[improve][tiered storage] Reduce cpu usage when offloading the ledger - #15063
Merged
Conversation
--- *Motivation* When offloading a ledger, the BlockAwareSegmentInputStreamImpl will wrap the ledger handler and make it can stream output. Then the JCloud will read the stream as the payload and upload to the storage. In the JCloud implementation, it read the stream with a buffer https://github.com/apache/jclouds/blob/36f351cd18925d2bb27bf7ad2c5d75e555da377a/core/src/main/java/org/jclouds/io/ByteStreams2.java#L68 In the current offload implementation, the read will call multiple times to construct the buffer and then return the data. After implement the read(byte[] b, int off, int len), the cpu usage reduced almost 10%. *Modifications* - Add read(byte[] b, int off, int len) implementation in the BlockAwareSegmentInputStreamImpl
zymap
added
type/enhancement
Member
Author
|
Before the change: After the change: |
horizonzy
reviewed
Apr 8, 2022
horizonzy
reviewed
Apr 8, 2022
hangc0276
reviewed
Apr 8, 2022
horizonzy
reviewed
Apr 8, 2022
Member
Author
|
@hangc0276 @horizonzy Could you please take another look? |
hangc0276
reviewed
Apr 13, 2022
Member
Author
|
/pulsarbot run-failure-checks |
Member
https://github.com/apache/pulsar/pull/15063/checks?check_run_id=6224983508. The first test have fixed recently, I'm not sure is it a new situation. |
Member
Author
|
@horizonzy rebased |
Member
Author
|
ping @eolivelli |
horizonzy
reviewed
May 10, 2022
| } | ||
|
|
||
| if (!entriesByteBuf.isEmpty() | ||
| && bytesReadOffset + entriesByteBuf.get(0).readableBytes() <= blockSize) { |
Member
There was a problem hiding this comment.
Needn't check every time, maybe it just need check then entriesByteBuf's 0 element changes.
Member
|
Add a test for hybrid read way (batch and 1 byte), it should work well. |
horizonzy
reviewed
May 10, 2022
Member
|
👍 |
Member
|
Hi @zymap Sorry I made a mistake just now that the PR was re-opened. |
codelipenghui
pushed a commit
to codelipenghui/incubator-pulsar
that referenced
this pull request
Jun 7, 2022
…apache#15063) * [imporve][tiered storage] Reduce cpu usage when offloading the ledger --- *Motivation* When offloading a ledger, the BlockAwareSegmentInputStreamImpl will wrap the ledger handler and make it can stream output. Then the JCloud will read the stream as the payload and upload to the storage. In the JCloud implementation, it read the stream with a buffer https://github.com/apache/jclouds/blob/36f351cd18925d2bb27bf7ad2c5d75e555da377a/core/src/main/java/org/jclouds/io/ByteStreams2.java#L68 In the current offload implementation, the read will call multiple times to construct the buffer and then return the data. After implement the read(byte[] b, int off, int len), the cpu usage reduced almost 10%. *Modifications* - Add read(byte[] b, int off, int len) implementation in the BlockAwareSegmentInputStreamImpl (cherry picked from commit 938ab7b)
nicoloboschi
pushed a commit
to datastax/pulsar
that referenced
this pull request
Jun 7, 2022
…apache#15063) * [imporve][tiered storage] Reduce cpu usage when offloading the ledger --- *Motivation* When offloading a ledger, the BlockAwareSegmentInputStreamImpl will wrap the ledger handler and make it can stream output. Then the JCloud will read the stream as the payload and upload to the storage. In the JCloud implementation, it read the stream with a buffer https://github.com/apache/jclouds/blob/36f351cd18925d2bb27bf7ad2c5d75e555da377a/core/src/main/java/org/jclouds/io/ByteStreams2.java#L68 In the current offload implementation, the read will call multiple times to construct the buffer and then return the data. After implement the read(byte[] b, int off, int len), the cpu usage reduced almost 10%. *Modifications* - Add read(byte[] b, int off, int len) implementation in the BlockAwareSegmentInputStreamImpl (cherry picked from commit 938ab7b)
mattisonchao
pushed a commit
that referenced
this pull request
Jun 7, 2022
…#15063) * [imporve][tiered storage] Reduce cpu usage when offloading the ledger --- *Motivation* When offloading a ledger, the BlockAwareSegmentInputStreamImpl will wrap the ledger handler and make it can stream output. Then the JCloud will read the stream as the payload and upload to the storage. In the JCloud implementation, it read the stream with a buffer https://github.com/apache/jclouds/blob/36f351cd18925d2bb27bf7ad2c5d75e555da377a/core/src/main/java/org/jclouds/io/ByteStreams2.java#L68 In the current offload implementation, the read will call multiple times to construct the buffer and then return the data. After implement the read(byte[] b, int off, int len), the cpu usage reduced almost 10%. *Modifications* - Add read(byte[] b, int off, int len) implementation in the BlockAwareSegmentInputStreamImpl (cherry picked from commit 938ab7b)
4 tasks
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
7 participants
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.
Motivation
When offloading a ledger, the BlockAwareSegmentInputStreamImpl will
wrap the ledger handler and make it can stream output. Then the JCloud
will read the stream as the payload and upload to the storage.
In the JCloud implementation, it read the stream with a buffer
https://github.com/apache/jclouds/blob/36f351cd18925d2bb27bf7ad2c5d75e555da377a/core/src/main/java/org/jclouds/io/ByteStreams2.java#L68
In the current offload implementation, the read will call multiple times
to construct the buffer and then return the data.
After implement the read(byte[] b, int off, int len), the cpu usage reduced
almost 10%.
Modifications
Verifying this change
(Please pick either of the following options)
This change is a trivial rework / code cleanup without any test coverage.
(or)
This change is already covered by existing tests, such as (please describe tests).
(or)
This change added tests and can be verified as follows:
(example:)
Does this pull request potentially affect one of the following parts:
If
yeswas chosen, please highlight the changesDocumentation
Check the box below or label this PR directly.
Need to update docs?
doc-required(Your PR needs to update docs and you will update later)
no-need-doc(Please explain why)
doc(Your PR contains doc changes)
doc-added(Docs have been already added)