Compaq Infortrend Manuale Utente
Redundant Controller
10-17
The two controllers continuously monitor each other. When a controller
detects that the other controller is not responding, the working controller
will immediately take over and disable the failed controller. However, it is
not predictable which one of the controllers should fail. It is necessary to
connect all other interfaces to both controllers so that a surviving controller
can readily continue all the services provided for the RAID system.
detects that the other controller is not responding, the working controller
will immediately take over and disable the failed controller. However, it is
not predictable which one of the controllers should fail. It is necessary to
connect all other interfaces to both controllers so that a surviving controller
can readily continue all the services provided for the RAID system.
Active-to-Active Configuration
Users can freely assign any logical configuration of drives to both or either of
the controllers, then map the logical configurations to the host channel
IDs/LUNs. I/O requests from host computer will then be directed to the
Primary or the Secondary controller accordingly. The total drive capacity
can be divided and equally serviced by both controllers.
the controllers, then map the logical configurations to the host channel
IDs/LUNs. I/O requests from host computer will then be directed to the
Primary or the Secondary controller accordingly. The total drive capacity
can be divided and equally serviced by both controllers.
The active-to-active configuration engages all system resources to
performance. Users may also assign all logical configurations to one
controller and let the other act as a standby.
performance. Users may also assign all logical configurations to one
controller and let the other act as a standby.
Active-to-Standby Configuration
By assigning all the logical configurations of drives to one controller, the
other controller will stay idle and becomes active only when its counterpart
fails.
other controller will stay idle and becomes active only when its counterpart
fails.
Cache Synchronization
The Write-back caching significantly enhances controller performance.
However, if one controller fails in the redundant controller configuration,
data cached in its memory will be lost and data inconsistency might occur
when the existing controller attempts to complete the writes.
However, if one controller fails in the redundant controller configuration,
data cached in its memory will be lost and data inconsistency might occur
when the existing controller attempts to complete the writes.
Data inconsistency can be avoided using one or several of the I/O channels
as the communications path between the controllers. The cached data is
always synchronized in each other's memory. Each controller saves an exact
replica of the cache content on its counterpart. In the event of controller or
power failure, the unfinished writes will be completed by the existing
controller.
as the communications path between the controllers. The cached data is
always synchronized in each other's memory. Each controller saves an exact
replica of the cache content on its counterpart. In the event of controller or
power failure, the unfinished writes will be completed by the existing
controller.
Battery Support
Unfinished writes will be cached in memory in write-back mode. If power to
the system is discontinued, data stored in the cache memory will be lost.
Battery modules can support cache memory for a period of several days
allowing the controller to keep the cached data. When two controllers are
the system is discontinued, data stored in the cache memory will be lost.
Battery modules can support cache memory for a period of several days
allowing the controller to keep the cached data. When two controllers are