A Design Of Financial Disaster Recovery System On The Mainframe Characterized By Zero-data-loss

Large-scale financial enterprises in China one after another established theirown global data centers with large mainframe system platforms after theyconsolidated their data. Consolidated datacenter has been a place where securitytechnical risks and management risks of information system are centralized. Once adisaster happens to data centers, its local or global branches have to stop theirbusiness and services, while losing important customer data from them. However, ifdisaster recovery backup is in place, information system may be able to recoverfrom damage or disasters. It is of great importance for data security and continuousbusiness operation.Disaster recovery is a procedure which the existing critical data/system and corebusiness may be recovered and resumed within a required period of time by takingtechnical/management actions and using relevant resources after disaster. Generally,core financial system is a 7X24 available system. Therefore, it is required to ensurenot only a sustaining business system, but system and data security in emergencies.And a financial disaster recovery system design on the mainframe which meets therequirement of zero-data-loss and ensures a continuous business operation is aninevitable choice. Disaster recovery can be considered from three technical aspects: Applicationrecovery, Network recovery, Data recovery. To the disaster recovery system in thepaper, it uses the technology of the parallel sysplex to application recovery, uses thetechnology of the class-connection PPRC to data recovery, and uses the technologyof the dynamical route to network recovery. The disaster recovery system not onlyensure the data consistency but also supports the ability of automatic applicationrecovery, it can achieve the aim of zero-data-loss.Parallel sysplex integrates many servers together via Coupling Facility, andCoupling Facility is the key as the heart of the parallel sysplex. It is composed of thespecial coupling facility machine and the macro codes(CFCC) . CFCC is the specialcode which can support the function of the parallel sysplex. Coupling Facility has agreat deal of storage spaces. It can dynamically assign the storage spaces withspecial structure types. It includes three structures: Cache Structure, List Structureand lock Structure. Cache Structure provides the storage spaces of share data. ListStructure is used for sharing queues and message, Lock Structure is used for thesynchronism and collision detection between resources, and provides themanagement of the all resources.The GDPS/PPRC offering is the most function-rich member of the GDPSfamily, providing a disaster recovery capability and, potentially, a continuous ornear continuous availability solution. Because it exploits Metro Mirror,GDPS/PPRC can be deployed in configurations where the production and recoverysites are within relatively close proximity (less than 100 km).GDPS/XRC(Extended Remote Copy) is a combined hardware and softwaresynchronous remote copy solution. Consistency of the data is maintained via theConsistency Group function within the System Data Mover(SDM).Because of theasynchronous nature of XRC, it is possible to have the secondary DASD at greaterdistances than would be acceptable for PPRC.The disaster recovery system design in this paper is a remote recovery solutionon IBM ES9000 mainframe system platforms. It uses sophisticated Parallel Sysplextechniques to build a basic environment and implements advanced GDPS/PPRC andGDPS/XRC techniques between two sites via optical fiber bandwidth multiplexing.The system include production center SITE1, backup centre SITE2 in the same cityand remote backup centre REMOTE. In production center SITE1, the mainframesystem has four system member and a GDPS control system( k system), and thesame system architecture as the SITE2. Parallel Sysplex can use not only the CFhost but also the logical partition which can be installed as the internal couplingfacility as the backup of CF. system timer TIMER provides the time synchronism ofthe parallel system. The four system member and coupling facility CF1 in SITE1compose a parallel coupling system which provides the production environment.The backup center in SITE2 provides the system backup environment in same city.The external coupling facility of Parallel Sysplex between two sites datacentersconnect via the optical fiber. All system in SITE1 and SITE2 build up the GDPSsystem. The GDPS control system (k system) controls the whole GDPS systemenvironment. Thus, the users control the physical-separate mainframe systems asone logical system. Any one system can share the workload of the whole systems. Itis to said that all the systems in Parallel Sysplex could be regarded as one logicalsystem. So if one system shuts down, the workload of it can be taken over byanother running system. This design integrates Parallel Sysplex and GDPStechniques to ensure the disaster recovery ability of the permanently-running systemwith loads balanced.The Disaster Recovery system is very complex, so it needs a great amounthardware devices and a great deal money. It needs high costs, but the speed ofdisaster recovery is satisfied, it already reaches the highest class of the internationaldisaster backup class SHARE 78. The design is comfortable to the disaster recoveryconstruction of large-scale financial and insurance enterprises which require strictlyreal-time data transaction.