Adoption & Ecosystem

Who uses it

The organizations below are listed in the project's own ADOPTERS.md as production users. Each entry reflects what that file states.

Organisation	Use case	Source
Uber Technologies	2000+ Cadence domains (use cases) across 20+ environments, some hosting 400+ domains: infrastructure rollout, ML training, payments, onboarding	ADOPTERS.md
NetApp (Instaclustr)	Orchestrates maintenance of fleets of tens of thousands of nodes, and offers managed Cadence	ADOPTERS.md
DoorDash	ETA, fulfillment, order management, catalog, and ads	ADOPTERS.md
Cloudera	Core of the control plane: provisioning and backup-restore, including Cloudera Data Warehouse	ADOPTERS.md

Third-party coverage independently names Uber, DoorDash, and Coinbase as users (Instaclustr blog). HashiCorp was an early external adopter in the project's history (ia40 interview).

Adoption signals

Observed from the GitHub API on 2026-06-27 for cadence-workflow/cadence:

• Stars: 9,358
• Forks: 898
• Watchers: 1,418
• Contributors: about 178
• Open issues: 192

The project was accepted into the CNCF Sandbox on 2025-05-22 (CNCF). The latest stable release is v1.4.0 (2026-02-27). Governance is a Technical Steering Committee of four plus a maintainer group (MAINTAINERS.md), with the CNCF as neutral host.

Ecosystem

• Official SDKs: Go and Java. Community Python and Ruby SDKs also exist (README.md:40).
• Web UI: cadence-web, served on localhost:8088.
• DSL layer: iWF runs as a framework on top of Cadence (README.md:42).
• Deployment: the cadence-charts Helm chart, with a guided Kubernetes install path (README.md:32-34).
• Storage and infrastructure: Cassandra, MySQL, PostgreSQL, or SQLite for core state; Elasticsearch, OpenSearch, or Pinot for visibility; Kafka for async workflows. The pluggable backends are loaded in the server entry point (cmd/server/main.go:30-36).

Alternatives

Alternative	Differs by
Temporal	The 2019 fork of Cadence by its original creators, under the MIT license; same durable-execution model, but moved from Thrift to protobuf and from a custom RPC to gRPC (ia40, FAQ).
Netflix Conductor / Conductor OSS	Defines workflows in a JSON DSL rather than native-language code.
Apache Airflow / Argo Workflows	DAG-based batch and data-pipeline schedulers; Cadence is general durable execution with arbitrary control flow and long waits.
AWS Step Functions / Azure Durable Functions	Managed services in the same family; they are the conceptual roots of Cadence rather than self-hosted engines.

Pick Cadence when you want self-hosted durable execution with workflows written as native Go or Java code and you can operate the database it needs. Pick a managed Step Functions or Durable Functions when you want no infrastructure to run. Pick Airflow or Argo Workflows when the work is a scheduled data DAG rather than long-running stateful logic.