StarRocks, a Linux Foundation project, is a next-generation sub-second MPP OLAP database for full analytics scenarios, including multi-dimensional analytics, real-time analytics, and ad-hoc queries. StarRocks received InfoWorld’s 2023 BOSSIE Award for best open source software.

StarRocks

[ Qustion. Shared Data Mode - Disk Performance ]

Hi, Folks!
In Shared-Data mode, StarRocks relies on PVCs, so I cannot use NVMe disks.
In my environment, the local storage is backed by gp3 EBS volumes. Given this limitation, how should I optimize disk I/O performance in Shared-Data mode?
I want to compare StarRocks with Druid, but Druid uses NVMe instance-store disks, so the difference in disk cache performance, disk I/O, and latency will likely make the comparison unfair or misleading. How can I minimize this performance gap?

• <https://docs.starrocks.io/docs/data_source/data_cache/#cache-medium|Cache Medium>
```If disk performance is not optimal, you can increase the number of disks to achieve load balancing and reduce I/O pressure on individual disks.```
In Shared-Data mode, it seems inevitable that there will be a performance gap compared to NVMe-based systems. I will need to minimize disk I/O as much as possible following the recommended guidelines and use higher-throughput EBS volumes to compensate. :thinking_face:

in shared-data mode, data is persisted in S3, you can use NVMe for local disks cache too for StarRocks. Then the gap will be bridged compared to Druid