Reliably

Overview

Reliably is a resilience engineering platform for teams that investigate operational risk through chaos experiments. It suits engineering groups that want experiment design, execution options, and safety controls in one workflow, particularly those already using Chaos Toolkit. Its breadth is appealing, but the free plan’s single seat and narrow cloud-run limits make it better for evaluation than sustained team use.

Key features

Experiment design and review

The builder provides more than 300 actions, including attacks and probes, without requiring a separate experiment language. Teams can also import JSON Chaos Toolkit experiments and reuse custom templates. This lowers the barrier to assembling experiments and migrating existing ones, though template and experiment caps on Free constrain how far a team can build a reusable library.

Scheduling, scores, freshness indicators, timelines, and dashboards help teams organize runs and review activity. Supported targets include AWS, Azure, Google Cloud, Kubernetes, Prometheus, and Spring. Those targets and several execution paths make Reliably flexible across different workflows, but do not establish that every target works with every execution method.

Execution, integrations, and safety

Experiments can run through GitHub Actions, Reliably-managed cloud execution, the CLI, Docker, Kubernetes jobs, or cron jobs. Slack posts experiment activity to a channel in near real time, while Honeycomb receives traces through OpenTelemetry. Other documented integrations include Dynatrace, Google Cloud, Grafana, OpenAI ChatGPT, pre-checks, safeguards, and auto-pause.

Pre-checks can prevent a run from starting, and safeguards can interrupt it when configured conditions fail. These controls matter for teams concerned about experiment blast radius, but they depend on configuration rather than removing the need to plan experiments carefully. Environments store variables and encrypted secrets for use by the app, CLI, or GitHub workflows.

Reliably describes itself as open source, built around an open-source Chaos Toolkit platform, and says its approach avoids lock-in. Its hybrid deployment model and support for self-hosted use may suit teams that need execution choices beyond a hosted cloud run.

Pricing

Reliably uses a freemium model. The paid plans are billed monthly, and the increasing runtime, team, concurrency, and retention allowances are the main reasons to move up.

Free is unusually constrained for collaborative work: it allows just one team member, one concurrent execution, and two weeks of retained data, as well as limiting cloud runs by both monthly minutes and duration. Start removes the experiment cap and extends retention, but still meters cloud runtime. Scale removes the monthly cloud-minute limit and concurrency cap, while retaining a 60-minute execution limit. Enterprise extends retention and adds the optional on-premises installation and support SLA.

Platforms

Reliably supports Linux, macOS, Windows, web, and self-hosted use. Kubernetes support is included, as are cloud fault injection, network fault injection, experiment scheduling, and blast-radius controls. That combination suits teams spanning local tooling and infrastructure workflows, though the exact deployment and execution setup will depend on their chosen path.

Who it's for

Reliably is best suited to engineering teams that want to create and schedule resilience experiments, inspect results, and apply pre-run and in-run controls without adopting a new experiment language. Teams with existing Chaos Toolkit JSON experiments can bring them in, while teams that need more than one seat or meaningful retention should budget for a paid plan. It is less compelling for a solo user who only needs a standalone open-source tool, or for teams whose regular cloud experiments exceed the plan’s execution caps.

Pros and cons

• Pros: The builder’s 300-plus actions, template reuse, and Chaos Toolkit imports support experiment creation without starting from a new language.
• Pros: GitHub Actions, managed cloud runs, CLI, Docker, and Kubernetes execution give teams several ways to fit experiments into their workflows.
• Pros: Pre-checks, safeguards, encrypted secrets, and blast-radius controls address practical safety concerns around running experiments.
• Cons: Free is limited to one seat, five custom experiments, one template, one concurrent run, and two weeks of retention.
• Cons: Free cloud usage is capped at 180 minutes monthly and 10 minutes per run; Start still caps cloud time at 300 monthly minutes.
• Cons: Scale costs £330 a month, and additional members are charged at $20 per month.

Alternatives

Choose Chaos Toolkit if a free, open-source CLI under the Apache License 2.0 is preferable to Reliably’s managed experiment workflow. ChaosBlade is another free open-source option, with support for three system platforms, four programming-language applications, and more than 200 experiment scenarios.

Toxiproxy is a free, MIT-licensed open-source TCP proxy, while Krkn is a free alternative. Pick Chaos Mesh if a free, open-source chaos engineering platform that runs on Kubernetes is the priority. chaosd is a free Apache-2.0 option distributed as a Linux binary and requires glibc v2.17 or later.

FaultMesh offers a free Starter plan for up to three services, with core fault types, basic resilience scoring, and community support; its Growth plan is $49 a month. LitmusChaos offers both a free hosted SaaS plan and a free open-source platform for installation on a cluster. For a broader comparison, see Chaos Engineering Platforms.

Verdict

Reliably is a strong fit for teams that want flexible chaos-experiment creation, multiple execution paths, and explicit safety controls around a Chaos Toolkit-based workflow. The chief reason to choose it is the combination of experiment management and safeguards; the chief reason to look elsewhere is the steep step from a highly limited Free plan to paid cloud capacity, especially for teams needing frequent runs, more seats, or longer retention.