High Availability Cluster Multi-Processing

Learn about High Availability Cluster Multi-Processing (HACMP™) in the Virtual I/O Server.

HACMP supports certain configurations that utilize the Virtual I/O Server, virtual SCSI and virtual networking capabilities. For the most recent support and configuration information, see the HACMP for System p® Web site. For HACMP documentation, see High Availability Cluster Multi-Processing for AIX®.

For IBM® i client partitions, you must use mirroring to create redundancy. For details, see Mirroring for client logical partitions.

HACMP and virtual SCSI

Be aware of the following considerations when implementing HACMP and virtual SCSI:

The volume group must be defined as Enhanced Concurrent Mode. Enhanced Concurrent Mode is the preferred mode for sharing volume groups in HACMP clusters because volumes are accessible by multiple HACMP nodes. If file systems are used on the standby nodes, those file systems are not mounted until the point of failover. If shared volumes are accessed directly (without file systems) in Enhanced Concurrent Mode, these volumes are accessible from multiple nodes, and as a result, access must be controlled at a higher layer.
If any one cluster node accesses shared volumes through virtual SCSI, then all nodes must. This means that disks cannot be shared between a logical partition using virtual SCSI and a node directly accessing those disks.
All volume group configuration and maintenance on these shared disks is done from the HACMP nodes, not from the Virtual I/O Server.

HACMP and virtual Ethernet

Be aware of the following considerations when implementing HACMP and virtual Ethernet:

IP Address Takeover (IPAT) by way of aliasing must be used. IPAT by way of Replacement and MAC Address Takeover are not supported.
Avoid using the HACMP PCI Hot Plug facility in a Virtual I/O Server environment. PCI Hot Plug operations are available through the Virtual I/O Server. When an HACMP node is using virtual I/O, the HACMP PCI Hot Plug facility is not meaningful because the I/O adapters are virtual rather than physical.
All virtual Ethernet interfaces defined to HACMP should be treated as single-adapter networks. In particular, you must use the ping_client_list attribute to monitor and detect failure of the network interfaces.
If the Virtual I/O Server has multiple physical interfaces on the same network, or if there are two or more HACMP nodes using the Virtual I/O Server in the same frame, HACMP is not informed of, and does not react to, single physical interface failures. This does not limit the availability of the entire cluster because the Virtual I/O Server routes traffic around the failure.
If the Virtual I/O Server has only a single physical interface on a network, failure of that physical interface is detected by HACMP. However, that failure isolates the node from the network.