While you could manually write shell scripts to fill up disks or use iptables to drop packets, dedicated chaos engineering tools make experiments safer, repeatable, and auditable — especially in Kubernetes environments where manual intervention is impractical at scale.
Chaos Mesh (Open Source, CNCF)
A powerful, cloud-native chaos engineering platform that runs inside Kubernetes. It uses Custom Resource Definitions (CRDs) to define experiments as YAML files, which means your chaos experiments are version-controlled, reviewable, and repeatable — just like any other infrastructure-as-code.
Chaos Mesh supports a wide range of fault types: pod kill, network latency/loss/partition, CPU/memory stress, disk I/O faults, and even time skew (simulating clock drift between nodes). You can schedule experiments to run automatically during business hours and define fine-grained selectors to target specific pods by label.
LitmusChaos (Open Source, CNCF)
Another CNCF project focused on Kubernetes. LitmusChaos provides a library of pre-built "ChaosExperiments" (also defined as CRDs) and a workflow engine for chaining multiple experiments together into complex scenarios. It integrates well with CI/CD pipelines, allowing you to run chaos experiments as part of your deployment pipeline — a practice called Continuous Chaos.
Gremlin (Commercial SaaS)
A popular commercial platform designed to make chaos engineering safe, simple, and enterprise-ready. It provides a polished UI, pre-built attack templates, strong safety features (including a "Halt" button that immediately stops all active attacks across your entire infrastructure), and detailed reporting. Gremlin works across Kubernetes, VMs, bare metal, and cloud services — making it useful for organizations with heterogeneous infrastructure.
AWS Fault Injection Simulator (FIS)
AWS's native chaos engineering service. It integrates directly with AWS services (EC2, ECS, EKS, RDS, etc.) and uses IAM for access control, making it the natural choice for teams running primarily on AWS. You define experiments in JSON/YAML and can trigger them via the AWS Console, CLI, or SDK.
Choosing a tool: For Kubernetes-native teams, Chaos Mesh or LitmusChaos are the natural starting points — they're free, well-documented, and deeply integrated with the Kubernetes API. For multi-cloud or VM-heavy environments, Gremlin's breadth of support is valuable. For AWS-centric teams, FIS reduces operational overhead by eliminating the need to manage a separate chaos platform.