Cadence
cadenceworkflow.io
- Free plan
- Linux
- Web
- API
- Self-hosted
What Cadence is
Cadence is an open-source workflow engine for building distributed applications that need to preserve state and recover from interruptions. Its durable workflows can resume after process restarts, infrastructure failures, or pauses. The project lists official SDKs for Go, Java, and Python, plus a community Ruby client. Scheduling tools support recurring runs, overlap policies, pauses, updates, backfills, and execution history. Deployment choices include embedded SQLite, Docker Compose, and Kubernetes with Helm. For production storage, the project lists Cassandra, MySQL, or PostgreSQL for workflow history, and Elasticsearch, OpenSearch, or Apache Pinot for visibility. Kafka supports search indexing; Prometheus, StatsD, and M3 support metrics. TLS can be configured for backing services, while SDK connections to the frontend support mutual TLS. A custom data converter can compress or encrypt payloads, or replace large payloads with external-storage references. Cadence is free and self-hosted. Workflow code must be deterministic because replay may be used to rebuild state after failure.
Who it suits
Cadence suits application developers building durable workflows and platform engineers operating it as shared infrastructure. Teams need to account for its self-hosted deployment and deterministic-code requirement.
Good to know, and what to watch
- Free open-source plan has no usage limits or per-execution fee.
- Workflows can resume after failures and pauses.
- SDKs are listed for Go, Java, and Python.
- Schedules include backfills and execution history.
- Supports event triggers, retries, observability, and human approvals.
- Deployment is self-hosted.
- Workflow code must be deterministic for replay.
- Ruby client is community-provided.
Plans
Every plan and priceOpen source Cadence
Open source
The verdict
Cadence combines durable workflow execution with scheduling, integrations, and configurable payload protection. It is a fit for teams prepared to operate the infrastructure and write deterministic workflow code.
What the maker publishes
- Deployment
- Deployment options include embedded SQLite, Docker Compose, and Kubernetes with Helm; the page also lists a managed Uber cluster for Uber internal teams. source
- Durable workflows
- Workflows are durable functions that Cadence can resume across process restarts, infrastructure failures, and pauses. source
- Founded
- 2017 source
- Languages
- The project lists official SDKs for Go, Java, and Python, and a community Ruby client. source
- Messaging and metrics
- The integrations page lists Kafka for search indexing and Prometheus, StatsD, or M3 for metrics. source
- Notable constraint
- Workflow code must be deterministic because Cadence may replay it to reconstruct state after a failure. source
- Payload protection
- A custom data converter can compress or encrypt workflow payloads, or replace large payloads with references to external object storage. source
- Scheduling
- Schedules support recurring workflow runs, overlap policies, pausing, updates, backfills, and execution history. source
- Security
- TLS is configurable on backing service connections, and SDK connections to the frontend support mutual TLS. source
- Storage integrations
- The production integrations page lists Cassandra, MySQL, or PostgreSQL for workflow history, and Elasticsearch, OpenSearch, or Apache Pinot for visibility. source
- Support
- The project points users to Stack Overflow, CNCF Slack, GitHub Issues, and community contributors for support. source
- What it does
- Cadence is an open-source workflow engine for building fault-tolerant, stateful distributed applications. source
- Who it is for
- Cadence targets application developers writing workflows and platform engineers operating Cadence as shared infrastructure. source
Checked against the maker's pages on 9 Oct 2026.