PowerHA Brings New Data Protection Options For Midsize Shops
February 18, 2013 Alex Woodie
The upcoming IBM i 7.1 TR6 version of PowerHA SystemMirror introduces a new technology that will allow midsize organizations to protect their systems in an entirely new way. The enablement of LUN-level switching in smaller IBM storage servers will allow midsize IBM i shops to take advantage of an HA topology that has been widely used in the Unix and Windows world but until now has only been available to IBM i shops running the largest storage servers.
The TR6 release of PowerHA SystemMirror introduces support for LUN-level switching in three IBM storage area networks (SANs), including the Storwize V7000, Storwize V3700, and SAN Volume Controller (SVC). With LUN-level switching, multiple IBM i server nodes have access to the same pool of DB2 for i, IFS, and other data residing in an iASP on the SAN. If one of the IBM i server nodes in the cluster goes down or is taken offline for maintenance, processing continues on the other IBM i node, providing a high level of continuity.
LUN-level switching, which is enabled on these Fibre Channel SAN devices through support for N_Port ID Virtualization (NPIV) technology, previously had only been available when running PowerHA in conjunction with IBM’s enterprise-class SANs, like the DS8000. The big SANs gained support for LUN-level switching with the initial release of IBM i 7.1 in 2010. Midsize IBM i shops are more likely to adopt the Storwize V7000 or V37000 (the SVC is a higher-end offering), and now they can make LUN-level switching part of their clustering topology for high availability and disaster recovery.
Steve Finnes, the worldwide product offering manager for PowerHA on AIX and IBM i, explains how it works. “By virtue of the fact that a set of LUNs is simultaneously connected to other servers, ownership can be switched between those servers as part of the clustering operation,” he tells IT Jungle. “That’s the construct. In that scenario, there is no replication going on whatsoever. That topology, by the way, is more typical in the AIX world. In the IBM i world, most of our customers would like to be doing replication to another site.”
PowerHA customers have been after him for a while to add this new capability, Finnes says. “That’s what a lot of our customers have been waiting for. They’ve been bugging me about that for quite a while. They wanted to be able to do the same topology that we’ve been doing with the DS8000 for several years,” he says. “What’s great about LUN-level switching in the V7000 is it gives clients in the midmarket space the capability to build the same kind of clusters that our enterprise class customers are using today.”
LUN-level switching is ideally used to boost resilience of a cluster residing in a single data center. If geographic separation between the primary and secondary systems is needed–which is usually the case in DR scenarios–then customers typically augment the local LUNs with a SAN-based data replication solution, either Metro Mirror (for synchronous replication up to 40 kilometers away) or Global Mirror (an asynchronous solution without a distance limitation). Another PowerHA offering called geographic mirroring, which uses an IP-based replication technology, is usually implemented with internal disk on IBM i servers, so it typically wouldn’t be utilized with LUN-level switching, which requires SANs.
The broader application of LUN-level switching gives a bigger range of customers access to a wider array of storage and data resilience topologies, which is the goal of IBM and its PowerHA offering, Finnes says. “We offer a lot of different topologies and scenarios,” he says. “And the storage folks are working on even more technology that’s going to enable more types of topologies.”
For example, Finnes’ team in Rochester is working on getting IBM i support for HyperSwap, which is a mainframe concept that IBM introduced in the AIX versions of PowerHA SystemMirror last October. HyperSwap is a key technology enabling the creation of fault-tolerant storage subsystems, but it is unclear when it will come to IBM i.
“At the end of the day, it comes down to customer discussions and customer requirements. What’s the RPO [recovery point objective]? If you have a recovery point of zero, then you need to have a synchronous component, or you need LUN-level switching,” he says.
The new release of PowerHA SystemMirror brings several other enhancements. All PowerHA functionality on PowerHA, SVC, and Storwize V7000 SANs can be accessed through the PowerHA GUI, which is an extension of IBM Navigator for i. IBM also introduced a new command interface, Work with CAD Monitored Resource Entries (WRKCADMRE), to improve the sorting, filtering, and sorting of MREs [monitored resource entries].
This release gives administrators the capability to add or remove any administrative domain entries whether or not all nodes in a cluster are active. Also, administrative domain, FlashCopy, and copy services functions are now fully cluster enabled, which means work with these functions can be initiated from any active cluster node. Geographic mirroring has been enhanced by enabling administrators to change the synchronization priority while the iASP is varied on. Lastly, PowerHA submitted jobs are now associated with the QCSTJOBD job description instead of the QBATCH job description.