deepflow
eBPF Observability - Distributed Tracing and Profiling
What is deepflow?
eBPF Observability - Distributed Tracing and Profiling
Overview of deepflow
What it is
DeepFlow is an open-source project that provides deep observability for complex cloud-native and AI applications. It implements zero-code data collection using eBPF for metrics, distributed tracing, request logs, and function profiling. The project integrates SmartEncoding to achieve full-stack correlation and efficient access to all observability data, removing the need for developers to instrument code manually. It offers monitoring and diagnostic capabilities covering from code to infrastructure for DevOps and SRE teams.
Key features
- Universal Map for Any Service: Provides a universal map with zero code via eBPF for production environments, including application services, AI services, and infrastructure services in any language. Supports Wasm plugins for private protocols and calculates full-stack golden signals to pinpoint performance bottlenecks.
- Distributed Tracing for Any Request: Zero-code distributed tracing powered by eBPF supports applications in any language and infrastructures such as gateways, service meshes, databases, message queues, DNS, and NICs. Automatically collects full-stack network performance metrics and file I/O events for each span.
- Continuous Profiling for Any Function: Collects profiling data at a cost below 1% with zero code, plots OnCPU/OffCPU/GPU/Memory/Network function call stack flame graphs, and locates full-stack performance bottlenecks in business functions, library and framework functions, runtime functions, shared library functions, kernel functions, and CUDA functions. Automatically relates profiling data to distributed tracing data.
- Seamless Integration with Popular Stack: Can serve as storage backend for Prometheus, OpenTelemetry, SkyWalking, and Pyroscope. Provides SQL, PromQL, and OLTP APIs to work as data source in popular observability stacks. Injects meta tags for all observability signals including cloud resource, K8s container, K8s labels, K8s annotations, and CMDB business attributes.
- Performance 10x ClickHouse: SmartEncoding injects standardized and pre-encoded meta tags into all observability data, reducing storage overhead by 10x compared to ClickHouse String or LowCard method. Custom tags and observability data are stored separately, enabling almost unlimited dimensions and cardinalities with uncompromised query experience.
Use cases
DeepFlow is designed for complex cloud-native and AI applications in production environments. It provides deep observability without requiring code instrumentation, making it suitable for DevOps and SRE teams to monitor and diagnose performance issues across applications, AI services, and infrastructure. It covers any service, any request, and any function, enabling full-stack performance bottleneck identification and correlation.
Getting started
DeepFlow offers three editions: Community (for developers), Enterprise (for organizations), and Cloud (SaaS, currently in beta). For installation instructions of the Community Edition, please refer to the deployment documentation in the repository README. A complete demo is also available for experience.
License and pricing
The project is licensed under Apache-2.0 and is open source. The Community Edition is free to use. Pricing for the Enterprise and Cloud editions is not stated in the provided materials.
Based on the project's public README and repository metadata.