Commit 34b48db66e08 (block: remove artificial max_hw_sectors cap) caused performance regressions for streaming I/O workloads across a range of storage from SATA disks to enterprise storage arrays. I was unable to actually show a performance gain for any storage I have access to. However, the patch was introduced to boost performance on software RAID over SATA disks, and I don't have access to such a configuration.
The compromise is to reinstate BLK_DEF_MAX_SECTORS, but to bump it up to 2560 from 1024. This number should allow full stripe writes to 10 data disks with a chunk size of 128KB, and at the same time does not regress the storage I have access to. [PATCH 1/2] Revert "block: remove artifical max_hw_sectors cap" [PATCH 2/2] block: bump BLK_DEF_MAX_SECTORS to 2560 block/blk-settings.c | 4 +++- drivers/block/aoe/aoeblk.c | 2 +- include/linux/blkdev.h | 1 + 3 files changed, 5 insertions(+), 2 deletions(-) -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
