chaosd vs Gremlin

Chaosd is free Apache-2.0 software for injecting and recovering faults on physical machines. It can create process, network, CPU or memory pressure, disk, and host faults. Use it as a command-line tool or run it as a background service controlled through HTTP requests. Chaosd is deployed separately from Chaos Mesh; PhysicalMachineChaos experiments require service mode on each targeted physical or virtual machine. The documented environment requires glibc 2.17 or later, and the binary download described is a Linux amd64 tarball. Network experiments using tc or iptables require sudo or root privileges. Deployment prerequisites include tc, ipset, and iptables; source builds additionally list stress-ng and byteman. Experiments can be searched and recovered by UID. TLS certificates are shown for service security, and running without generated certificates is not recommended.

Gremlin is a chaos engineering platform for engineering teams seeking failure modes and ways to validate reliability before failures become outages. It injects faults into services, hosts, containers, and serverless workloads so teams can examine system responses. Blast radius controls limit experiments, which can halt and roll back automatically if issues arise. Teams can sequence experiments into incident scenarios, schedule them, and use GameDay Manager to prepare, run, analyze, and store results. The platform also includes dependency discovery, standardized test suites and reliability scores, disaster recovery testing, detected risks, and organization-wide benchmarking. Documented integrations include Slack, Datadog, Dynatrace, AWS, Azure, Google Cloud Platform, Grafana Cloud k6, Jira, and custom webhooks. Supported environments include public clouds, Linux, Windows, Kubernetes, and on-premises deployments through self-hosted Private Edition. Gremlin states that it is SOC 2 Type II certified and encrypts data in transit with TLS and at rest with AES-256. A 30-day trial is available without a credit card and does not auto-convert. Platform pricing is by custom quote per service; agents, targets, and test volume are sized during the pricing conversation.