By: MartinL

MartinL — Tue, 22 Nov 2011 12:20:00 +0000

What is really missing here is the “real RAID5 performance figures” which are 4K Random Writes. Reads on a redundant RAID5 do not need to go through the RAID stack as soon as fastpath is installed. Sequential writes do but are basically an equiation of compute power and memory bandwidth. Same if you look on the LSI web page performance figures – all marketing figures that avoid what is really important on a typical Web or OLTP server. Additionally, I would want to see performance figures on a RAID5 in non redundant state and during rebuild as again, in that case the RAID stack matters.

By: LostBenji

LostBenji — Sun, 01 May 2011 22:26:27 +0000

Good,quick review. The test rig could be cleaned up a little but thats just trivial and has no bearing on the outcome. As for the LSI card, they are decent performer but still have the appearence of being rough and raw. I would have liked to see the Cache as a DDR III SODIMM to allow users to change or upgrade rather than requiring expensive additions. As for the SSD’s, I feel this review would benefit from using several arrays including the excessively priced C300’s as well as SSD’s in the domestic areas. Also, I understand that the site is for SSD’s but using a mixture of SATA and SAS HDD’s for more comparison. The other area most hide away is the negitive effects of RAIDing SSD’s in respect of TRIM being lost. The only RAID that is supposed to support TRIM in RAID is Intel Matrix with version 9.xx software however, still buggy. It’s nice seing the high speeds but no good if the system is a mess after a month of use.

Comments on: LSI 9265-8i 6Gbps MegaRAID Card RAID 5 Tested! – Final Thoughts

By: MartinL

By: LostBenji