About
Pixie is a powerful, open-source observability and live-debugging platform purpose-built for Kubernetes environments. Originally created by New Relic and now a CNCF sandbox project, Pixie leverages eBPF (extended Berkeley Packet Filter) technology to automatically collect metrics, events, traces, and logs without requiring any application code instrumentation or modification. Running entirely within your Kubernetes clusters (v1.16+, Linux kernel v4.14+), Pixie ensures that sensitive telemetry data never leaves your infrastructure, eliminating the cost and complexity of shipping petabytes of data to external platforms. This in-cluster architecture gives teams deep visibility without privacy trade-offs. Pixie provides three primary interaction modes: the Pixie Live interface for navigating data with auto-generated debug views, the `px` CLI for running health checks and charts without leaving your dev workflow, and PXL scripts — a Python-like scripting language for writing, sharing, and publishing reusable debug sessions as code. Use cases include service health monitoring, Golang performance profiling, HTTP/gRPC request tracing, database query analysis, Kubernetes infrastructure health checks, and canary analysis. Pixie supports popular managed Kubernetes platforms including EKS, GKE, AKS, Minikube, and self-managed clusters. Pixie is ideal for platform engineers, SREs, and developers who want frictionless, code-free observability deeply integrated into their Kubernetes-native workflow.
Key Features
- Zero-Code Auto-Instrumentation: Uses eBPF dynamic probes to automatically capture metrics, events, traces, and logs from your applications without any code changes or redeployment.
- In-Cluster Data Storage: All telemetry data remains entirely within your Kubernetes cluster, eliminating data egress costs and maintaining full data privacy.
- PXL Scriptable Debugging: Write, run, and share debug sessions as code using PXL scripts. Access community-contributed scripts via the Pixienaut ecosystem or publish your own.
- CLI-First Developer Workflow: The `px` command-line interface lets developers run health checks, view charts, and debug production issues without ever leaving their terminal.
- Kubernetes-Native Architecture: Natively supports EKS, GKE, AKS, Minikube, and self-managed clusters. Covers service health, request tracing, DB query profiling, and canary analysis.
Use Cases
- Monitor Kubernetes service health and detect performance regressions without adding instrumentation code to microservices.
- Trace HTTP, gRPC, and database requests across distributed services to identify latency bottlenecks and failed calls.
- Profile Golang applications and analyze slow database queries in production using auto-captured execution data.
- Run canary analysis and compare application behavior before and after deployments using scriptable PXL debug sessions.
- Share reusable debug workflows and observability scripts with your team or the broader Pixienaut open-source community.
Pros
- No Instrumentation Required: eBPF-based auto-instrumentation means teams get full observability in seconds without modifying application code or adding SDKs.
- Data Stays In-Cluster: Telemetry never leaves your Kubernetes environment, reducing cost, latency, and compliance risk associated with external data ingestion.
- Open Source & Community-Driven: As a CNCF sandbox project with a rich community of Pixienauts, teams benefit from community scripts, contributions, and long-term vendor-neutrality.
- Flexible Interaction Modes: Supports CLI, browser UI, and script-based workflows so developers can choose the interface that best fits their process.
Cons
- Kubernetes-Only: Pixie is designed exclusively for Kubernetes environments, making it unsuitable for non-containerized or non-Kubernetes workloads.
- Kernel Version Dependency: Requires Linux kernel v4.14+ and Kubernetes v1.16+, which may be a constraint in legacy or restricted infrastructure environments.
- Learning Curve for PXL Scripts: Advanced debugging and customization require familiarity with PXL, Pixie's custom scripting language, which has a learning curve for new users.
Frequently Asked Questions
What is Pixie and who is it for?
Pixie is an open-source Kubernetes observability and live-debug platform designed for developers, SREs, and platform engineers who need deep visibility into distributed applications without the overhead of manual instrumentation.
How does Pixie instrument applications without code changes?
Pixie uses eBPF (extended Berkeley Packet Filter) dynamic probes that hook into the Linux kernel to capture metrics, events, traces, and logs at the OS level — completely transparently to the application.
Does Pixie send my data to an external service?
No. Pixie runs entirely inside your Kubernetes cluster and stores all telemetry data in-cluster. No customer data is transmitted to external servers, ensuring full data privacy and reducing egress costs.
What Kubernetes and Linux versions does Pixie support?
Pixie requires Kubernetes v1.16 or later and Linux kernel v4.14 or later. It supports EKS, GKE, AKS, Minikube, and self-managed Kubernetes clusters.
Is Pixie free to use?
Yes, Pixie is fully open source and free to use. It is a CNCF sandbox project originally created and contributed by New Relic, Inc., with an active open-source community.
