IBM Offers Tech Preview for iASP-Based HyperSwap
October 21, 2015 Alex Woodie
The future of HyperSwap on IBM i became a little more clear this month with IBM‘s October 5 announcement. That’s when IBM shared details on a new Technology Preview for HyperSwap supporting independent auxiliary storage pools (iASPs), which will allow more fine-grained failover of IBM i environments that are protected in PowerHA Enterprise Edition environments.
HyperSwap is a data resiliency feature that IBM debuted in its PowerHA Express Edition last year with the launch of IBM i version 7.2. The technology, which was originally developed years ago for System z mainframes, relies on the peer-to-peer remote copy (PPRC) synchronous data replication protocol (i.e. Metro Mirror) to replicate DB2 for i data between two DS8000-series arrays sitting in the same data center or campus.
In the event of an unexpected failure of the primary array, HyperSwap automatically switches the production IBM i LPAR to the secondary array, using IBM’s live partition mobility (LPM) functionality. (The secondary array must be located within 25 miles because it’s riding on PPRC.) The role swap happens very quickly–within two to three minutes, according to IBM. The IBM i applications sitting above all this just keep running, with no OS-level role swaps, as required with logical replication solutions.
One of the criticisms of HyperSwap when it first came out was that it required the entire system to be moved over. That’s because the HyperSwap relationship between the server and the DS8000 array was defined at the system level, which meant that all of SYSBAS and any defined iASPs had to be moved.
That will no longer be a limitation of HyperSwap when iASP level switching becomes a fully supported in the OS and PowerHA, IBM’s Jenny Dervin explains in an October 13 developerWorks article:
“In a future 7.2 technology refresh, PowerHA Enterprise Edition will support DS8000 HyperSwap at an IASP level,” she writes. “Now, the HyperSwap relationship can be defined on an ASP within the partition, allowing SYSBAS and the IASPs to be in separate HyperSwap relationships. This is necessary because for a truly resilient environment, SYSBAS may be replicated in a different direction than the IASP.
“With this support, HyperSwap can be used for near zero downtime in the event of a planned or unplanned storage outage, and can be combined with PowerHA LUN switching technology to provide a minimal downtime solution for planned and unplanned server outages,” Dervin continues.
To be eligible for the Tech Preview, an IBM i shop must be using iASPs for data replication with PowerHA SystemMirror for i Enterprise Edition. They also must be running the new IBM i 7.2 TR3 release, have a second partition configured as the target node in a PowerHA switched LUN configuration, and have two DS8800 (or above) storage servers.
While HyperSwap debuted last year in PowerHA Express Edition, it now appears IBM is moving it to Enterprise Edition (at least that’s where the Tech Preview will take place). This makes sense, considering that only the biggest IBM i shops will be using DS8800 arrays.