[Bug] [broker]Disk iops cannot be higher than 4k,seems to be affected by Direct I/O
- Dominant language
- Java
- Stars
- 15.3k
- Forks
- 3.8k
- Avg merge
- 1d 14h
- Merged PRs (30d)
- 160
Description
### Search before asking
- [X] I searched in the [issues](https://github.com/apache/pulsar/issues) and found nothing similar.
### Version
3 Aliyun ecs,broker and bookie share a disk
disk have 16800 iops
pulsar-version:3.0.0
CentOS Linux release 7.9.2009 (Core)
3.10.0-1160.88.1.el7.x86_64
### Minimal reproduce step
Through the fni tool, when testing disk performance, it is found that turning on/off direct has a great impact on disk iops

### What did you expect to see?
I hope to see that the disk io of pulsar can exceed 4k
### What did you see instead?

### Anything else?
_No response_
### Are you willing to submit a PR?
- [ ] I'm willing to submit a PR!
Contributor guide
Research direction
Start by reproducing the reported disk I/O behavior on Pulsar 3.0.0 with the broker and bookie sharing an Aliyun ECS disk, comparing direct I/O enabled and disabled. Check the broker and bookie disk-I/O paths to identify why throughput appears capped near 4k, then verify the result with the same disk-performance test.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- java
- Domain
- backend, performance
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100