09-06-2010 05:06 AM
I am trying to measure the seq 100%read performance of Intel X25V SSD on linux machine. Issue that I see here is that I am getting different IOPS with different states of the disk i.e consider the below two cases
1) erase the disk -> run rand 100%write -> seq 100% read
2) erase the disk -> run seq 100% write -> seq 100% read
IOPS are different in both the above cases (varying by 5K to 6K). Same behaviour is seen even with seq read done after some seq/rand writes.
Is this behavriour due to wear-leveling done in SSD ? If yes, what is the best way to capture seq 100% read performance ?
Am I missing anything in setup ?
Thank you in advance.
Thanks & Regards,
09-07-2010 03:08 AM