Beyond Argo CD and Flux: Progressive Delivery with PipeCD

GitOps is great, until your deployment pipeline needs more than “sync this manifest.” Argo CD and Flux have made GitOps mainstream, but progressive delivery, multi-cloud targets, and fine-grained rollout control often require additional tooling or custom workflows in Kubernetes environments.

PipeCD is a CNCF Sandbox project that approaches this differently. Instead of bolting canary and blue-green support on top of a sync engine, PipeCD treats deployment pipelines as first-class citizens: defined in Git, versioned, and fully auditable.

Drawing from contributions to PipeCD’s deployment pipeline and operator components, this talk explores the internals from implementation details rather than only user-facing workflows.

By the end of this talk, attendees will be able to evaluate whether PipeCD fits their delivery stack, understand the architectural tradeoffs compared to Argo CD and Flux, and define a basic progressive delivery pipeline in PipeCD from scratch.

We will cover:

• why GitOps workflows sometimes need progressive delivery primitives

• how PipeCD’s agent-based architecture works under the hood

• defining canary and blue-green deployment pipelines in YAML

• differences in design philosophy across PipeCD, Argo CD, and Flux

• what breaks in production and how PipeCD’s rollback model handles it

• a short walkthrough of a PipeCD deployment pipeline running in a Kubernetes cluster

This is not a “PipeCD is better” talk. It is about understanding the tradeoffs between different GitOps approaches so you can make better infrastructure decisions.

1. Why GitOps sync engines alone are not enough for progressive delivery and where PipeCD fills that gap

2. How PipeCD's agent-based architecture works, the piped component, control plane separation, and what it means operationally

3. What a canary or blue-green pipeline actually looks like in PipeCD YAML, not conceptually

4. Honest tradeoffs between PipeCD, Argo CD, and Flux to make an informed decision for your stack

5. What breaks in real deployments and how PipeCD's rollback model handles it