Understanding the Benefits of Placing Your Database Data under System Management

Database data benefits from system management in the following ways:

Database data can benefit from data isolation achieved through the use of SMS storage groups.
SMS services help you automate allocation of production database data sets (as well as production data bases from end user and test data bases) on separately managed volumes. You should isolate database data on separate volumes because of its high performance and availability requirements, its specialized backup and recovery management procedures, and its migration and expiration criteria.
Both production and end-user databases can benefit from the improved performance of enhanced dynamic cache management.
Using SMS dynamic cache management lets you specify a hierarchy of performance for database data, based on your knowledge of the online applications' use of databases and the requirements of database data types, such as recovery or system data sets. If you are already using volume-level caching and the extended functions of a cache-capable 3990 storage control, you can use the data set-level caching provided by SMS dynamic cache management to improve performance and increase automation of performance management.
Production databases can benefit from improved SMS allocation algorithms.
SMS tries to balance allocations in (and across) storage groups that have large numbers of volumes. It also uses a randomization technique for volume selection from the secondary volume list in order to avoid skews resulting from the addition of new volumes, or owing to differences in workload between the time of allocation and the time of actual use. With SMS allocation, users do not have to worry about specifying volume serial numbers or ensuring that the specified volumes have adequate free space. Additionally, with SMS allocation algorithms, volumes are not fragmented too much for the request for space to be satisfied. While some production databases might require special placement of data for critical performance, or for separation of software striped data (as in DB2®), all other data should benefit from SMS allocation algorithms.
Database data backup processes can benefit from point-in-time copy using either concurrent copy or virtual concurrent copy.
For concurrent copy, virtual concurrent copy, or flash copy, implement this high-performance backup with DFSMSdss and a cache-capable 3990 storage control. It is supported by CICS®, IMS™, and DB2. With concurrent copy, you are only required to quiesce the affected databases briefly, instead of shutting down the entire online system during the backup process.

Virtual concurrent copy support is provided through IBM® RAMAC Virtual Array devices with the SnapShot feature. With virtual concurrent copy, you create a "fast" point-in-time version of a data set so that it is unavailable for normal application processing for a very minimal period of time. The version is then available for application testing, reporting, or backup operations.
End-user and test databases can benefit from the automatic availability management provided by DFSMShsm.
Database management systems, except for CICS VSAM file control systems, typically maintain their own catalogs that contain allocation-related information about the data sets that support the online environment. Catalog accuracy is most critical when a database must be recovered. Because there is no automated facility to keep the database management system informed of backups done by DFSMShsm, you should not use DFSMShsm to back up production databases. Because end-user databases usually have less stringent recovery requirements, DFSMShsm can effectively manage them. DFSMShsm's automatic space management can migrate database data after a period of disuse and automatically recall it when needed by the user.
Management of database data is improved by implementing standards enforcement through the use of data classes and the data class ACS routine.
Storage and database administrators can jointly develop data classes for commonly used types of database data. With data classes, you can standardize the allocation parameters used to create new database data sets. You can also enforce data set naming standards through the use of data class ACS routines.
Use dual copy and RAID architecture to increase availability of critical data sets.
Some recovery data sets are duplexed by IMS and DB2 to provide greater reliability. You can use a cache-capable 3990 storage control's dual copy capability to provide protection against hardware outages for CICS, and extend availability to other IMS and DB2 data sets.

You can also use virtual concurrent copy support, through an IBM RAMAC Virtual Array device with the SnapShot feature, to create test data bases from production data bases instantaneously, without using additional storage resources. Virtual concurrent copy support lets you restart/rerun interrupted backups, which cannot be done using concurrent copy.
Database applications using non-shared resources (NSR) can benefit from improved performance and elapsed times in applications that access the data directly and sequentially.
Database applications that perform sequential data access can benefit from striping.
Applications, such as data warehousing that uses large sizes, can benefit from extended addressability.

Table 1 summarizes the SMS services for database data.

Table 1. SMS Services for Database Data
Data Type	Space Management	Performance Management¹	Availability Management	Volume Selection
Production Databases	None	Cache, DFW	Concurrent copy for point-of-consistency backup	Yes
End-user Databases	Automatic migration	Cache, DFW	Automatic backup using concurrent copy	Yes
Recovery Data Sets	Automatic Migration	Cache, DFW	Dual copy	Yes
System Data Sets	None	Cache, DFW	Automatic backup using concurrent copy	Yes
1. DFW refers to the DASD fast write extended function of a cache-capable 3990 storage control.

You can design your ACS routines so that SMS restricts the allocation of data sets in CICS, IMS, and DB2 storage groups to production databases and selected system data sets. Only specially-identified users, such as the database or storage administrator can allocate data in these storage groups. Most data sets that support the database environment, including the recovery and system data sets, are directed to the PRIMExx storage group. The storage and database administrators have special SMS authority to assign data sets with critical performance and availability requirements to specific volumes. Dual copy and RAID technology provide high availability for selected data sets that are not duplexed by the database management system. Use DASD fast write and cache to provide superior performance for databases and recovery data sets.

DFSMS supplements the backup and recovery utilities provided by the database management system as follows:

DFSMSdss uses concurrent copy capability and virtual concurrent copy support to create point-in-time backups.
Data base utilities (except for CICS) invoke DFSMSdss for concurrent copy and virtual concurrent copy support for point-in-time backups and backup-while-open.
DFSMShsm backs up system data sets and end-user database data that is less critical than production database data. You can use the backup-while-open function with concurrent copy or virtual concurrent copy support to back up CICS VSAM data sets while they are open for update.
DFSMShsm carries out direct migration to migration level 2 for archived recovery data sets on DASD.
End-user and test database data is migrated by DFSMShsm through the storage hierarchy, based on database data usage.