Performance factors for HP ProLiant Serial Attached Storage (SAS)

12
Figure 8. Random performance comparison of ten10K SFF RAID-5drives and ten15K SFF RAID-5drives
Figure 9 below provides evidence for reading a single sequential stream of data, at higher queue
depths, a RAID-6 array of four SFF drives can approach the same performance achieved by a RAID-5
array of four SFF drives.
Figure 10 indicates RAID-6 single stream sequential write rate is limited to about 67% the rate of the
RAID-5 array. A significant performance variance occurs because in an array configuration consisting
of four drives, two drives in the RAID-6 array are dedicated to parity, while only one of the four drives
in the RAID-5 array is dedicated to parity; (2/3 = 67%).
40%
60%
80%
100%
120%
140%
160%
4KiB RAN
READ Aligned
4KiB RAN
WRITE Aligned
4KiB OLTP
Aligned
64KiB RAN
READ Aligned
64KiB RAN
WRITE Aligned
1MiB RAN
READ Aligned
1MiB RAN
WRITE Aligned
Relative Performance - ( % )
1x10-Drv, P411, fw:3.50, 512MB, R5(256KB), 1-D2700, 1LD
1x10-Drv,P411,fw:3.50,512MB,R5(256KB),1-D2700,1LD
100%= 10K rpm SFF SAS
RAID-5 Random Performance; P411
10-146GB 15K 6Gb SAS SFF vs 10-146GB 10K 6Gb SAS SFF