refactor: parallel write buffer writing for large object disk cache #619
Conversation
This file contains 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
Signed-off-by: MrCroxx <[email protected]>
MrCroxx
added
feature
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #619 +/- ##
==========================================
+ Coverage 81.70% 81.77% +0.06%
==========================================
Files 66 67 +1
Lines 8425 8522 +97
==========================================
+ Hits 6884 6969 +85
- Misses 1541 1553 +12 ☔ View full report in Codecov by Sentry. |
Li0k
reviewed
Jul 19, 2024
foyer-storage/src/large/batch.rs
Outdated
| .map(|group| { | ||
| // TODO(MrCroxx): Refine to logic. | ||
| // Do not filter empty group here. | ||
| // An empty group can be used to trigger makring evictable region in flusher. |
There was a problem hiding this comment.
Sharper sight than typos 🥵.
Signed-off-by: MrCroxx <[email protected]>
MrCroxx
commented
Jul 20, 2024
|
|
||
| rx: mpsc::UnboundedReceiver<Submission<K, V, S>>, | ||
| notify: Arc<Notify>, |
There was a problem hiding this comment.
Replace Notify with CondationVariable later, for the implementation of Notify also includes a mutex.
MrCroxx
commented
Jul 20, 2024
Comment on lines
+303
to
+311
| let mut indices = group.indices; | ||
| for haddr in indices.iter_mut() { | ||
| haddr.address.region = region.id(); | ||
| } | ||
| }; | ||
| try_join(try_join_all(futures), future).await?; | ||
| if let Some(init_time) = batch.init_time.as_ref() { | ||
| metrics.storage_queue_rotate.increment(1); | ||
| metrics.storage_queue_rotate_duration.record(init_time.elapsed()); | ||
| indexer.insert_batch(indices); | ||
| for tx in group.txs { | ||
| let _ = tx.send(Ok(true)); | ||
| } | ||
|
|
There was a problem hiding this comment.
Move notification to a dedicated thread to increase the throughput.
Signed-off-by: MrCroxx <[email protected]>
This was referenced Jul 20, 2024
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's changed and what's your intention?
This PR refined the write model of the large object disk cache.
With the previous design, entries submitted to the same flusher will be serialized in the flusher runner thread (tokio task), whose throughput is limited by the 1 CPU core. With the new design, every entry is serialized in its submit thread, and the serialization between entries can be paralleled.
The parallelism is achieved by allocating with a mutex and waiting with
WaitGroup#615.Checklist
make all(ormake fastinstead if the old tests are not modified) in my local environment.Related issues or PRs (optional)
#596 #615
close #581