Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.
Original topic: tikv 磁盘容量大小问题
[TiDB Usage Environment] Production
[TiDB Version] 4.0
[Encountered Issue] Currently, we need to select the appropriate hard disk for the TiDB server. The documentation suggests that the size of PCIe SSD should not exceed 2TB and regular SSD should not exceed 1.5TB. The given reason is to prevent the impact of data recovery in case of disk failure. Assuming I don’t need to consider such an impact, can the disk size of a single KV node exceed 2TB?
The main issue with a single TiKV being too large is the excessive number of regions. A large number of region heartbeats can also affect performance. For inactive regions, the hibernate-regions feature can reduce the number of heartbeats. If there is no impact from heartbeats or there are not too many requests on TiKV, exceeding 2T is not a problem.
Hello, I would like to ask if creating a logical volume from multiple disks will affect the performance of KV?
Using LVM to create RAID 0? It doesn’t have much impact.
There is no RAID, just a logical volume directly.
The best requirement for TiDB is the ext4 file system with the nodealloc mount option. As long as this requirement is met, it is sufficient. In terms of performance, regardless of the disk type, RAID0 or striped volumes are better.
Okay, thank you for the explanation.
This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.
|
username: Hacker_suny
|
username: 
|
username: