tidiness is not a reason to replace every tool at once.

Which option fits?

Pilot SigNoz Cloud when logs, metrics, and traces are the core requirement, OTel is already in place, usage-based budgeting is attractive, and the team does not want backend operations work.

Pilot self-hosted SigNoz when deployment and storage control are concrete requirements and the team can own capacity, backups, upgrades, and availability. Confirm Community and paid-feature boundaries first.

Keep or choose Datadog when existing cross-product investigations, integrations, and organizational workflows have demonstrated value, and the team wants a vendor-operated backend while managing module and data spend. SaaS removes backend operations, not collection, permission, or billing work.

A migration is justified when the new path has passed the important incident and alert checks, and its cost and operating responsibilities fit the team—not simply when its entry price is lower.

Official references

• SigNoz architecture, installation, and license
• SigNoz editions and pricing, metric-sample billing, and enterprise responsibilities
• Datadog OTel compatibility, log correlation, and pricing

Sources checked October 21, 2026. Prices use public USD terms; recheck Region, taxes, contracts, and product editions before purchasing.