Originally posted by: SystemAdmin
We are planning the implementation of a PowerHA 6.1 geographic cluster, with two sites (local and remote), using asynchronous GLVM mirroring.
The servers are JS43 blades. Each blade is divided into several AIX 6.1 LPARs, running under VIOS and managed by IVM.
The storage on both sites is EMC CLARiiON.
We decided on asynchronous GLVM mirroring because according to the PowerHA 6.1 documentation, EMC SRDF support is limited to DMX-3, DMX-4, and V-MAX, and VIOS and NPIV storage are also not supported.
According to the PowerHA 6.1 documentation, Asynchronous GLVM mirroring is only supported on Non Concurrent Scalable Volume Groups with mirror pools set to be super strict.
However, according to PowerHA support announcement for NPIV
(
http://www-03.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/FLASH10691 ) PowerHA support for Virtual SCSI/NPIV requires the volume group to be defined as “Enhanced Concurrent Mode".
We have a few technical questions:
1. Is this kind of solution supported / tested ?
2. What are the planning considerations for the Non Concurrent Volume Groups ?
Can we use either Virtual SCSI or NPIV ?
Should we set any hdisk attributes in the VIOS LPAR in case of Virtual SCSI (SCSI reserve policy , etc.) ?
Should we set any hdisk attributes in the client LPARs in case of Virtual SCSI or NPIV ?
3. Since Non Concurrent Volume Groups don't support disk heartbeat, what methods
can we use for NON IP heartbeat - both between nodes in the local site, and between
local and remote sites ?
#PowerHA-(Formerly-known-as-HACMP)-Technical-Forum#PowerHAforAIX