Specifications

24 en | RAID Planning SCSI to SATA RAID Subsystem | Administrator’s Manual
F.01U.027.802 | V1 | 2006.11 Bosch Security Systems
mization scheme and optimize each array/controller for your application. Stripe size and write
policy can be adjusted on a per logical drive basis.
2.1.6 Multi-path, hub, or switch?
Unpredictable situations like a cable coming loose can cause system down time. Fibre Chan-
nel redundant path configurations using a flexible LUN mapping method can guarantee there
is no single point of failure. You may avail a logical drive on two separate host channels and let
the multi-path software running on the host manage the access to this logical drive. The use of
a Fibre Channel switch makes cabling and topology more flexible. Change host-side protocol
and other associated settings to adjust the RAID array to your demands.
2.1.7 Optimization mode
Fig. 2.1 Optimization Setting
You should select an optimization scheme best suited to your applications before configuring
a RAID array. Once the optimization mode is selected, it will be applied to all arrays in the sys-
tem.
Two options are available: Sequential I/Os and Random I/Os.
You may refer to Section 5.2 Caching Parameters and Section 7.2 Caching Parameters for the
stripe size variables and their relation to RAID levels.
Numerous controller parameters are tuned for each optimization mode. Although stripe size
can be adjusted on a per logical drive basis, users are not encouraged to make a change to the
default values.
For example, smaller stripe sizes are ideal for I/Os that are transaction-based and randomly
accessed. However, using the wrong stripe size can cause problems. When an array of the 4KB
stripe size receives files of 128KB size, each drive will have to write many more times to store
data fragments of the size of 4KB.
The default values in optimization modes guarantee the optimal performance for most applica-
tions. For all the controller parameters that are related to system performance and fault- toler-
ance see Section 2.4 Tunable Parameters.
2.1.8 What RAID level?
Different RAID levels provide varying performance features and fault tolerance levels. For
explanation of RAID levels see Section 1 RAID Functions: Introduction.
Data availability comes with the tradeoff that more processing power is consumed because
the RAID subsystem needs to generate/retrieve parity data. For example, in a performance-ori-
ented configuration without availability concerns, you may configure an adequate number of
drives into one or several RAID0 logical drives to maximize the array performance.
2.1.9 Any spare drives?
(Swap Drive Rebuild / Spare Drive Rebuild)
Spare drives allow for the unattended rebuilding of a failed drive, heightening the degree of
fault tolerance. If there is no spare drive, data rebuild has to be manually initiated by replacing
a failed drive with a healthy one.
As is often ignored, a spare drive (whether dedicated or global) must have a capacity no
smaller than the members of a logical drive.