Sonnet Technologies Web Management Tool - Manual (Page 16)

10
Creating a Custom Setup (continued)
9. Click Next.
10. Choose the number of partitions for the RAID group. See
Figure 13 on page 15.
• A RAID group may have several Terabytes of total data
capacity because of the size of the included drives.
Partitions allow you to break up large RAID groups into
smaller, more manageable groups.
• Most host systems can address only 2TB per LUN.
Partitioning increases storage efficiency by providing more
LUNs without using lower capacity RAID groups.
• Partitioning allows the creation of multiple logical
volumes.
Note: If you don't want to use partitions, click the All Unallocated
button.
a. Click Create One Partition, or enter the desired partition
size for the first partition from the available RAID group
capacity.
b. Click Create.
c. If you have created more than one partition, repeat entering
the partition size and clicking Create as often as you need
to partition the remaining capacity. Whenever you have
completed designating partitions, click
All Unallocated to
put all the remaining capacity into one partition.
11. Click Next. The storage capacity is allocated.
12. RAID partitions are mapped onto the Fibre Channel network
as FC LUNs (SCSI-FCP LUNs). Select the method you wish to
use to map the partitions. See
Figure 14 on page 16.
• If you select Auto, all mapping for all RAID groups
attached to the RX1600Fibre is changed, destroying any
previous mapping.
• If you do not wish to change the mapping of your other
RAID groups, select Manual. Manual mapping allows you
to make LUN assignments for each RAID partition in the
selected RAID group.
a. From the RAID Configuration page presented, under
Select the mapping method, click the Manual radio
button.
b. Click any partition to map that partition to a Port and
LUN.
13. Choose an Initialization method (refer to Initialization on
page 7).
1.2 Configure Storage into RAID Groups
14. Choose the Interleave by clicking the drop-down box and
selecting a value. See Figure 15 on page 16.
15. Select a Sector Size. The RAID group sector size must be
evenly divisible by the sector size of any member disk.
• 512 bytes is the default size for most operating systems.
• For Windows XP (32-bit support) select 4KB sectors
to enable large volume support (greater than 2TB, up to
16TB).
16. Select a SpeedRead feature. SpeedRead looks ahead during
reads and stores the data in cache memory. The optimum
setting depends on your actual I/O and storage. You may
adjust this setting later.
• Enabling SpeedRead may boost performance when you
are running video playback and other applications which
access data sequentially.
• Disabling SpeedRead is a better choice for audio
applications.
• SpeedRead Auto is usually the best choice for database
applications.
17. Choose a Prefetch option. The number of extra stripes that
are read when the SpeedRead setting is set to enabled or
auto.
18. Choose an Auto-Rebuild feature if it is available for your
RAID configuration (refer to Auto-Rebuild on page 7).
19. Choose a Rebuild Priority level. Rebuild Priority allows
you to determine whether rebuild or I/O transactions take
precedence during rebuild operations. If you choose low
priority, for example, rebuilds take longer but the rebuild has
minimal impact on performance.
Support Note: The interleave value chosen when
creating a RAID group makes a significant impact on
performance. Fusion RAID storage systems shipped from Sonnet
with pre-installed hard disk drives are now optimized for use
with video editing (larger files) applications, typically with an
interleave value of 512KB or 1MB selected. If you intend to
use your storage system primarily for storage of smaller files
(database, office documents, etc.), choose a smaller interleave
value of 64KB or 128KB.
Support Note: Rebuild priority affects the performance
of your Fusion storage system when a drive is replaced and
a degraded RAID group is rebuilt. Selecting Low rebuild priority
enables you to continue working at the best performance level
possible, but the RAID group will take much longer to rebuild.