RAID 5 Guidelines

RAID 5 can withstand only a single device failure.

A RAID 1 (mirror) volume can withstand multiple device failures in some cases (for example, if the multiple failed devices are all on the same submirror). A RAID 5 volume can only withstand a single device failure. A RAID 0 volume cannot withstand any device failures.

A RAID 5 volume provides good read performance if no error conditions, and poor read performance under error conditions.

When a device fails in a RAID 5 volume, read performance suffers because multiple I/O operations are required to regenerate the data from the data and parity on the existing drives. RAID 1 (mirror) volumes do not suffer the same degradation in performance when a device fails.

RAID 5 volumes can cause poor write performance.

In a RAID 5 volume, parity must be calculated and both data and parity must be stored for each write operation. Because of the multiple I/O operations required to do this, RAID 5 write performance is generally reduced. In mirrored metadevices, the data must be written to multiple mirrors, but mirrored performance in write-intensive applications is still much better than in RAID 5 volumes.

RAID 5 volumes involves a lower hardware cost than RAID 1 (mirror) volumes.

RAID 5 volumes have a lower hardware cost than RAID 1 (mirror) volumes. Creating RAID 1 volumes requires twice the disk storage (for a two-way mirror). In a RAID 5 volume, the amount required to store the parity is: 1/#-disks.

RAID 5 volumes can not be used for existing file systems.

You can not encapsulate an existing file system in a RAID 5 volume (you must backup and restore).

RAID 5 Volumes and RAID 0 (Stripe) Volumes

Generally speaking, RAID 0 (stripe) performance is better than RAID 5 volumes. RAID 5 volume read performance is comparable or slightly slower than RAID 5 performance and is substantially lower than RAID 0 (stripe) volume performance for write operations, because the RAID 5 volume requires multiple I/O operations to calculate and store the parity.

Write intensive applications: Because of the read-modify-write nature of RAID 5 volumes, volumes with greater than about 20 percent writes should probably not be RAID 5 volumes. If data protection is required, consider creating RAID 1 volumes.

RAID 5 volumes writes will never be as fast as RAID 1 (mirror) writes, which in turn will never be as fast as unprotected writes. The NVRAM cache on the SPARCstorage Array closes the gap between RAID 5 and RAID 1 configurations.

Full Stripe Writes: RAID 5 read performance is always good (unless the volume has suffered a disk failure and is operating in degraded mode), but write performance suffers because of the read-modify-write nature of RAID 5 volumes.

In particular, when writes are less than a full stripe width or do not align with a stripe, multiple I/Os (a read-modify-write sequence) are required. First, the old data and parity are read into buffers. Next, the parity is modified (XOR's are performed between data and parity to calculate the new parity--first the old data is logically subtracted from the parity and then the new data is logically added to the parity), and the new parity and data are stored to a log. Finally, the new parity and new data are written to the data stripe units.

Performance in degraded mode: When a slice of a RAID 5 volume fails, the parity is used to reconstruct the data; this requires reading from every column of the RAID 5 volume. The more slices assigned to the RAID 5 volume, the longer read and write operations (including resyncing the RAID 5 volume) will take when I/O maps to the failed device.