./fio --filename=/root/bla --direct=1 --rw=randrw --refill_buffers --norandommap --randrepeat=0 --ioengine=libaio --bs=4k --rwmixwrite=100 --iodepth=16 --numjobs=16 --runtime=60 --group_reporting --name=4ktest --size=20G
This is how I test the write IOPS ^
Are the variables good to get results that matter?
I'm asking because I can get 80K write IOPS on a Crucial MX100 512GB SSD directly connected to the motherboard while I can only get 17K write IOPS on a raid array that has 4 x M500 480GB and 4 x MX100 512GB SSD on RAID 10 running on a LSI 2108 chip with drive caches enabled and RAID card cache enabled as well.
Does 2108 really suck this hard or is my command not an accurate way to determine the random 4k write IOPS with 16QD?
This is how I test the write IOPS ^
Are the variables good to get results that matter?
I'm asking because I can get 80K write IOPS on a Crucial MX100 512GB SSD directly connected to the motherboard while I can only get 17K write IOPS on a raid array that has 4 x M500 480GB and 4 x MX100 512GB SSD on RAID 10 running on a LSI 2108 chip with drive caches enabled and RAID card cache enabled as well.
Does 2108 really suck this hard or is my command not an accurate way to determine the random 4k write IOPS with 16QD?
Last edited by a moderator: