Dell PowerVault MD3000 - Manual (Page 49)

Using Your RAID Enclosure
49
Hardware Redundancy and Failover
This section provides basic information that will help you determine how to connect a host to the RAID
controller modules in a RAID enclosure in order to achieve redundancy and failover.
NOTE:
The best practice for ensuring data access is to avoid a configuration with a single point of failure. A single
point of failure is any component or path that is not duplicated (redundant) or whose failure can cause loss of data
access.
Redundancy and Non-Redundancy
Redundancy means that a storage array has duplicate components, or alternate ways to provide essential
services. This redundancy ensures access to the virtual disks in case of a failure of a single component.
In a typical RAID configuration, most of the components are redundant; however, the configuration
itself might not be fully redundant. For example, there might be components whose individual failure
would cause loss of access to the virtual disks. In a fully redundant storage array, all components must be
duplicated. In addition, the storage array must be configured in such a way as to ensure that duplicate
components can be accessed in case of a failure. The manner in which the storage array is cabled is an
essential component of creating a successfully configured redundant storage array.
Multi-Path Software
Multi-path software (also referred to as the failover driver) is software resident on the host server that
provides management of the redundant data path between the server and the RAID enclosure. For the
multi-path software to correctly manage a redundant path, the configuration must provide for redundant
HBAs and cabling.
The multi-path software identifies the existence of multiple paths to a virtual disk and establishes a
preferred path to that disk. If any component in the preferred path fails, the multi-path software
automatically re-routes I/O requests to the alternate path so that the storage array continues to operate
without interruption.
NOTE:
Multi-path software is included with MD Storage Manager.
Preferred and Alternate Controllers and Paths
A preferred controller is a RAID controller module designated as the owner of a virtual disk or disk group.
The preferred controller is automatically selected by MD Storage Manager when a virtual disk is created.
You can change the preferred RAID controller module owner of a virtual disk after it is created.
Ownership of a virtual disk is moved from the preferred controller to the second controller (also called
the alternate controller) when the preferred controller is:
• Physically removed
• Updating firmware
• Involved in a fatal event that caused failover over to the alternate controller