Remote copy is an extended function on the IBM 3990 Model 6 Storage Control, along with PTFs on the appropriate levels of MVS/DFP and DFSMS/MVS.

Remote copy offers two options for your disaster recovery and workload migration needs: extended remote copy (XRC) and peer-to-peer remote copy (PPRC). Both options address the problem of unrecoverable data that occurs between the last, safe backup of an application system to a recovery system and the time when the application system fails.

Disasters occur in many forms. Some disasters may happen suddenly and stop all processing at a single point in time. A more likely scenario is the disaster that interrupts operations in stages, occurring over several seconds or even minutes. You must, therefore, plan for recovery from a potential disaster that causes intermittent and gradual system failures. IBM's remote copy function addresses this real-world situation.

Without Remote Copy

The data at the recovery site is not kept current; therefore, any updates made since the last periodic backup may be lost. Without remote copy you must determine what has occurred since the last backup, and then try to manually enter the missing data.

With Remote Copy

Remote copy allows recovery to take place faster and more completely. Remote copy improves the current disaster recovery process by allowing a much shorter recovery time with little or no data loss. Remote copy promotes faster recovery because it constantly updates the records at the recovery site to match the primary application records. In the event of a disaster, data loss is minimized or even eliminated.

The remote copy XRC and PPRC options maintain current copies of your vital application data as follows:

• With PPRC, no DASD data is lost between the last update at the application system and the recovery at the recovery system site.

• With XRC, the only data that is lost is data that is "in transit" between the time when the application system fails and the recovery at the recovery system site.

Remote copy is based upon two systems: a primary application system at one location and a recovery system at another location. Both systems can be located in the same building, or at remote locations. The recovery system only needs to be in place for the time when a recovery is required, and can be a stand-alone system. Each system has specific DASD that will handle data that you have identified as remote copy-managed. In case of a disaster at your primary location, data is recovered from the recovery system DASD by your recovery system.

Once established, remote copy:

• Provides a real-time copy of both SMS and non-SMS-managed data.

• Makes changes to your data on a remote DASD subsystem as you make those changes at your application location.

• Is application independent; you do not need separate copy facilities for multiple applications or databases.

• Supports all DASD data needed for application recovery.

• Runs virtually unattended. Data is written to the primary DASD and copied to the remote secondary DASD without any involvement from the application users.

• If your application system fails, recovery involves a takeover by a recovery system, which may be located miles from the application system.

Note: You may want to automate operations with a program such as NetView, to help switch applications from the application site to the recovery location. Your installation can make this switch with a minimal impact on end users.

In addition to disaster recovery, remote copy is used for workload migration from one location to another. Remote copy offers an excellent way to move, or migrate, data from one set of DASD volumes to another set with minimal impact to continuous availability applications. Using remote copy for workload migration is actually a subset of the disaster recovery function.

Another way to accomplish workload migration is to use the PPRC Dynamic Address Switching (P/DAS) function to redirect application I/O from one set of DASD volumes to another. P/DAS is described in Topic 15, "PPRC Dynamic Address Switching (P/DAS)" in topic 15.0.

Remote copy can copy data from one volume to another if the number of bytes per track and number of tracks per cylinder are the same between the two devices, and the target device has an equal or greater number of cylinders. The only way to copy data between unlike devices or from larger devices to smaller capacity devices is by using DFSMSdss (or DFDSS, or a comparable product) logical dump, restore, or copy functions.

The advantages of using remote copy over other data migration methods include:

• Minimal application down time -- The remote migration copy is made at the same time that the application is updated. When the migration copy has completed, you simply quiesce application programs, ensure that all updates are on the migration volumes, and start up the applications using the new volumes.

• Minimal application impact -- The application's performance is not affected during the migration. Users may continue to update their data normally, with a pause to switch to the new volumes.

• Migration automation -- There is less operator intervention required to perform the migration because there is no need for a dump and restore process.

Depending upon your specific needs, you can choose between one of the two:

(1) remote copy disaster recovery options: extended remote copy (XRC)

(2) peer-to-peer remote copy (PPRC).

These two options differ in several respect Effect on primary application DASD input/output (I/O) performance

Extended remote copy, an enhancement to DFSMS/MVS 1.1.0 and above, is a combined hardware and software solution to the problem of accurate and rapid disaster recovery. XRC also provides a DASD and workload migration solution.

XRC is designed for those sites that:

• Need to maintain the highest levels of performance on their application host system.

• Need to support extended distances between volume copies.

• Can accept a gap of a few seconds between writes on the application host system and the subsequent write updates on the recovery system.

XRC provides an asynchronous copy operation, over distance, with minimal performance impact to application system DASD I/O write operations. XRC supports both ESCON and parallel channel operation. With ESCON channels, the XRC function provides for the recovery system to reside at up to ESCON distances from the application system.

XRC is implemented in a cooperative way between IBM 3990 storage subsystems and DFSMS/MVS host system software. XRC automatically sends copies of updated data to the recovery system. This is done asynchronously to data updates on the application system, and with a minimal increase to DASD write response time at the application.

To maintain high performance at the primary location, XRC allows the primary application's I/O operation to signal completion before receiving confirmation of the write at the recovery system's DASD. Because this secondary copy may be only seconds behind the primary write, there may be little or no data loss if a system failure occurs when data is "in transit" between the two locations.

Peer-to-peer remote copy, available on MVS/DFP 3.1.0 and above, is a hardware solution to the problem of accurate and rapid disaster recovery, and also provides a workload migration and DASD migration solution.

Peer-to-peer remote copy is designed for those sites that:

• Need the recovery system to always be fully current with the primary application system.

• Can accept some performance impact to application write I/O operations at the primary location.

A PPRC data copy to the recovery storage subsystem is synchronous with the primary volume's I/O operation. This means that when the application system writes data to a primary volume's storage subsystem DASD, and this data has been transferred to cache and nonvolatile storage (NVS), the storage control sends channel end status to the application host system channel. The application site storage control then initiates an I/O channel program to the recovery site storage control to transfer the updated data.

The application site storage control returns device end status to the application system when the transfer to the recovery site storage control cache and NVS is complete. The application system then notifies the application program that the operation is complete.

The PPRC copy function does not consider the application system DASD write operation complete until the data sent to the recovery storage subsystem has received channel end and device end status from the secondary volume's storage control. Each application system write to the recovery subsystem causes an increase to the application system response time.