Skip to content

Reduce File IO for BucketList writes #4696

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

Open
SirTyson opened this issue Apr 23, 2025 · 0 comments
Open

Reduce File IO for BucketList writes #4696

SirTyson opened this issue Apr 23, 2025 · 0 comments
Labels

Comments

@SirTyson
Copy link
Contributor

Currently, we do lots of unnecessary file IO when writing a new batch to the BucketList. Specifically, we create a fresh bucket from a vector of entries, write it to disk, and open the file to index the bucket. Then we open the new bucket file and the current level 0 curr bucket file and kick of a very low priority background thread. This background thread does a merge via file IO, then again reindexes the merge output. During all this, we are blocking the main thread, waiting on this low priority background work.

Instead of launching a background thread, we should do all this work on the main thread. We should also only index buckets once, and since we have all the entries in memory already anyway, we should avoid file IO for level 0 merges.

@SirTyson SirTyson added the bug label Apr 23, 2025
@SirTyson SirTyson mentioned this issue Apr 23, 2025
6 tasks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

1 participant