The function allows failed applications to automatically be recovered onto an available spare server by pre-allocating spare servers to managed servers.
Depending on the server's boot method, one of the three following switchover methods can be used to recover applications on a spare server:
HBA address rename
This method is used in SAN boot environments where servers start from boot disks located in SAN storage arrays. If the primary server fails, its World Wide Name (WWN) is inherited by the spare server, which then automatically starts up from the same SAN disk. This is made possible by the I/O virtualization (*) capabilities of the HBA address rename function, which is able to dynamically reconfigure the WWN of an I/O adapter (HBA).
* Note: Refer to "2.2.9 I/O Virtualization".
Profile exchange
This method is used in environments where servers start from boot disks located in SAN storage arrays or on a storage device connected to the LAN. If the primary server fails, the World Wide Name (WWN), MAC address, boot configuration, and network configuration set in its profile in advance using I/O virtualization using VIOM or ISM are inherited by the spare server, which then automatically starts up from the same boot disk.
For details on profiles, refer to the manuals of ServerView Virtual-IO Manager or ServerView Infrastructure Manager.
* Note: Refer to "2.2.9 I/O Virtualization".
Storage affinity switchover method
This method is used in environments where SPARC M10/M12 and SPARC Enterprise servers start from boot disks located in SAN storage arrays. When a primary server fails in a SAN boot environment, changing the following configuration using storage management software enables access and startup from the same boot disk. When HBA WWNs are fixed, reconfiguring storage devices enables continuation of operations.
Zoning settings for the Fibre Channel switches connected to servers
Host affinity settings for storage CAs
* Note: Refer to "8.1.6 Settings when Switching Over SPARC M10/M12 or SPARC Enterprise Servers".
For the configurations for each redundancy method, refer to "Table: Functions Available for Each Target Operating System" in "2.2 Function Overview" in the "Design Guide VE".
The following LAN switch settings can also be exchanged between primary and spare servers during server switchover.
VLAN
Port groups (For PRIMERGY BX900/BX400 LAN switch blades operating in IBP mode)
Several servers can share one or more common spare servers, irrespective of the kind of servers used (physical or virtual), or the applications that are running on them.
Spare servers can also be shared between physical and virtual servers. This is done by combining Auto-Recovery with the high availability feature provided with the server virtualization software used.
Note that the Auto-Recovery function differs from clustering software (such as PRIMECLUSTER) in the following respect:
Server failure detection
The Auto-Recovery function can detect hardware failures using server management software (such as ServerView Agents) and server management devices (management blades, management boards, or remote management controllers). It cannot detect system slowdowns.
