[BEAM-7745] StreamingSideInputDoFnRunner/StreamingSideInputFetcher have suboptimal state access pattern during normal operation

• Type: Improvement
• Status: Resolved
• Priority: P3
• Resolution: Fixed
• Fix Version/s: 2.33.0
• Component/s: runner-dataflow
• Assignee: Steve Niemitz
• Reporter: Steve Niemitz
• Created: 15/Jul/19 18:13
• Updated: 16/Aug/21 18:28
• Resolved: 16/Aug/21 18:28

Description

I spent some time tracking down sources of uncached state fetches in my job, and one large category was the interaction of StreamingSideInputDoFnRunner + StreamingSideInputFetcher.

Basically, during standard operations, when the main input is NOT blocked by the side input, the side input fetcher will perform an uncached state read for every input element.  Changing it to cache the blockedMap state gave me a ~30-40% increase in throughput in my job.

The interaction is a little complicated, and there's a couple optimizations here I can see.

 

Primarily, the blockedMap is only persisted if it is non-empty.  Because the WindmillStateCache won't cache a null value, this means that the "nothing is blocked" signal is never actually cached, and will issue a state read to windmill for each input element.  The solution here seems like it is to persist an empty map rather than a null when there are no blocked elements.

 

Issue Links

• links to: GitHub Pull Request #15235