Setting Up Application Clusters In A Multi-Site San Environment; Using The Primary Site Designation In Two-Site Configurations; Failure Modes And Recovery Steps; Failure Conditions And Resolutions - HP StorageWorks P4000 User Manual

When using VMWare ESX 4 or higher and its native MPIO in a Multi-Site SAN, you cannot configure
more than one subnet and VIP. Multiple paths cannot be routed across subnets by the ESX/ESXi 4
initiator. With VMware ESX, the preferred configuration is a single subnet and with ESX servers
assigned to sites in SAN/iQ.

Setting up application clusters in a Multi-Site SAN environment

Just as you can set up a Multi-Site SAN for data availability even when a complete site is offline, you
can set up application clusters (Microsoft Cluster Server, VMware ESX Cluster, Linux Cluster, and so
forth) so that the cluster nodes are split between the sites. By having the automated failover/failback
capabilities in the SAN, many application cluster technologies can leverage this capability to do
"stretch" application clustering. Consult the application server specific information for information
about setting up clustering to support this type of failover. In general, you must perform three tasks:
1. Set up an application server node in each site.
2. Configure the application server to connect to the volumes on the SAN.
3. If using multiple subnets in the storage cluster, log on to both VIPs in the cluster.
If you are using the HP P4000 DSM for MPIO, see "Using the HP P4000 DSM for MPIO," in the HP
StorageWorks P4000 Windows Solution Pack User Manual.

Using the Primary Site designation in two-site configurations

The Multi-Site SAN software supports designating a particular site as Primary so that it does not go
offline when the secondary site goes offline, or when the network link between sites goes offline. This
is specific to the two-site configuration where, without this designation, there would be no way for
the system to determine which site should remain online in the event of a network separation (classically
known in the clustering/distributed systems world as a "split-brain" system).
To set up this type of configuration:
1. Designate one of the sites as primary.
2. Start a majority of managers in the primary site.
Contact Customer Support for more information.

Failure modes and recovery steps

In general, there are three failure modes that a configuration like this can encounter that you should
be aware of.
the issue.
Table 5 Failure conditions and resolutions
Failure condition
Secondary site offline
Network link between sites fails
Table 5 on page 19 summarizes the failure conditions and what you should do to rectify
Resolution steps
None. The primary site will continue to operate.
None. The primary site will continue to operate.
P4000 Multi-Site HA/DR Solution Pack user guide 19