IBM Rejiggers PowerHA CBU And HyperSwap
October 11, 2016 Alex Woodie
IBM today unveiled several enhancements to PowerHA for i as part of its semi-annual Technology Refresh for the IBM i stack. First up is a new Capacity BackUp (CBU) system that will be more affordable and easier to use. The company also announced that it will officially be supporting PowerHA’s HyperSwap functionality on SAN Volume Controller (SVC) appliances, not just the DS800. Cloud-based IBM i storage and Unix-only resiliency features round out what’s new in PowerHA.
IBM is rolling out a rehashed CBU for Power Systems offering that will be better than last year’s CBU program. IBMers say last year’s CBU offering, which was based on the Model E880, was both too complicated and too expensive.
Under the new CBU offering, IBM i and AIX PowerHA customers can get a second Power Systems box with a discount of 75 percent compared to the cost of the first box. Overall, the new Model E880 CBU system can be as low as 30 percent of the cost of the production E880 system, IBM says. That will alleviate some of the financial concerns that customers had about the CBU program.
IBM is also making it easier to put standby RAM into use on the CBU box. Previously, it was a fairly simple matter to repurpose the CPU resources on the CBU box during a failover. However, it was difficult to get the memory properly licensed and freed for use. With the new Power Enterprise Pools feature, CBU customers can get that standby RAM on the CBU box available almost immediately.
IBM also is officially adding support for the HyperSwap function for PowerHA customers who store their data on SVC appliances. IBM actually announced this feature in May as part of the IBM i 7.3 announcement. But apparently it wasn’t “officially” supported–until November 11, that is.
HyperSwap, to refresh your memory, is a data resiliency feature introduced two-and-a-half years ago with the launch of IBM i version 7.2. The technology uses the short-range, synchronous PPRC data replication protocol (also called Metro Mirror) to replicate DB2 for i data between two IBM storage arrays. In the event of an unexpected failure of the primary array, HyperSwap automatically switches the production IBM i LPAR (or iASPs since IBM i 7.2 TR 3 shipped) from the first array to the second array. The failover is all handled auto-magically by the system, without burdening the administrator with costly demands for attention.
Up to this point, HyperSwap has only supported DS8000s, the big Power-based arrays used by IBM’s largest customers. However, many of IBM’s midsize customers–and even some of its big customers too–are adopting the smaller X86-based storage arrays, like the SVC and, in particular, the Storwize series of arrays, which also run SVC code. With HyperSwap support, now these midrange SVC and Storwize users can benefit from the same high-end data resiliency capabilities that System z mainframe shops have enjoyed for years.
There are a some caveats worth mentioning about HyperSwap, which has confused the IBM i installed base to some extent–not to mention IBM’s business partner community–since it was announced.
The biggest caveat is that it doesn’t really provide high availability, since it’s based on the PPRC protocol, which can only be used by two storage arrays that are within 25 miles of each other. For a large regional disaster, such as Hurricane Matthew’s raking of Florida’s Atlantic coast last weekend, that distance may not be enough. IBM is apparently working to get HyperSwap to support the IP-based protocol used in the geo mirror feature of PowerHA–which is asynchronous and can support two sites anywhere in the world– but as of yet, it’s not supported.
IBM is also talking about how the new cloud-based IBM i storage function that it announced can be used in a PowerHA context. IBMers are intrigued with the possibility of replicating data offsite using the geo mirror protocol, and then using the FlashCopy functionally of PowerHA to quickly take a snapshot of an iASP, and then send that snapshot up to the SoftLayer cloud where it’s kept in a big object-based storage system. This sort of setup could benefit many IBM i shops that are currently cobbling data resilience and backup solutions together. (Look for our coverage on cloud storage in a future story on IT Jungle.)
Lastly, we have the new Geographically Dispersed Resiliency (GDR) feature of PowerHA, which only supports AIX and Linux at the moment. According to IBM, GDR provides a simplified way for customers to restart a virtual AIX or Linux machine at a remote location. The gist of GDR is that customers avoid the need to license software for the backup machine, and also automates disaster recovery operations that many PowerHA shops are doing manually at the moment. Just like PowerHA started off as an AIX-only product and then came to the IBM i later, the GDR functionality is starting there and will be coming to IBM i sometime in the first half of 2017, according to Steve Sibley, director of worldwide product management for IBM’s Power Systems line.