-
Notifications
You must be signed in to change notification settings - Fork 50
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Pipeline architecture v2 (preview) #1913
Conversation
Co-authored-by: Haris Osmanagić <haris@meroxa.io>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nice work! Mostly questions in this review, so I can have a more meaningful review next time.:)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We're going to ship this as part of 0.12.2 behind a feature flag. We'll deal with the optimizations later.
For the record, I just resolved all open conversations so we could release this PR. |
Created issue to address what was remaining #1951 |
Description
This PR adds a new internal pipeline architecture which in conjunction with the SDK changes (ConduitIO/conduit-connector-sdk#193) improves the pipeline throughput up to 5 times (from ~50k rec/s to ~250k rec/s). The new architecture uses a single worker per source connector which runs in its own goroutine and processes records in batches end-to-end. It lowers the overhead of passing records through channels between nodes, because the whole pipeline is executed in the same goroutine from start to end.
The improved pipeline architecture is disabled by default and can be enabled using the flag
-preview.pipeline-arch-v2
. Note that the new architecture currently supports only 1 source and 1 destination per pipeline.Important
This is an isolated change, the functionality does not change unless the user supplies the flag
-preview.pipeline-arch-v2
.Quick checks