Performance factors for HP ProLiant Serial Attached Storage (SAS)

Figure 10. Sequential Write performance comparison between RAID-6 and RAID-5 SFF drives

Figure 11 below shows the difference in random performance between a RAID-6 array of ten 15K

SFF SAS drives and a RAID-5 array with the same ten drives. Note random read performance is

almost identical between the two RAID levels.

The two RAID configurations’ read equivalent performance illustrates that for random read operations,

all drives in the array are available for each random I/O operation.

However, it is different for write operations. For RAID-5, each random write I/O requires reading two

drives (data drive and parity drive) and writing on two drives (data drive and parity drive). For RAID-

6 ,each random write I/O to the array must read three drives (data drive and two parity drives) with

writing operations performed on three drives (data drive and two parity drives). Therefore, with RAID-

6, random write operations are limited to less than 70% of the compared RAID-5 random write rate.

The decision for choosing RAID-5 or RAID-6 is a tradeoff between performance and higher reliability.

40%

60%

80%

100%

120%

140%

160%

64KiB

SEQ WRITE

256KiB

SEQ WRITE

512KiB

SEQ WRITE

1MiB

SEQ WRITE

Relative Performance - ( % )

1x4-Drv, P411, fw:3.50, 512MB, R6(256KB), 1-D2700, 1LD

1x4-Drv,P411, fw:3.50, 512MB, R6(256KB),1-D2700,1LD

RAID-6 vs RAID-5, Sequential Write Performance

4-146GB 15K 6Gb SFF SAS, P411

100%= RAID-5