Exploring ChaosBlade: Alibaba’s Open‑Source Chaos Engineering Platform for Cloud‑Native Environments

Project Overview

ChaosBlade is an open‑source chaos engineering project launched by Alibaba in 2019. It consists of the chaosblade experiment tool and the chaosblade‑box platform, aiming to help enterprises address high‑availability challenges in cloud‑native environments.

The experiment tool supports three system platforms, four programming languages, and more than 200 experiment scenarios with over 3,000 parameters, allowing fine‑grained control. The platform can host both its own tool and external tools such as LitmusChaos, and is already used by over 40 enterprises including Industrial and Commercial Bank of China, China Mobile, Xiaomi, and JD.com.

Core Capabilities

Rich experiment scenarios : cover basic resources (CPU, memory, network, disk, process, kernel, file), multi‑language services (Java, C++, NodeJS, Go), and Kubernetes resources (containers, pods, nodes).

Diverse execution methods : white‑box UI, the blade CLI, kubectl, or programmatic APIs.

Easy scenario extension : all scenarios follow a chaos experiment model with separate executors, making them simple to add.

Automated deployment : the tool can be deployed automatically on hosts or clusters without manual steps.

Hosted open‑source tools : the platform can manage mainstream tools like chaosblade and LitmusChaos.

Unified user interface : users operate experiments through a single UI regardless of the underlying tool.

Multi‑dimensional experiments : support experiments from host level to Kubernetes resources and application layer.

Cloud‑native integration : deployed via Helm, integrated with Prometheus monitoring, and compatible with other cloud‑native components.

Architecture Design

The chaosblade‑box architecture (illustrated below) enables automated deployment of hosted tools, unified experiment modeling, and resource targeting through a target manager. Experiments are executed by invoking the appropriate chaos engine, and metrics are collected via Prometheus for future reporting.

Deployment instructions are available at https://github.com/chaosblade-io/chaosblade-box/releases.

Customer Cases

More than 40 enterprises have adopted ChaosBlade, including major Chinese firms such as Industrial and Commercial Bank of China, China Mobile, Xiaomi, and JD.com, using it to improve system resilience.

Future Roadmap

ChaosBlade will continue to build on cloud‑native foundations, delivering a multi‑cluster, multi‑environment, multi‑language chaos engineering platform. The project will expand experiment scenario coverage, support additional Kubernetes resources, standardize multi‑language scenario implementations, simplify deployment, host more tools, provide scenario recommendations, integrate business and system monitoring, and generate comprehensive experiment reports.

Community contributions are welcomed to advance the field of chaos engineering and help enterprises achieve highly available distributed systems.