HP StorageWorks Storage Mirroring user's guide (T2558-96073, February 2008)

16 - 2
Configuration requirements
Storage Mirroring failover can be implemented with very little configuration necessary in small or simple networks, but
additional configuration may be required in large or complex environments. Because an infinite number of network
configurations and environments exist, it is difficult to identify all of the possible configuration steps required for failover. You
need to carefully review the configuration requirements, including those for all environments and those sections specific to
your environment.
Review the following configuration sections.
 All environments on page 16-2
 Domain controllers on page 16-2
 Routers on page 16-2
 Cluster configurations on page 16-2
 WINS on page 16-3
 DNS on page 16-5
 Macintosh shares on page 16-9
 NFS shares on page 16-10
All environments
All environments, regardless of the configuration, should meet the following requirements.
 The source and target should be similarly configured (operating system, applications, service packs, patches, and so on)
so that the applications and services running on the source are available on the target after failover.
 Required applications should be preinstalled on the target and left idle so they do not open or modify replicated files.
Failover scripts will start and stop required applications.
Domain controllers
Failover should be implemented on member servers whenever possible. However, there are certain environments where the
use of a domain controller as the source is unavoidable. Domain controllers can be successfully failed over with Storage
Mirroring when necessary, but the domain controller functionality is not failed over.
 Keep in mind that unavailability of some of the FSMO roles (Flexible Single Master Operation) can cause immediate issues
such as the inability to extend the Active Directory schema or to add a domain to a forest.
 If your source is a global catalog server, you should have other global catalog servers throughout the network to ensure
that the failure of source will not impact users
Routers
Depending on the type of router you are using, failover may or may not be advisable. For example, failover across a NAT
router is not advisable, because you would only be able to monitor the IP address of the NAT, not the IP address of the source.
In that case, failover would occur when the NAT router was unavailable, not when the source was unavailable.
Failover across a firewall can be performed. By default, a destination host unreachable message will trigger failover. If you do
not desire this functionality, it can be disabled using the FailoverOnRouteFailure server setting. See the server settings in the
Scripting Guide for more information on this option.
Cluster configurations
If your target is a cluster, you will need to create two failover scripts.
1. On the target cluster, select Start, Program Files, Accessories, NotePad.
2. Type the following case-sensitive command:
Cluster resource “IP_Address_Resource_Name” /OFFLINE
Substitute the name of the IP Address resource assigned to the target virtual server.
3. Save the file as preback.bat on the shared drive of the cluster.
4. Delete that command replace it with the following case-sensitive command:
Cluster group “Group_Name” /ONLINE
Substitute the name of the group on the target virtual server.
5. Save this file as postover.bat on the shared drive of the cluster and close Notepad.