VTrak E-Class Product Manual
The write policy automatically changes to Write Thru. When the battery comes
back online, the write policy automatically changes back to Write Back.
To enable the Adaptive Writeback Cache option, see "Making Controller Settings"
on page 107 or page 156.
Also see "Replacing a Cache Battery" on page 224.
Cache Mirroring
VTrak subsystems with two controllers include a Cache Mirroring feature. Cache
Mirroring causes the local controller to mirror write data to the remote controller.
That means, when there is write data in the cache of the controller managing the
target logical drive, the same write data is copied to the cache of the other
controller as well. This arrangement protects the data from loss, in the event that
the local controller fails before the data is written to the logical drive.
Cache Mirroring works whether the write cache policy of your logical drives is set
to Write Back or Write Through. However, you only realize the advantage of
Cache Mirroring when the write cache policy is set to Write Back.
With Cache Mirroring enabled, any write data in the controller cache that has not
been written to the logical drive, will be written to the logical drive, even if the
controller fails. Enable Cache Mirroring when you require failover/failback
protection.
With Cache Mirroring disabled, any write data in the controller cache that has
not been written to the logical drive, will be lost if the controller fails. On the other
hand, write performance increases because of greater bandwidth. Disable Cache
Mirroring when you require maximum performance.
To use Cache Mirroring you must:
•
Have two controllers in the subsystem
•
Set Redundancy Type to Active-Active.
See "Setting Redundancy for the Subsystem" on page 62 or page 152
•
Enable Cache Mirroring under subsystem settings.
See "Setting Redundancy for the Subsystem" on page 62 or page 152
On subsystems with two controllers, when Cache Mirroring is disabled, LUN
Affinity is enabled automatically. See "LUN Affinity" on page 249
Failover and Failback
When one controller fails, the surviving controller takes over logical drive access
until the failed controller is brought back online or is replaced. For example,
Cache Mirroring is enabled and your logical drives are assigned to Controller 1.
The following actions will happen:
248