DB2 Parallel Edition for AIX, System/390 Parallel Query Server, and System/ 390 Parallel Transaction Server support DB2 technology. These products allow users to leverage their investments in DB2 knowledge and skills, and enable desktop systems to access data in a client/server environment, where the server is a parallel database. All three offerings result from years of parallel database re-search, building on the strength of DB2 and IMS technology, such as reliability and data integrity.
Here are some technical highlights of the advanced new products.
Because DB2 Parallel Edition shares the same architecture as the SP2, the combined solution can potentially offer near-linear scalability -- that is, as new processors are added, very little overhead is added. The "shared-nothing" architecture does not share any resources -- such as memory or disk -- so there is little system overhead because of many processors attempting to use a shared resource.
DB2 Parallel Edition matches the SP2 architecture in the following ways:
Because data are partitioned across the processors, loading data and reclustering data, database backup and restore, and forward recovery can all run in parallel. Indexes are created in parallel on each of the processors.
Today, SPQS uses a traditional S/390 DB2 implementation as a front-end system to connect with end users, so end-user query tools that use DB2 directly can access the SPQS. This capability means customers can leverage existing investments in DB2 skills, utilities, and procedures to manage the server. A stand-alone SPQS offering is planned in 1995. The underlying platform is the new generation of S/390 multiprocessor complexes.
SPQS can split the work of a single query across 48 processors. Because the data are shared by all the processors, the individual split queries can process the data in parallel. The workload for each processor can be managed so no processor is overloaded. This shared architecture allows users to manage workload balance and allows for non-disruptive growth; processors and disk storage devices can be added independently and without a database reorganization. This architecture offers superior availability, since data are not specifically tied to any processor.
The System/390 Parallel Transaction Server builds on the strengths of MVS, CICS, IMS, and DB2 to extend the range of current transaction processing in a Parallel Sysplex environment. It enables users to read and write shared data across the server and other sharing machines. This ability allows for easy integration into an existing S/390 environment, and reduces the total cost of computing by leveraging the S/390 microprocessor technology.
The shared architecture provides improve-ments in availability, security, integrity, and management. Sharing data means that work can be scheduled to run on an under-utilized processor. When a planned or unplanned outage occurs, there is an alternate path to the data, which users can access easily. Security and integrity are managed automatically, simplifying system supervision.
DB2 Version 4 introduces query parallelism into the data sharing environment to provide improved response time of read-only, long-running queries. Examples include data-intensive queries that scan large volumes of data and require considerable I/O processing. Complex queries that contain many functional complexities, such as complicated predicates, column and scalar functions, multiway joins, and grouping or ordering clauses -- and require a high amount of CPU processing -- will also benefit.
Query parallelism enables true multitasking within a query. A large query can be broken into multiple smaller queries. These smaller queries can run simultaneously on multiple processors within a tightly-coupled multi-processor system while processing data in parallel. Also, parallel I/O techniques may be used for accessing data partitions.
See also:
DB2: What Happens Next?