Jaeger is a free, open source distributed tracing platform originally developed at Uber and now part of the Cloud Native Computing Foundation. CNCF lists Jaeger as a graduated project, with graduation dated October 31, 2019.
Jaeger pricing is different from SaaS observability pricing. There is no Jaeger license fee, paid tier, user seat, or hosted plan from the Jaeger project. The real cost is the infrastructure and engineering time needed to run collectors, query services, storage, sampling, retention, upgrades, and on-call operations.
In this guide, we review Jaeger pricing, self-hosting costs, v2 architecture, storage backends, user-review limitations, and alternatives such as CubeAPM, Datadog, Grafana Tempo, and SigNoz.
What Is Jaeger?
Jaeger is a distributed tracing system for monitoring and troubleshooting microservices. It maps how requests move across services, queues, databases, and APIs so teams can identify latency, errors, and service bottlenecks. The official Jaeger site describes it as a tool for connecting request flow across distributed systems and finding performance issues or errors.
Jaeger is best understood as a tracing backend, not a full observability platform. It accepts and stores trace data, then provides a UI for searching traces, viewing span waterfalls, and analyzing request paths. Official Jaeger v2 API docs state that Jaeger accepts OTLP trace data, but only tracing data, because Jaeger does not store other telemetry types.
That distinction matters for pricing. Jaeger can be very cost-effective for teams that only need tracing and already have platform engineering capacity. But teams that also need logs, infrastructure metrics, RUM, synthetics, alerting, and dashboards usually need to run Jaeger alongside other tools.
Supported Languages, Integrations, and Data Sources
Jaeger now recommends OpenTelemetry SDKs for instrumentation. Its native Jaeger client SDKs were retired in 2022, and the official migration page says new applications should use OpenTelemetry APIs, SDKs, and instrumentation.
| Area | Jaeger support |
| Languages | Go, Java, Python, Node.js, C++, C#, and others through OpenTelemetry SDKs |
| Ingestion | OTLP over gRPC, OTLP over HTTP, and OTLP JSON for trace data |
| Infrastructure | Kubernetes, containers, VMs, cloud, hybrid, and on-prem environments |
| Storage | OpenSearch, Elasticsearch, Cassandra, ClickHouse experimental, Kafka buffer, memory, Badger |
| Ecosystem | OpenTelemetry Collector, Prometheus, Grafana, Envoy, Istio, Helm, Kubernetes |
Key Features of Jaeger
Jaeger’s core feature is distributed tracing. It records operations as spans, links them into traces, and shows how a request moves across services. This helps engineering teams find where latency, errors, retries, or downstream failures happen in a microservice chain.
Jaeger helps teams narrow root cause by showing the service path, timing, and span relationships for a request. It does not automatically solve every incident, but it gives engineers the trace-level evidence needed to isolate slow or failing services.
Jaeger can show service dependencies based on trace data. However, the official FAQ notes that with distributed storage backends such as Cassandra, Elasticsearch, or OpenSearch, dependency graphs are not always calculated by scanning all traces directly; Jaeger uses external jobs such as Spark or experimental Flink jobs to extract dependency data.
Jaeger requires persistent storage for production deployments. Official Jaeger storage docs list Cassandra, Elasticsearch, and OpenSearch as the primary supported distributed backends, with ClickHouse supported experimentally behind a feature gate. The same page says the Jaeger team recommends OpenSearch for large-scale production over Cassandra.
Sampling is one of Jaeger’s most important cost-control levers. CNCF’s Jaeger v2 release post says v2 supports remote and adaptive head-based sampling, and adds tail-based sampling through the OpenTelemetry Collector ecosystem.
Jaeger includes a web UI for searching traces, viewing span waterfalls, and inspecting service paths. The UI is useful for tracing workflows, but it is more focused than modern commercial observability platforms that combine traces with logs, metrics, dashboards, anomaly detection, and alerts.
Jaeger v2 is built on the OpenTelemetry Collector framework. CNCF says v2 natively supports OTLP and avoids translating OTLP into Jaeger’s older internal data format. It also inherits Collector features such as health checks, internal monitoring, zPages, authentication, and Collector-style pipelines.
Jaeger v2 Architecture in Brief
Jaeger v2 is a major architectural shift. The official download page says Jaeger v2 is based on the OpenTelemetry Collector framework, while Jaeger v1 reached end-of-life on December 31, 2025 and no longer receives updates. New deployments should use v2.
In v1, Jaeger commonly used separate binaries for agent, collector, ingester, and query. In v2, CNCF says Jaeger ships as a single jaeger binary that can be configured for roles such as collector, ingester, query, and all-in-one through YAML configuration.
| Role | Function |
| collector | Receives trace data and writes it to storage |
| query | Serves APIs and the Jaeger UI |
| ingester | Reads spans from Kafka and writes them to storage |
| all-in-one | Runs collector and query roles together |
| Kafka buffer | Optional queue between collectors and storage |
The all-in-one pattern is useful for development, demos, and very small deployments. For production, Jaeger needs persistent storage such as OpenSearch, Elasticsearch, or Cassandra. Larger deployments may add Kafka between collection and storage to absorb ingestion spikes.
Jaeger Pricing in 2026
Jaeger is free and open source under the Apache 2.0 license. The Jaeger GitHub repository lists Apache 2.0 as the project license, and there is no official paid Jaeger tier, enterprise edition, or per-seat charge from the Jaeger project.
So the honest answer is simple: Jaeger software costs $0.
The real cost is the total cost of ownership around it.
| Cost area | What it includes |
| Compute | Collector, query, ingester, and supporting services |
| Storage | OpenSearch, Elasticsearch, Cassandra, or other backend |
| Data transfer | Internal traffic, cross-zone traffic, and egress |
| Engineering time | Setup, scaling, retention, sampling, upgrades, incidents |
| Additional tooling | Metrics, logs, dashboards, alerts, RUM, synthetics |
Is There a Free Tier in Jaeger?
Jaeger does not have a SaaS-style free tier because it is not sold as a SaaS product by the Jaeger project. The whole project is free to download and run. The cost begins when you choose where and how to run it: Kubernetes, VMs, managed databases, storage volumes, Kafka, backups, and the engineers who maintain the stack.
This makes Jaeger attractive for teams with strong platform engineering skills. It is less attractive for teams that want full observability with minimal day-two operations.
How Jaeger Cost Is Measured
Because Jaeger has no license fee, pricing should be modeled as TCO.
The main drivers are trace volume, retention, sampling, storage backend, compute, and engineering time. Storage usually dominates at production scale because Jaeger trace data must be written, indexed, retained, queried, and eventually expired.
Jaeger also has a narrower scope than full observability platforms. Official Jaeger docs say only tracing data is accepted because Jaeger does not store other telemetry types. That means logs, metrics, RUM, synthetics, dashboards, and alerting require separate tooling.
What Drives Jaeger Costs?
Storage is the biggest production cost driver. Jaeger requires persistent storage, and official docs list Cassandra, Elasticsearch, and OpenSearch as primary supported distributed backends. ClickHouse is available experimentally behind a feature gate.
Trace volume grows with traffic. Sampling controls how much trace data is retained. Jaeger v2 supports remote and adaptive head-based sampling and tail-based sampling through the OpenTelemetry Collector ecosystem.
Longer retention means larger storage clusters. Jaeger’s storage docs note that primary storage is typically used with shorter TTLs, while archive storage can be configured separately for traces that need longer retention.
Jaeger is free software, but someone still owns deployment, upgrades, storage tuning, sampling policies, security, backups, and incident response. This cost is easy to miss in a “free open source” calculation.
Jaeger does not store logs, metrics, RUM, or synthetics. Official API docs state that only tracing data is accepted. Teams that need full observability usually add Prometheus, Loki, Grafana, alerting, log storage, and other systems.
Cross-zone, cross-region, and high-volume internal traffic can add cost. This depends heavily on the cloud provider and deployment topology, so it should be modeled with real traffic patterns.
Jaeger User Reviews
Public review data for Jaeger needs careful handling. G2 shows a Jaeger listing with 4.4/5 from 15 reviews, but visible review snippets include comments that appear to describe Uber’s rideshare app rather than the Jaeger tracing platform. Because of that, the rating should not be treated as a clean buyer-review signal for Jaeger as observability software.
A better approach is to combine public review snippets with technical documentation and community/analyst commentary. Jaeger is widely recognized as a strong open source tracing backend, but it is not a full-stack observability product.
What Users and Practitioners Praise
- Jaeger is commonly praised for giving deep visibility into request paths across microservices. Its main strength is that it lets engineers inspect how a request moved through services and where latency or errors appeared.
- Teams also value its open source model, OpenTelemetry compatibility, and flexible storage options. Jaeger v2 strengthens this by aligning the project with the OpenTelemetry Collector and supporting OTLP natively.
What Users and Practitioners Criticize
- The biggest criticism is operational complexity. Running Jaeger in production means operating collectors, query services, storage backends, retention, sampling, and sometimes Kafka. This is not the same experience as subscribing to a managed observability platform.
- The second limitation is scope. Jaeger is focused on tracing and does not store other telemetry types. Teams that need logs, metrics, RUM, synthetics, and alerting need additional tools.
- The third limitation is analytics depth. Jaeger’s UI is useful for trace search and waterfall analysis, but teams that want richer correlation, dashboards, anomaly detection, and automated alerting usually need to pair it with other systems.
Jaeger Alternatives: How It Compares to Competitors
Jaeger vs CubeAPM
Jaeger is a free, self-operated tracing backend. CubeAPM is a self-hosted, vendor-managed observability platform with per-GB pricing across APM, traces, logs, infrastructure monitoring, RUM, synthetics, error tracking, dashboards, RBAC, SSO, MFA, and audit logs. CubeAPM lists Pro pricing at $0.15/GB.
| Category | Jaeger | CubeAPM |
| Deployment | Self-hosted, self-operated | Self-hosted, vendor-managed |
| Pricing | Free software, infrastructure cost | $0.15/GB |
| Scope | Tracing | Full-stack observability |
| Operations | Team owns pipeline | Vendor handles upgrades/patches |
| Best for | Teams with platform engineering capacity | Teams wanting full-stack observability with lower ops burden |
CubeAPM is stronger when the buyer wants OpenTelemetry-native observability, predictable per-GB pricing, and data control without running every part of the pipeline internally.
Jaeger vs New Relic
New Relic is a managed full-stack observability platform covering APM, infrastructure monitoring, logs, browser monitoring, synthetics, errors, dashboards, alerts, and telemetry analytics. Unlike Jaeger, it is not tracing-only. New Relic pricing is based mainly on data ingest and users: its pricing page lists 100 GB of free data ingest per month, then $0.40/GB for Original Data and $0.60/GB for Data Plus beyond the free allowance on Standard and Pro plans. New Relic docs also state that billing is based on GB ingested, regardless of telemetry type.
| Category | Jaeger | New Relic |
| Deployment | Self-hosted | SaaS |
| Pricing | Free software + infrastructure | 100 GB free, then usage + users |
| Scope | Tracing | Full-stack observability |
| Operations | Team runs the pipeline | Vendor-managed |
| Best for | Teams wanting open source tracing control | Teams wanting managed observability with a strong free tier |
New Relic is stronger when teams want a managed platform with broad observability coverage and do not want to operate trace storage themselves. Jaeger is stronger when the team only needs tracing, wants full infrastructure control, and has the engineering capacity to run the stack.
Jaeger vs Dynatrace
Dynatrace is an enterprise observability platform built around automated discovery, dependency mapping, APM, infrastructure monitoring, log analytics, digital experience monitoring, Kubernetes monitoring, and AI-assisted root-cause analysis. Dynatrace pricing is usage-based under its platform subscription model, with capabilities drawing from a shared capacity pool. Its public rate card lists examples such as $0.01 per host-hour for Foundation & Discovery, $0.002 per hour for Kubernetes pod monitoring, $0.20 per GiB for log ingest and processing, and $0.001 per synthetic request.
| Category | Jaeger | Dynatrace |
| Deployment | Self-hosted | SaaS or managed platform |
| Pricing | Free software + infrastructure | Usage-based platform subscription |
| Scope | Tracing | Enterprise full-stack observability |
| Operations | Team owns storage and scaling | Vendor-managed |
| Best for | Open source tracing-first teams | Large teams needing AI-assisted observability |
Dynatrace is stronger for large enterprises that want automated discovery, deep dependency mapping, Davis AI, and a managed observability platform across applications, infrastructure, logs, user experience, and Kubernetes. Jaeger is better for teams that want a lighter, open source tracing backend and are comfortable managing the operational work themselves.
Jaeger vs Datadog
Datadog is a managed SaaS observability platform with infrastructure monitoring, APM, logs, RUM, synthetics, and many other modules. Datadog’s public pricing list shows Infrastructure Pro at $15 per infra host per month when billed annually, while Datadog’s APM billing docs list APM Host at $31 per underlying APM host per month, including 1 million indexed spans and 150 GB of ingested spans per APM host.
| Category | Jaeger | Datadog |
| Deployment | Self-hosted | SaaS |
| Pricing | Free software + infrastructure | Modular per-host and usage pricing |
| Scope | Tracing | Full observability |
| Operations | You operate it | Vendor-managed |
| Best for | Open source tracing control | Managed enterprise observability |
Datadog is easier to operate and broader than Jaeger, but it can become expensive as teams add infrastructure monitoring, APM, logs, synthetics, RUM, indexing, and retention.
Jaeger vs Grafana Tempo
Grafana Tempo is an open source distributed tracing backend built for the Grafana ecosystem. Grafana describes Tempo as available through Grafana Cloud Traces and says the free tier includes up to 50 GB of traces.
| Category | Jaeger | Grafana Tempo |
| Deployment | Self-hosted | Self-hosted or Grafana Cloud |
| Pricing | Free software + infrastructure | Free OSS or Grafana Cloud pricing |
| Storage model | Search-oriented backends | Object-storage-oriented design |
| Ecosystem | Standalone tracing backend | Strong Grafana/Loki/Mimir fit |
| Best for | Flexible tracing backend | Teams already using Grafana stack |
Tempo is often a natural fit for teams already using Grafana, Loki, and Mimir. Jaeger is better when teams want a mature standalone tracing backend with multiple storage-backend options.
Jaeger vs SigNoz
SigNoz is an OpenTelemetry-native observability platform that combines logs, metrics, and traces. SigNoz pricing states that its $49/month plan includes access to all features and $49 worth of usage, such as 163 GB of logs/traces or 490 million metric samples, with unlimited teammates and hosts.
| Category | Jaeger | SigNoz |
| Deployment | Self-hosted | Self-hosted or cloud |
| Pricing | Free software + infrastructure | Free self-hosted; cloud usage pricing |
| Scope | Tracing | Logs, metrics, traces |
| Backend | OpenSearch, Elasticsearch, Cassandra | ClickHouse |
| Best for | Pure tracing control | Unified OTel observability |
SigNoz is broader than Jaeger because it covers more telemetry types in one product. Jaeger is more focused and better suited to teams that specifically want a tracing backend and already have other observability tools.
Is Jaeger the Right Choice?
Jaeger Works Best For
- Jaeger is a strong fit for Kubernetes-native teams with platform engineering capacity. It works especially well when the team wants to own the trace pipeline, storage backend, retention, and sampling strategy.
- It also fits teams that want full control over trace data and do not want vendor lock-in. Jaeger’s OpenTelemetry support makes instrumentation portable, and the project’s Apache 2.0 license keeps the software free to use and modify.
- Jaeger is also a good fit for tracing-first use cases. If distributed tracing is the main requirement and logs/metrics are already handled elsewhere, Jaeger can be a practical and mature choice.
Jaeger May Not Be the Right Fit For
- Jaeger is not ideal for teams that want unified observability out of the box. It does not store logs, metrics, RUM, or synthetic monitoring data, so teams must assemble the rest of the stack themselves.
- It may also be a poor fit for teams without operational bandwidth. Production Jaeger requires storage operations, sampling decisions, scaling, security, upgrades, and incident handling.
- It is also not the best fit for buyers who want rich dashboards, alerting, analytics, and cross-signal correlation without stitching together multiple tools.
Conclusion
Jaeger remains one of the strongest open source options for distributed tracing. It is CNCF-graduated, Apache 2.0 licensed, and now OpenTelemetry-native in v2. The move to the OpenTelemetry Collector framework makes Jaeger more aligned with modern telemetry pipelines and OTLP-based instrumentation.
The pricing story is simple but often misunderstood. Jaeger costs $0 to license, but production Jaeger is not free to operate. The real cost comes from compute, storage, data transfer, retention, sampling, upgrades, and engineering time.
Jaeger is a strong choice for tracing-first teams with platform engineering capacity. Teams that want full-stack observability with lower operational burden should compare it with platforms such as CubeAPM, Datadog, Grafana Tempo, and SigNoz before deciding.
Disclaimer: Pricing, packaging, features, and ratings can change. The cost examples in this article are editorial estimates based on publicly available information as of June 2026. Jaeger itself is free and open source; the figures shown reflect estimated infrastructure and operational cost. Always confirm current pricing, usage limits, and terms directly with each vendor before purchase.
FAQs
1. How much does Jaeger cost?
Jaeger software costs $0. It is open source and licensed under Apache 2.0. The real cost is infrastructure and engineering time.
2. Is Jaeger really free?
Yes. The Jaeger project does not sell a paid tier or enterprise edition. You pay for the infrastructure and operations required to run it.
3. Does Jaeger include metrics and logs?
No. Jaeger accepts OTLP trace data, but official docs state that only tracing data is accepted because Jaeger does not store other telemetry types.
4. What is the difference between Jaeger v1 and v2?
Jaeger v2 is based on the OpenTelemetry Collector framework and ships as a single configurable binary. Jaeger v1 reached end-of-life on December 31, 2025 and no longer receives updates.
5. What storage backends does Jaeger support?
Jaeger’s primary supported distributed storage backends are Cassandra, Elasticsearch, and OpenSearch. ClickHouse is supported experimentally behind a feature gate, and Kafka can be used as an intermediate buffer.
6. What drives Jaeger cost?
The biggest cost drivers are storage backend size, trace volume, retention, sampling strategy, compute, data transfer, and engineering time.
7. What are the best Jaeger alternatives?
Strong alternatives include CubeAPM, Datadog, Grafana Tempo, and SigNoz. CubeAPM is best for self-hosted, vendor-managed full-stack observability. Datadog is best for managed SaaS breadth. Grafana Tempo is best for Grafana-stack teams. SigNoz is best for open source unified OTel observability.