[mihirpandya]

Thanks for the comment. Fair point about Kafka :), although there are good tools now when it comes to deploying Kafka easily (Strimzi, Confluent, etc.).

I like to think of Atlas as connective meshing for Argo which adds a reactive element. Atlas can connect tasks/tests, deployments, and automated stability under one standardized interface. It also adds a lot of automation, picking up feedback/changes at any point during the lifecycle and making changes based on it (rollback, progression, etc).

[johnmarcus]

can't workflows do these things?

[mihirpandya]

Argo Workflows would need to be fully configured with Argo Events and Argo CD to make this happen, specifically with additional Workflows for error processing, multiple sensors for input, and a secondary tier of custom sensors for watching tasks or Argo CD apps past the pipeline run. The workflows (main pipeline and handling pipelines) would likely require custom logic as well. Definitely feasible, but can be complicated.