| |
Controller
Redundancy
Equally
as important is the necessity to eliminate the controller from being
a single point of failure by employing redundant RAID controllers.
A redundant controller is simply the concept of using two or more
controllers in the same RAID system. By using more than a single
controller to govern and control a logical
volume, access to data contained within
the logical volume carries a far higher guarantee. Some controllers
incorporate a dedicated port for redundant controller connection.
The controllers invariable need to be the same make, model, and
firmware version.
All electronic components are liable to unpredictable failure. If
the controller contains a cache
memory to speed I/O requests, even a simple
memory parity error could cause the controller to temporarily fail
until the memory is replaced. As a RAID system is designed to offer
a stable storage architecture, the more redundancy the RAID system
incorporates the more stability it offers. Controllers are simply
another component. Using redundant controllers increases the availability
of the RAID system and reduces the chances of downtime.
By using redundant controllers together with clustered
servers the chances of downtime may be
reduced even further. Most high-level controllers also have the
ability to operate simultaneously with redundant host server as
well as redundant controllers. Uninterruptible and redundant power
supplies are an example of further redundancy that should also be
incorporated.
By using more than one controller in a dedicated environment, you
can safeguard your system even further and offer greater guarantees
over data availability and up-time. Servers that are attached to
dual RAID controllers are effectively secured against controller
failure - and although it would be exceptional to experience two
controllers failing simultaneously, more than two controllers may
even be used.
The working relationship between the dual RAID controllers is commonly
classified into an active-active
or active-passive
relationship. Each controller may participate in the in servicing
I/O requests and share the workload, or one may be designated the
Master or Primary controller and the other the Slave or Secondary
controller and will be placed on stand-by, actively monitoring the
master for any signs of error or failure in the same manner as a
spare drive
in the array.
Related
topics:
|
