Open-source IBM Instana alternatives

Netdata is a real-time infrastructure monitoring platform. You install an agent on servers, containers or Kubernetes nodes and it starts collecting metrics every second, showing them on dashboards without lengthy setup. Its origin story is a frustration with monitoring tools that offered few metrics at low resolution and cost too much to run.

The project highlights zero-configuration deployment, machine-learning-based anomaly detection and automated analysis, and an efficient agent that uses few resources. Data stays distributed on the monitored machines instead of being shipped to one central store. Integrations cover databases such as MySQL, PostgreSQL and MongoDB, plus Docker and Kubernetes, and the topics list mentions export paths to Prometheus, Grafana and InfluxDB. Netdata also advertises AI-assisted troubleshooting and an MCP interface.

The agent is GPL-3.0 licensed and written largely in Go, and Netdata Cloud offers a hosted layer for viewing many nodes together. It suits operations teams and lean engineering groups that want deep visibility quickly rather than building a metrics stack from several separate tools.

SigNoz is an observability platform, open source and built on OpenTelemetry. It aims to replace a fragmented monitoring stack with a single place for logs, metrics, traces, alerts, exceptions and dashboards, and it presents itself as useful for both engineering teams and their AI agents.

The APM view follows how each service performs, covering latency, error rate, throughput, Apdex, the busiest endpoints, and calls to databases and external services. Log management supports ingesting, searching and correlating logs with traces and metrics through a visual query builder. Dashboards can be built with a query builder, PromQL or ClickHouse SQL, infrastructure monitoring covers Kubernetes clusters, pods, nodes and host resources, and LLM and AI observability traces prompts, tool calls, tokens, latency and cost.

You can choose how to run it. SigNoz Cloud is fully managed, with a 30-day free trial and usage-based pricing that starts at $49. Enterprise options cover cloud, bring-your-own-cloud and self-hosted deployments with extra support and controls. The free Community edition runs in your own infrastructure via Docker, Kubernetes or Linux. The repository license is listed as 'Other' on GitHub.

OpenObserve, often shortened to O2, is open-source observability software spanning logs, metrics, traces and analytics, real user monitoring on web, Android and iOS, session replay, pipelines, SLOs and AI and LLM observability. It is pitched as a lower-cost option next to Datadog, Splunk and Elasticsearch for teams that want complete observability without the complexity or price.

Its architecture uses Parquet columnar storage and an S3-native design, which the project says can cut storage costs by up to 140 times compared with Elasticsearch, with petabyte-scale capacity. Topics reference OpenTelemetry, Prometheus, Jaeger and Kibana, showing the standards and tools it interoperates with. The README has sections on architecture, comparisons, production readiness, security and compliance, and an enterprise edition.

OpenObserve is AGPL-3.0 licensed, with a hosted cloud option and a self-hosted deployment, and documentation and a Slack community are provided. It suits engineering teams looking to consolidate logging, metrics and tracing in one tool while keeping storage costs under control.

Pinpoint is an APM tool, short for application performance management, built for large-scale distributed systems. Inspired by Google's Dapper, it traces transactions as they move across the components of a distributed application so teams can see how the system is structured and where problems or bottlenecks arise.

Its ServerMap visualizes application topology and how components connect, while real-time monitoring and code-level visibility into each transaction help diagnose slow or failing requests. Agents can be installed without changing application code, and the project says their overhead on resource usage is low. Agents are available for Java, PHP and Python applications.

Pinpoint is written in Java and licensed under Apache-2.0. You deploy and operate the Pinpoint components yourself, with a quick-start guide, an installation guide and a Kubernetes deployment option provided, and a live demo is available online for a first look.

HyperDX is an open-source observability platform for working out why production is broken. It brings logs, metrics, traces, errors and session replays into one place, and works as a search and visualization layer on top of ClickHouse and OpenTelemetry, which the authors compare to Kibana for ClickHouse. It is a core component of ClickStack.

It is schema agnostic and can sit on top of an existing ClickHouse schema. Searching uses full-text and property syntax such as level:err, with SQL as an option, events can be compared with event deltas, high-cardinality events can be charted on dashboards, and alerts can be set up in a few clicks. Live tail shows the newest logs and traces, JSON strings can be queried natively, OpenTelemetry works out of the box, and APM views cover HTTP requests down to database queries.

One deployment option is ClickStack, a bundle of ClickHouse, HyperDX, an OpenTelemetry Collector and MongoDB, and HyperDX can instead be pointed at your own ClickHouse instance. It can also be used with ClickHouse Cloud. The project is MIT licensed and written in TypeScript.

Coroot is an open-source observability and application performance monitoring tool. Its pitch is that collecting metrics, logs and traces is not enough on its own, so it turns the data into actionable insights, including AI-powered root cause analysis, predefined inspections and a service map.

It gathers metrics, logs, traces and continuous profiles automatically with eBPF, so legacy and third-party services can be observed without code changes, and it is also vendor-neutral through OpenTelemetry instrumentation. Features include a service map, application health summaries, SLO tracking and alerting, distributed tracing, log pattern clustering with logs-to-traces correlation using ClickHouse, and one-click profiling that drills down to the line of code. Predefined inspections audit each application without configuration.

Coroot is written in Go and licensed under Apache-2.0, with a live demo, documentation and a separate Coroot Enterprise edition. Topics include database monitoring, microservices and Prometheus. It suits SRE and platform teams that want observability for Kubernetes and Linux workloads without manually instrumenting every service.

Pixie is an open-source observability platform made for applications running on Kubernetes. It gives a high-level view of a cluster, including service maps, resource usage and application traffic, and lets you drill into individual pods, flame graphs and full request bodies without adding instrumentation code to your services.

It relies on eBPF to gather telemetry automatically, covering requests, resource and network metrics and application profiles. Data is collected, stored and queried on the cluster itself, an approach the project calls in-cluster edge compute, which keeps raw telemetry local and limits overhead. Scripts written in a Python-based, dataframe-style language let you build custom views, and the project is hosted by the Cloud Native Computing Foundation.

Pixie is licensed under Apache-2.0 and is installed into your own Kubernetes cluster, whether a local one such as minikube or a managed cluster on GKE, EKS or AKS. It suits platform and application teams that want fast insight into service behavior without a separate agent-heavy monitoring stack.

Odigos is an open-source distributed tracing solution aimed at Kubernetes environments and virtual machines. Its core promise is instant tracing without changing application code, which helps platform engineers, DevOps professionals and SREs improve observability quickly.

It builds on OpenTelemetry and eBPF to instrument applications automatically, covering languages such as Go, Java, .NET, Node.js and Python according to the repository topics, and it can emit traces, metrics and logs. The data can be sent to many observability back ends, so it fits into existing monitoring stacks. Documentation, a demo video and a Slack community are linked from the README. It is written in Go.

Odigos is licensed under Apache-2.0 and installed into your own cluster. It suits teams that want OpenTelemetry coverage across many services without rolling out SDKs and code changes service by service.

Uptrace is an open-source application performance monitoring platform. It collects distributed traces, metrics and logs through OpenTelemetry and presents them in a single interface, so teams can monitor applications and troubleshoot problems without switching between separate tracing, metrics and logging tools.

The interface includes a query builder, a service graph, faceted filters, chart annotations and many prebuilt dashboards that appear automatically as metrics arrive. Alerting rules send notifications through email, Slack, WebHook or AlertManager. Data can be ingested via OpenTelemetry, Prometheus, Vector, FluentBit and CloudWatch, Grafana can use Uptrace as a Tempo or Prometheus data source, and single sign-on works through OpenID Connect.

Uptrace is written in Go with a Vue front end, stores telemetry in ClickHouse and keeps metadata such as alerts in PostgreSQL. It is licensed under AGPL-3.0 and can be self-hosted with Docker, and a cloud demo needs no login. The project says it is meant to cut monitoring costs, which suits engineering teams comparing self-hosted APM options with commercial observability suites.

Traceway is an observability platform built natively on OpenTelemetry. It brings logs, traces, metrics, exception tracking, session replay, AI tracing and on-call paging into one product, and you can start by pointing an OTLP exporter at it without a separate collector or a vendor-specific SDK for each language.

Logs are structured and linked to traces, so you can jump from a log line to its span. Traces show end-to-end waterfalls, endpoint views rank routes by latency percentiles and error rate, and exceptions are grouped into ranked issues with source map support. There is experimental profiling with flame graphs, session replay for web and Flutter apps, and LLM observability covering cost, tokens and latency.

The project is written in Go, licensed under MIT and says that every feature ships in the open-source build with no open-core split. You can self-host it for free, using a setup the README says takes about 90 seconds, or run it on Traceway Cloud if you prefer not to manage infrastructure. It suits engineering teams looking for one tool to find and diagnose production problems.