Allocate direct buffers for multipart upload #559
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Outdated suggestions cannot be applied.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.
Allow to allocate multipart upload buffers as direct buffer rather than on the heap.
Reasoning
We try to set a pretty large multipart upload part size on cluster to optimize throughput and reduce S3 requests.
At the same time, we try to keep kafka JVM heap size contained on most kafka installation in order to leave as much memory as possible for the page cache. As a matter of example, we will use 4GB heap size on machines with 64GB available memory.
The consequence of using pretty large multipart upload size on contained JVM heap size is that we can pretty easily run out of heap size if we suddenly have to upload many segments to tiered storage.
The strategy we propose is to allocate multipart buffers in direct memory so that we can more easily configure direct buffer budget.
Usage
We introduced
s3.multipart.upload.direct.buffers
boolean configuration flag (disabled by default).