|GDPS||Geographically Dispersed Parallel Sysplex (GDPS)|
|PPRC||Peer to Peer Remote Copy|
|RPO||Recovery Point Objective|
|RTO||Recovery Time Objective|
HyperSwap is a feature of GDPS and is used in a PPRC (Metro Mirror for ESS or Synchronous PPRC) environment. By design it enhances the resilience of Parallel Sysplex by facilitating the immediate switching of PPRC mirrored disk subsystems. The HyperSwap function significantly enhances the ability to provide non-interrupted (continuous) operations by switching I/O operations from the primary to the [PPRC mirrored] secondary disk subsystems in real time without requiring application restarts or system IPLs.
By building on GDPS automation, all aspects of a site or disk subsystem swap (planned or failure) can be accomplished via pre-defined GDPS scripts thus eliminating the primary DASD (disk subsystems) as a single point of failure to provide the next level of continuous operations support.
The HyperSwap function provides the ability to transparently switch an applications I/O operations to the secondary Metro Mirror (PPRC) volumes providing physical channel connectivity (ESCON or FICON) is available from the processor to the secondary subsystems.
This affords the ability to provide continuous operations from a single site or from multiple locations within metro distances. By implementing HyperSwap, disk failures, maintenance functions, and hardware implementations can be endured without incurring a service interruption.
HyperSwap is not implemented in this example as the primary and recovery CPUs are not shown to be participating in a Parallel Sysplex. This is a resilient configuration that supports an RPO (Recovery Point Objective) of zero and an RTO (Recovery Time Objective) of perhaps one to two hours.
The RTO of this environment can be improved by using the facilities of Parallel Sysplex with HyperSwap.
Figure 2 shows a similar hardware configuration but with the addition of a Coupling Facility and Sysplex Timer. With the addition of these two hardware components, a parallel Sysplex has been enabled supporting both processors and storage subsystems.
Metro Mirror (PPRC) continues to function in this example just as it was shown in figure 1. By adding GDPS/HyperSwap in this configuration, the RTO can be reduced to near-Zero if not eliminated entirely.
HyperSwap enables switching from the primary to secondary storage volumes much more rapidly than previously possible. For scheduled activities, such as planned maintenance to the primary storage subsystem(s), HyperSwap can dynamically reconfigure the parallel Sysplex environment to logically swap the PPRC primary and secondary volumes within just a few minutes.
For an unplanned event, such as a complete failure of the primary storage subsystem, HyperSwap re-directs all I/Os from the primary disk volumes to the secondary volumes without incurring any loss of data or requiring any IPLs.
In neither case is it required that the applications be quiesced or restarted, nor that the systems be IPL’d.
From the viewpoint of an application running on one of the processors, there is no outage experienced as the I/O workload is transferred from the primary to the secondary volumes. From the end-user standpoint, this may be viewed as a brief “period of degraded response time” rather than a physical outage where transactions would have to be reentered. The distinction between the two is largely one of semantics and influenced in part by how an outage is measured.
In this configuration, it is possible to achieve continuous operations within a single-site or metro-distance parallel Sysplex.
This document was printed from http://recoveryspecialties.com/