*************************************************** * DataPropagator Relational Questions and Answers * October 4, 1994 *************************************************** Q: CAN YOU GIVE ME A BRIEF OVERVIEW OF TODAY'S DATAPROPAGATOR RELATIONAL ANNOUNCEMENT? A: Today, IBM is announcing enhancements to DataPropagator Relational providing: o New sources - Capture for AIX - Capture for OS/2 o New targets - Apply for HP-UX - Apply for Solaris - Sybase and Oracle through IBM DataJoiner o Usability and performance improvements DataPropagator Relational provides robust replication capability for defining, synchronizing, automating, and managing copy operations between relational databases across your enterprise. Release 2 supports copying between DB2 family databases and other DRDA application servers in addition to replicating to Sybase and Oracle through IBM DataJoiner. Update propagation (copying only the changes to the copy) is announced from DB2 databases on MVS, OS/400, AIX, and OS/2. Full refresh (full replacement of the copy) is enabled from any DB2 family database. DataPropagator Relational can interoperate with any DRDA application server. Application servers can be sources for full refresh or replication targets for both refresh and update propagation. The DataPropagator Relational target databases are extended through DataJoiner, IBM's new multi-database server. Providing powerful heterogeneous distributed join and update capabilities, DataJoiner provides a common SQL interface to mixed environments. Together with DataJoiner, DataPropagator Relational can replicate data to Sybase and Oracle copies. DataPropagator Relational provides rich data enhancement capabilities to deliver detailed, subsetted, summarized, restructured, or derived data to the target table. The DataPropagator Relational design gives the user the flexibility to manage within system constraints, particularly regarding network resources, data transfer volumes, and OLTP response time requirements. DataPropagator Relational provides a GUI for ease of administration. Q: HOW WILL DATAPROPAGATOR RELATIONAL HELP MY BUSINESS? A: DataPropagator Relational supports a range of applications from replicating host data to LAN servers for application downsizing to building executive information systems or other decision support environments. Key to supporting this range of applications are: o Automatic data copy supporting synchronization requirements from minutes (for operational applications) to weeks (for decision support applications) o A design compatible with high transaction rate production systems o Tailoring data during the copy process to meet the user's or application's needs. Enhancements are specified through SQL leveraging skills already in your enterprise. o Change capture to maintain data histories for trend analysis,audit, or recovery from application logic errors. o Efficient use of network resources o An easy-to-use administration capability o Open interfaces for sophisticated customizing and interoperability with non-IBM data stores o Scalable solution capable of supporting 100s to 1000s of copies Q: WHAT KIND OF APPLICATIONS CAN DATAPROPAGATOR RELATIONAL ENABLE? A: For example, here are some of the kinds of applications that customers are using DataPropagator Relational for today: o Building decision support systems on mainframes, LAN servers, or stand-alone workstations o Offloading query processing from transaction processing systems to allow for transaction rate growth o Reducing batch windows by replacing extract jobs with background replication o Downsizing applications which must still share data with legacy systems o Supporting stringent audit requirements requiring the ability to query data as of any point in time. o Analyzing business trends using change data Q: DOES DATAPROPAGATOR RELATIONAL SUPPORT NON-IBM DATABASES? A: Yes, multi-vendor databases can be supported in three ways: o DataPropagator Relational interoperates with any DRDA application server, so for example, XDB could be a replication target or a source in refresh mode. o DataPropagator Relational and DataJoiner work together to support replication to Sybase and Oracle databases. These can also be sources in refresh mode. o DataPropagator Relational's data staging area provides an architected, published interface allowing other programs to populate or retrieve from the staging area. Q: WHY WOULD I NEED DATAPROPAGATOR RELATIONAL IF I HAVE DATAJOINER? A: DataPropagator Relational and DataJoiner are complementary offerings. They can be used together to extend replication function to multi-vendor environments. Fundamentally, DataPropagator Relational deals with data placement and DataJoiner with data access. Customers must examine their data access requirements to understand whether data should be cached locally to the application (replicated) or accessed remotely from its home location. Replicating data for local client use is desirable for those applications that have high performance requirements for data on a predictable and recurring basis. Since data is put to similar uses over time, users can anticipate their data requirements, and use IBM's Data Replication Solution products to place replicas where they are needed and where locality of reference can be exploited to enhance performance. Such applications must be tolerant of latency between the source and copies. Replication may also be desirable where ad hoc query access interferes with production system performance. Replication can be scheduled and managed for minimum impact. DataJoiner addresses the needs of users whose information needs are not easily anticipated. Long lead times to develop responses to their queries are prohibitively costly and put them at a distinct disadvantage with respect to their competitors. DataJoiner also addresses the needs of applications which require access to live data and any latency is not tolerable. Although the IBM Data Replication Solution products and DataJoiner satisfy different application requirements, many installations have applications whose profiles warrant using DataJoiner along with IBM's replication facilities. These tools can be profitably used in a complementary way. Q: WHAT ARE SOME OF DATAPROPAGATOR RELATIONAL'S STRENGTHS? A: We believe DataPropagator Relational's strengths include: o Ability to address a range of applications from operational to informational. o Use of standard SQL language to enhance data prior to propagation. o The ability to deliver different views of the same source data to different copy targets. o Keep high data volatility isolated between the source and data staging tables. This allows you to manage your network resources efficiently. o Create and maintain data histories providing for trend analysis capabilities. o Easy administration through graphical user interfaces. Q: I WANT TO COPY DATA FROM LEGACY DATA SOURCES. CAN DATAPROPAGATOR RELATIONAL DO THAT? A: DataPropagator Relational does not directly support non-relational data sources. However, DataRefresher and DataPropagator NonRelational interoperate with DataPropagator Relational to provide an end-to-end copy solution from legacy sources to DataPropagator Relational copy targets. Q: DO I NEED TO HAVE DATAHUB INSTALLED TO USE DATAPROPAGATOR RELATIONAL? A: DataPropagator Relational administration is built upon the DataHub control point and requires the DataHub/2 platform. The DataHub Support components are not required for DataPropagator Relational administration or operation. However, customers may find the complete DataHub environment valuable for database systems management tasks such as authorization management and utility execution. Q: WILL DATAPROPAGATOR RELATIONAL RUN FROM THE NEW DATAHUB FOR UNIX? A: DataPropagator Relational currently runs on the DataHub for OS/2 control point. Product enhancements should include replication administration from DataHub for Unix. Q: WILL DATAPROPAGATOR RELATIONAL SUPPORT TABLE JOIN ON COPIES? A: Yes, this is one of the advantages of a staging table design based on relational tables. Q: WILL DATAPROPAGATOR RELATIONAL SUPPORT DATA TRANSFORMATIONS? A: Yes, DataPropagator Relational supports data transformation through views defined on a base table and a translation table. Q: CAN DATAPROPAGATOR RELATIONAL COPIES BE SCHEDULED? A: DataPropagator Relational supports automated copy operation at user specified intervals. Time and date scheduling is not supported in Release 1, however we are investigating how best to provide this support. Time and date scheduling may be approximated through user specified predicates in the copy specification. Copies may also be initiated through application events. Q: CAN LOGGING BE TURNED OFF FOR CHANGES TO THE STAGING TABLES? A: Logging options are defined by the database management system. Where the database manager supports non-recoverable tables, logging may be turned off for the staging tables. Q: WILL DATAPROPAGATOR RELATIONAL ALLOW ME TO AUDIT UPDATES TO DB2 TABLES? A: DataPropagator Relational does not provide audit support within the transaction commit scope, however DataPropagator Relational captured change data may be analyzed by audit applications providing effective asynchronous audit capability. Both before and after image information can be captured with DataPropagator Relational. Q: WILL DATAPROPAGATOR RELATIONAL ALLOW ME TO UPDATE A COPY TABLE AND HAVE THAT UPDATE AUTOMATICALLY REFLECTED BACK TO THE BASE TABLE ON DB2? A: DataPropagator Relational currently supports a single updater model. Where two databases update distinct fragments of the data, changes may be propagated back and forth. This implies that the application program must ensure that updates do not occur to the same data at different locations. When the same data may be updated at different locations (sometimes known as 'update anywhere' or 'updatable copies'), then conflict detection and resolution logic is required. DataPropagator Relational does not support updatable copies in this release, however it will introduce this support in 1995. Updatable copy offerings are new and largely untested in production environments. Customers must consider their implications very carefully. Q: DOES DATAPROPAGATOR RELATIONAL SUPPORT DB2 V3.1'S DATA COMPRESSION FEATURE? A: No. However, DataPropagator Relational supports DB2 V3.1 and can support change data capture for non-compressed data. This is a recognized and accepted requirement. Q: ARE ALL COMPONENTS (CAPTURE, APPLY, AND ADMINISTRATION) OF THE SYSTEM REQUIRED TO BE UP AND RUNNING IN ORDER FOR MY COPIES TO BE MAINTAINED PROPERLY? A: All three components are designed to run autonomously. That is, any component can be started or stopped at any time, yet it will still run properly. This design produces a very fault-tolerant operational environment. Q: CAN THE CAPTURE AND APPLY COMPONENTS BE STARTED FROM THE ADMINISTRATION CONTROL POINT? A: The capture JCL could be submitted from the DataHub control point, but it is recommended that the capture component be started as part of system startup, ideally automated with a tool such as NetView. The apply component cannot be started from within DataHub, although any tool supporting remote command line (such as IBM's Distributed Console Access Facility) could provide such capability. Again it is recommended that the apply component be started as part of system startup. Q: CAN TRANSACTIONS TRIGGER COPY OPERATIONS? A: Yes. Copy operations may be triggered by an application update of the tables controlling the apply function. Q: DOES DATAPROPAGATOR RELATIONAL USE PRIVATE COMMUNICATION BETWEEN ITS COMPONENTS? A: No. DataPropagator Relational components communicate indirectly through the staging tables and control tables. DRDA connections provide table access between remote systems. Q: WHAT ARE THE SECURITY FEATURES THAT DATAPROPAGATOR RELATIONAL PROVIDES? A: DataPropagator Relational relies on the security features of the operating system and database manager. Registrar and subscriber authorities are no more than convenient names for groups of authorities granted to users. Q: CAN THE CAPTURE COMPONENT READ BOTH THE ACTIVE LOG AND THE ARCHIVE LOG ? A: No. The capture component reads only the active log. Q: DOES THE APPLY PROGRAM USE DYNAMIC OR STATIC SQL? A: Dynamic SQL. Q: IF I ALTER A TABLE ATTRIBUTE, MUST I STOP THE COPY OPERATIONS? A: No. Copy processing can continue without interruption. However, if changes in the table attributes should be reflected in the copy processes, then copy subscriptions must be cancelled and re-done, and the Capture component must also be reinitialized. Q: CAN DATAPROPAGATOR RELATIONAL BROADCAST CHANGES FROM A SINGLE TABLE TO MULTIPLE TARGETS AT DIFFERENT LOCATIONS? A: The DataPropagator Relational design is based on the Apply component 'pulling' information from the source or staging tables. As such, DataPropagator Relational does not 'broadcast' changes to target tables and locations. However, DataPropagator Relational can be configured to support distribution of copies to any number of target locations and databases from a single source providing a highly scalable solution. Q: WHAT IMPACT WILL DATAPROPAGATOR RELATIONAL HAVE ON THE PERFORMANCE OF MY DB2 SUBSYSTEM RUNNING THE CHANGE CAPTURE FUNCTION? A: Our experience indicates that added CPU overhead for change capture is very low (a few percent) depending on the volume of data you are capturing. Q: ARE THERE CONSULTING AND SERVICE OFFERINGS FROM IBM FOR DATAPROPAGATOR RELATIONAL? Consulting and services can be used with customers to plan for a DataPropagator Relational solution. Consulting and services offerings are available now to support all aspects of the Information Warehouse family -- from strategy, planning, design, implementation, operations, to evaluation. Contact your IBM representative. IBM (c) International Business Machines Corporation 1994 All Rights Reserved References in this publication to IBM products or services do not imply that IBM intends to make them available outside the United States. DataPropagator, IBM, AIX, OS/2, DataJoiner, DB2, DRDA, OS/400, DataRefresher, DataHub, NetView, and Information Warehouse are trademarks or registered trademarks of the International Business Machines Corporation. All other products are trademarks or registered trademarks of their respective companies.