Filter
Top Products
Raidtec FS3102 Hardware Installation Guide Version 1.0 29
If there are insufficient disks to form an entire new row, the RAID will be expanded by adding columns to the
stripe. This involves adding extra disks to each row (this will be a single row unless the RAID has already been
expanded by adding an additional row). When this is done, a re-striping operation is necessary. The system reads
each stripe in turn and writes it to an expanded stripe. This is a time consuming operation.
NOTE: Expansion by column will expand the RAID using the same disk capacity that was used in the existing
columns. If the disks that are being used to expand the RAID have smaller capacities than the columns already in
the RAID, the expansion will not be allowed. If the disks that are being used to expand the RAID have larger
capacities than the columns already in the RAID, the additional capacity on the larger disks will be lost.
RAID Verify
A RAID level 5 may be checked for valid parity by initiating a verify operation. The system reads each stripe and
checks for valid parity. At the end of the verify process the system sends an Alert if any stripes were found with
invalid parity.
Stripe Size
Each RAID can consist of up to a maximum of 12 disks.
Stripe Unit Size
The stripe unit size is set by the profile. See the Raidtec Manager specification for more details.
Caching
The Raidtec FS3102 comes with 256 MB of cache memory as standard. The system supports up to 1 GB as an
option. The cache memory is used for both read and write data. There are three kinds of caching implemented by
the system firmware.
Read Cached Data
The system records the number of read and write commands that occur in the same physical stripe. When the
number of commands exceeds a preset threshold, the system will allocate storage space to cache this data. This
increases performance because subsequent read operations can be serviced from memory instead of reading from
the disk.
Streaming Read Data
The system records the starting block and block size of each SCSI command received. When the start block of a
new command matches the last block of a previous command the system detects that there is a stream of
commands in progress and prefetches data up to 16 times the size of the original command. This increases
performance by lowering the latency of subsequent commands in the stream.
Write Cached Data
When write caching is enabled, good status is returned immediately, before the data is written to disk. This
increases performance in two ways. Firstly it lowers command latency to the host. Secondly, by coalescing
contiguous write data into larger segments before it is written to disk, seeks are reduced and bandwidths increased.