[#2494] feat(spark): Overlapping compression to avoid blocking shuffle writing #2511
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
zuston
changed the title
|
Maybe this is not a best strategy to reduce the compression cost, but this is acceptable. Actually the biggest cost of compression is the frequent memory allocation, if we could reuse the compression buffer in the final phase of rpc sending, I think the performance improvement will be significant. |
zuston
force-pushed
the
overlapCompression
branch
from
a094902 to
49b380c
Compare
|
cc @jerqi |
|
Could you give some data for this improvement? |
Still in evaluating and I will explore the compression improvements |
zuston
force-pushed
the
overlapCompression
branch
from
4309d2d to
6017641
Compare
Updated in the description. |
Got it. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What changes were proposed in this pull request?
Support overlapping compression to avoid block shuffle writing
Why are the changes needed?
for #2494 .
After applying the proposed improvements, the client observed that shuffle performance doubled in a 100GB Terasort benchmark. Specifically, the overall shuffle speed improved by over 100%, significantly reducing the job’s runtime and demonstrating the effectiveness of the optimizations in large-scale sorting workloads.
Does this PR introduce any user-facing change?
Yes.
How was this patch tested?
Will be added later.