Buyer’s Guide - October 2026

The best distributed tracing tools, ranked

Q: What is distributed tracing?

Distributed tracing follows a single request as it moves through services. Each unit of work is a span with a start time, duration, status and attributes, and spans that share a trace ID form a trace. A tracing backend stores those spans and lets you see the full path of a request, where time was spent, and which service returned an error.

Q: Do I need OpenTelemetry to use these tools?

Not strictly, but it is the safest choice. Every tool on this list can receive OpenTelemetry traces, directly or through the Collector, and the Jaeger project itself retired its own client libraries in favour of OpenTelemetry. Instrumenting with OpenTelemetry SDKs and routing through the Collector keeps you free to change backends later.

Q: Does Netdata support distributed tracing?

Yes. Netdata ingests OpenTelemetry traces over OTLP on the same agent that collects your metrics and logs. Spans are stored and indexed on your own infrastructure, and the Traces tab in Netdata Cloud gives you a duration heatmap, filters by service, span name and duration, a full waterfall for every trace, and every span attribute your instrumentation recorded.

Q: Should I sample traces?

At high request rates, usually yes. Head sampling in the SDK is simple; tail sampling in an OpenTelemetry Collector keeps errors and slow traces while dropping routine ones. Some tools add their own controls, such as Jaeger’s adaptive sampling or New Relic’s Infinite Tracing. Netdata stores every span it accepts and leaves sampling to your SDK or Collector.

Q: Do I still need infrastructure monitoring if I have tracing?

Yes. A trace tells you which span was slow; it rarely tells you that the node was swapping, the disk queue was full or the container was CPU-throttled. Pick a tracing tool that puts infrastructure metrics and logs in the same place, or plan how engineers will get from a trace to the host behind it.

OpenTelemetry made instrumentation portable, so choosing a tracing tool is now mostly a choice about where spans are stored, who operates that storage, and what else you can see next to a slow trace. We ranked ten tools on OpenTelemetry alignment, trace exploration, operational surface area and infrastructure context, and we say where each one is genuinely stronger than the others.

Start with Netdata Jump to rankings

The best distributed tracing tools, ranked product interface

Why this list exists

Distributed tracing used to be a decision about agents and instrumentation libraries. OpenTelemetry has mostly settled that: every tool on this list can receive OpenTelemetry traces, directly over OTLP or, for Zipkin, through the Collector, so the same SDKs can feed any of them. What is left to decide is everything that happens after the span leaves your service.

Three dimensions decide the outcome more than any feature checklist:

Where spans live and who runs that storage. A tracing backend is a write-heavy database. Self-hosted tools hand you Cassandra, Elasticsearch, ClickHouse or object storage to operate; SaaS tools hand you a bill that grows with span volume. Either way it is a real cost, and open source is a distribution model, not a free one.
How you find the trace that matters. Some tools start from a search form, some from a query language such as TraceQL, some from charts that surface outliers. Service maps and span-derived RED metrics are where the mature APM suites still lead.
What sits next to the trace. Most slow spans end at a saturated disk, a throttled container or a noisy neighbour. A trace without the infrastructure behind it is half an answer.

We do not quote list prices for any vendor. Pricing pages change, tiers hide limits, and a figure copied here would be stale within a quarter. Each card describes the pricing shape and what makes the bill grow, with a link to the vendor’s current pricing page. For Netdata, see our pricing page. For setup detail on sending OpenTelemetry data to Netdata, see the OpenTelemetry hub.

We shortlisted tools that a team adopting OpenTelemetry would realistically evaluate in 2026: the CNCF and open-source backends, the OpenTelemetry-native newcomers, and the commercial APM suites. Each had to receive OpenTelemetry traces, directly or through the OpenTelemetry Collector, have a release in the last twelve months, and have a documented production deployment path.

OpenTelemetry alignment and trace exploration carry the most weight because they decide whether the tool fits your instrumentation and whether engineers actually find the slow request. Operational surface area and infrastructure context follow, because they decide what the tool costs in people and whether a trace leads anywhere. APM depth (service maps, span metrics, profiling, RUM) is scored separately so that the full suites get credit where they earn it.

Tester credit

Compiled by the Netdata team - Updated October 5, 2026

Sources verified

Scoring criteria

OpenTelemetry alignment 20%
OTLP ingestion, works with standard SDKs and the Collector, no proprietary agent required
Trace exploration 20%
How quickly you get from a symptom to the right trace: search, query language, heatmaps, waterfall, span detail
Operational surface area 20%
Services and storage you must deploy, scale and upgrade to keep traces available
Infrastructure context 15%
Metrics and logs for the same nodes in the same place as the trace
APM depth 15%
Service maps, span-derived metrics and alerts, profiling, RUM
Data ownership 10%
Whether spans can stay on infrastructure you control

Vendor 01 / 10 · #netdata

Netdata

OpenTelemetry traces stored and indexed on the same agent that already collects your per-second metrics and logs, explored in the Netdata Cloud Traces tab.

Best for

Teams that want traces, metrics and logs from one agent instead of a tracing backend plus separate metrics and logs stacks
Organizations that need spans to stay on their own infrastructure
Operators who troubleshoot slow requests down to the node, container or disk underneath them

Pricing

Netdata Agents are open source and run on your own nodes
Traces are explored in Netdata Cloud; current plans and what each covers are on the pricing page
Spans are stored on your agents’ disks, so storage capacity and retention are yours to set

Pros

Ingests OpenTelemetry traces over OTLP/gRPC on the same Agent endpoint that receives OTLP metrics and logs
Every span indexed: span names, kinds, statuses and span, resource, scope, event and link attributes, plus a trace-ID index
Duration heatmap of every trace, with error spans in red; click a cell to narrow the trace list
Full waterfall and span details with Resource, Scope, Events, Links and Raw views, and a shareable deep link to any trace
Trace retention by size, age and file count, separate from metrics and logs, with optional offload to S3-compatible storage
Per-second metrics from 800+ integrations and logs for the same nodes, in the same UI, and AI and MCP access to traces

Where teams pair it

No service maps, and no RED metrics, alerts or anomaly detection generated from spans; teams that depend on those keep an APM suite for that job
OTLP/gRPC only today (OTLP/HTTP is on the roadmap); SDKs that default to HTTP need OTEL_EXPORTER_OTLP_PROTOCOL=grpc or a Collector in front
Traces are queried on the Agent that received them and explored in Netdata Cloud with a sign-in; centralize by pointing senders at one Agent or Parent, and there is no offline trace UI

Verdict

Netdata leads this list on the criteria that cost teams the most over a year: no separate tracing backend to operate, spans that stay on infrastructure you control, and the node behind every slow span in the same place as the trace. It is not the deepest trace analysis tool here. Honeycomb, Tempo’s TraceQL and the APM suites offer more ways to slice spans, and Datadog and Dynatrace map services in ways Netdata does not. If your bottleneck is running and correlating a tracing stack rather than querying it, Netdata is the shortest path.

Start free -> See distributed tracing ->

Vendor 02 / 10 · #jaeger

Jaeger

The CNCF-graduated open-source tracing backend, rebuilt in v2 on the OpenTelemetry Collector framework.

Best for

Teams that want a focused, self-hosted trace store and UI they fully control
Platform teams already operating Cassandra, Elasticsearch or OpenSearch
Environments that need trace exploration fully inside the network

Pricing

Open source under Apache 2.0; you run and operate it yourself
Cost is the storage backend and the people who scale it as span volume grows

Pros

Accepts OTLP over gRPC and HTTP, plus legacy Jaeger and Zipkin formats
Pluggable storage: Cassandra, Elasticsearch, OpenSearch, Badger or in-memory
Service dependency graph, trace search by service, operation, tags and duration, and trace comparison
Remote and adaptive sampling strategies served to SDKs and Collectors

Cons

Traces only; metrics, logs and alerting live in other systems
Production storage is a cluster you size, upgrade and keep on call
The Monitor tab’s service metrics need a span-metrics pipeline and a Prometheus-compatible backend

Verdict

Jaeger is the reference open-source tracing backend and a sound choice when a team wants exactly a trace store and nothing more. It does service dependencies and tag search better than Netdata. The trade is operational: you own the storage tier, and every other signal comes from somewhere else.

Read full Netdata vs Jaeger -> Jaeger website ->

Vendor 03 / 10 · #grafana-tempo

Grafana Tempo

Grafana Labs’ trace backend that stores traces in object storage and queries them with TraceQL.

Best for

Teams already standardized on Grafana, Mimir or Prometheus and Loki
High span volumes where object storage keeps long retention affordable
Engineers who want an expressive query language for traces

Pricing

Open source under AGPLv3; self-hosted Tempo is yours to operate
Grafana Cloud Traces is the managed option, billed by usage

Pros

Object storage (S3, GCS, Azure Blob) as the durable store
TraceQL for structural queries over spans and attributes
Metrics-generator produces span metrics and service graphs
Links from traces to logs, metrics and profiles once Loki, Mimir and Pyroscope are wired in

Cons

No UI of its own; Grafana is required
Usually one part of an LGTM assembly, with each component operated separately
Distributed mode means distributors, ingesters, queriers and compactors to run

Verdict

Tempo is the strongest open-source choice for teams who already live in Grafana and want cheap, long trace retention with a real query language. TraceQL and service graphs are genuine advantages over Netdata. The cost is the stack around it: Tempo alone is a trace store, not a place to troubleshoot.

Read full Netdata vs Grafana Tempo -> Grafana pricing ->

Vendor 04 / 10 · #signoz

SigNoz

OpenTelemetry-native APM with traces, metrics and logs in one ClickHouse-backed store.

Best for

Teams that want a Datadog-style APM experience from an open-source project
OpenTelemetry-first organizations comfortable operating ClickHouse

Pricing

Open-source core, self-hosted on ClickHouse
SigNoz Cloud is the managed option, billed by data ingested

Pros

Native OTLP ingestion with no proprietary agent
Service maps, span-derived RED metrics and trace-based alerts
One query surface across traces, metrics and logs

Cons

ClickHouse becomes part of your on-call surface when self-hosted
Infrastructure monitoring depth lags dedicated infrastructure tools
Smaller ecosystem than the Grafana or Elastic stacks

Verdict

SigNoz is the most complete open-source APM experience on this list, with service maps and span metrics Netdata does not offer. Choose it when APM features drive the decision and your team is ready to own a ClickHouse deployment or pay for SigNoz Cloud.

Read full Netdata vs SigNoz -> SigNoz pricing ->

Vendor 05 / 10 · #honeycomb

Honeycomb

SaaS observability built around wide events and fast, high-cardinality queries over traces.

Best for

Developer teams debugging unknown failures in distributed systems
Workloads where high-cardinality attributes (user, tenant, build) drive investigations

Pricing

SaaS, billed by event volume; spans count as events
Cost grows with traffic and instrumentation detail; tail sampling is the usual control

Pros

Fast ad-hoc queries over any span attribute, at high cardinality
BubbleUp highlights which attributes differ in an outlier group
SLOs built on trace data, and strong OpenTelemetry support

Cons

SaaS only; spans are stored in Honeycomb’s infrastructure
Infrastructure monitoring is not its focus
Sampling strategy needs deliberate design to control volume

Verdict

Honeycomb is the best tool here for open-ended trace analysis, and it beats Netdata on slicing spans by arbitrary attributes. It is a developer debugging tool first; teams that also need host and container visibility pair it with an infrastructure tool, and teams that need spans on their own infrastructure cannot use it.

Read full Netdata vs Honeycomb -> Honeycomb pricing ->

The full APM suites

The next four entries are complete APM platforms where tracing is one feature among many: service maps, span-derived metrics, profiling, real user monitoring and more. They have the deepest tracing features on this list. They rank lower here because they score lower on operational surface area, data ownership or cost predictability, not because their tracing is weak.

Vendor 06 / 10 · #datadog

Datadog APM

The broadest commercial APM: traces, service maps, profiling and RUM in one SaaS platform.

Best for

Teams already on Datadog for infrastructure and logs
Organizations that want service maps, profiling and RUM connected to traces out of the box

Pricing

Per APM host, with span ingestion and indexed span retention billed on top
Each product (infrastructure, logs, RUM, profiling) adds its own line item

Pros

Service map and span-derived service metrics with alerting
Continuous profiler linked to traces
Traces connected to RUM sessions, logs and infrastructure in one UI
Accepts OpenTelemetry data alongside its own tracing libraries

Cons

SaaS only; traces are stored in Datadog’s infrastructure
Retention and ingestion controls take ongoing tuning to keep the bill predictable
Deepest features assume Datadog’s own agent and libraries

Verdict

Datadog APM is the most feature-complete commercial tracing product here, and on APM depth it is ahead of Netdata. Pick it when you want one vendor for every signal and accept SaaS storage and a bill with several meters.

Read full Netdata vs Datadog -> Datadog pricing ->

Vendor 07 / 10 · #dynatrace

Dynatrace

Enterprise APM with automatic instrumentation, PurePath distributed traces and Smartscape topology.

Best for

Large enterprises that want automatic instrumentation across Java, .NET and other runtimes
Teams that rely on automatic topology mapping and AI-assisted root cause analysis

Pricing

Usage-based platform subscription; consumption depends on hosts, data and features used

Pros

OneAgent instruments supported runtimes automatically
PurePath traces with code-level detail
Smartscape topology and Davis AI root-cause analysis
Ingests OpenTelemetry traces

Cons

Heavier platform to adopt and license than a focused trace backend
Deepest value assumes OneAgent everywhere
Data is stored in Dynatrace’s platform

Verdict

Dynatrace has the most automated APM on this list, with topology and code-level trace detail that Netdata does not try to match. It suits enterprises that want instrumentation and dependency mapping done for them.

Read full Netdata vs Dynatrace -> Dynatrace pricing ->

Vendor 08 / 10 · #newrelic

New Relic

Full-stack SaaS observability with APM agents, distributed tracing and native OTLP ingestion.

Best for

Teams that want APM, infrastructure, logs and browser monitoring under one account
Organizations moving to OpenTelemetry but keeping a commercial backend

Pricing

Usage-based: data ingested plus user seats
The bill grows with trace volume and with the number of full-platform users

Pros

Mature APM agents plus native OTLP endpoints
Service maps and span-derived metrics
Infinite Tracing for tail-based sampling

Cons

SaaS only; trace data is stored by New Relic
Per-user pricing shapes who gets access
Data volume needs active management to keep costs forecastable

Verdict

New Relic is a solid all-in-one APM with good OpenTelemetry support and tail-based sampling. Its APM depth exceeds Netdata’s; the trade-offs are SaaS storage and a bill shaped by both data and users.

Read full Netdata vs New Relic -> New Relic pricing ->

Vendor 09 / 10 · #elastic-apm

Elastic APM

APM built on Elasticsearch and Kibana, with traces next to your existing Elastic logs.

Best for

Organizations already running Elasticsearch for logs or search
Teams that want self-managed APM on infrastructure they control

Pricing

Self-managed under Elastic’s licenses, or Elastic Cloud billed by resources or usage
Some APM features depend on subscription tier

Pros

Traces stored in Elasticsearch, searchable with the same tools as logs
Elastic APM agents plus OpenTelemetry ingestion
Service maps and APM views in Kibana

Cons

Elasticsearch storage and cluster operations grow with span volume
Feature availability varies by license tier
Infrastructure metrics are not its strongest area

Verdict

Elastic APM is the natural pick for teams with an existing Elastic deployment who want traces and logs in one search engine. If you do not already run Elasticsearch, adopting it for traces is a large operational commitment.

Read full Netdata vs ELK -> Elastic pricing ->

Vendor 10 / 10 · #zipkin

Zipkin

The original open-source distributed tracing system, still simple, small and widely understood.

Best for

Small deployments that want a lightweight trace store and UI
Estates with existing Zipkin and B3 propagation

Pricing

Open source under Apache 2.0; you run and operate it yourself

Pros

Simple to run, with in-memory, Cassandra, Elasticsearch and MySQL storage options
Dependency diagram and straightforward trace search
B3 propagation is widely supported

Cons

Traces only, with a basic UI by current standards
Slower development pace than newer projects
Usually fed OpenTelemetry data through a Collector’s Zipkin exporter rather than OTLP directly

Verdict

Zipkin remains a reasonable lightweight trace store, especially where B3 propagation is already in place. For new OpenTelemetry deployments, Jaeger or Tempo offer more active development and more features.

Zipkin website ->

How to pick

A simple decision tree

The ten tools above cluster into four buying patterns. Pick the pattern first, then the tool.

If your constraint is fewer systems to operate and spans on your own infrastructure

Start with Netdata (#1). Traces, metrics and logs arrive at one agent, and the Traces tab sits next to the nodes they ran on. Check that you can live without service maps and span-derived metrics.

If you want a focused open-source trace store

Jaeger (#2) for a self-contained backend and UI, Grafana Tempo (#3) if you already run Grafana and want object storage and TraceQL.

If APM features drive the decision

SigNoz (#4) for open source, Datadog (#6), Dynatrace (#7) or New Relic (#8) for commercial suites with service maps, span metrics and profiling. Elastic APM (#9) if you already run Elasticsearch.

If you need open-ended, high-cardinality trace analysis

Honeycomb (#5).

Whatever you choose, instrument with OpenTelemetry. It keeps the backend decision reversible, and an OpenTelemetry Collector can send the same spans to two tools while you compare them on real traffic.

Frequently Asked Questions

What is distributed tracing?

Do I need OpenTelemetry to use these tools?

Does Netdata support distributed tracing?

Can Netdata replace Jaeger or Tempo?

For OpenTelemetry-instrumented services, yes: point your SDKs or Collector at a Netdata Agent over OTLP/gRPC and explore traces in Netdata Cloud, next to your metrics and logs. Netdata does not ingest the native Jaeger or Zipkin protocols, so send spans through OpenTelemetry. If your SDKs export OTLP/HTTP, switch them to gRPC or put an OpenTelemetry Collector in between; native OTLP/HTTP support is on the roadmap. See Netdata vs Jaeger and Netdata vs Grafana Tempo for what you gain and give up.

What is the best open-source distributed tracing tool?

It depends on what you want to operate. Jaeger is the reference self-hosted trace backend, Grafana Tempo is the best fit for Grafana users who want object storage and TraceQL, and SigNoz is the most complete open-source APM. Netdata’s agent is open source and stores spans on your own nodes, with exploration in Netdata Cloud. In every case open source means you run it: the operating cost is real.

Should I sample traces?

What is the difference between distributed tracing and APM?

Distributed tracing records and shows the path of requests through services. APM suites build more on top: service maps, RED metrics and alerts derived from spans, code-level profiling, real user monitoring and error tracking. Jaeger, Tempo and Zipkin are tracing backends; Datadog, Dynatrace, New Relic and Elastic are APM suites. Netdata does distributed tracing next to per-second infrastructure metrics and logs, without service maps or code profiling.

Where should trace data be stored?

That is the biggest architectural choice on this list. SaaS tools store spans in the vendor’s infrastructure. Self-hosted tools store them in a database or object storage you operate: Cassandra, Elasticsearch or OpenSearch for Jaeger, object storage for Tempo, ClickHouse for SigNoz. Netdata stores spans on the disk of the agent that receives them, with optional offload to S3-compatible storage, and Netdata Cloud brokers access without storing trace data.

Do I still need infrastructure monitoring if I have tracing?

The best distributed tracing tools, ranked

How we evaluated distributed tracing tools

Netdata

Jaeger

Grafana Tempo

SigNoz

Honeycomb

Datadog APM

Dynatrace

New Relic

Elastic APM

Zipkin

A simple decision tree

If your constraint is fewer systems to operate and spans on your own infrastructure

If you want a focused open-source trace store

If APM features drive the decision

If you need open-ended, high-cardinality trace analysis

Frequently Asked Questions

Book Your Free Demo