VR is an intelligent, intuitive, performance optimization tool designed to dramatically improve system performance while significantly reducing system overhead by automating tuning activities.
VR reduces overhead by optimally allocating buffers for VSAM and non-VSAM files to provide the best throughput and response time for I/O requests.
For VSAM optimizations, VR intelligently selects either the NSR (Non-Shared Resource) or LSR (Local Shared Resources) buffering techniques, and then uses its internal optimization rules and services to make the most efficient and effective use of that buffering technique.
VR supports all current VSAM macro formats, and can intelligently use those macro formats in conjunction with the selected buffering technique to accelerate application processing. When needed, VR can exploit the use of all 256 LSR buffer pools. VR also provides VSCR (Virtual Storage Constraint Relief) by placing VSAM control blocks and buffers above the 16MB line.
For non-VSAM optimizations, VR ensures optimal tuning and buffer allocation by selecting a buffer count that will optimize the use of SAM/E chained scheduling for asynchronous sequential access (QSAM).
By buffering VSAM and non-VSAM files more efficiently, your system can process applications more quickly and with less demand on the CPU. VR allows users to defer CPU upgrades, thereby saving both time and money.
The primary challenges faced by data centers are the need to reduce batch processing time, the need and ability to meet service level agreements, the requirement to maximize system performance without extensive programming changes and the ability to determine system-wide impact of application tuning activities. VR addresses these challenges by:
- Eliminating CPU time that is wasted during I/O operations.
- Automatically reducing system overhead and batch window constraints.
- Eliminating manual tuning efforts.
- Allowing users to defer or eliminate CPU upgrades.
- Ensuring service level agreements are more easily met.
VR allows you to:
- Reduce batch window processing constraints by performing VSAM and non-VSAM batch processing in a fraction of the time. Application processing can sometimes be reduced to 1/10th of the time when VR is implemented.
- Eliminate manual tuning efforts. Technical resources that might otherwise be utilized for manual tuning can be reassigned elsewhere without any loss in performance.
- Eliminate or defer the need for CPU upgrades. Fluid application processing reduces CPU requirements while making better use of main memory, thereby extending the life of the CPU and offsetting the cost of expensive upgrades.
- Meet service level agreements. On-line systems will come up on time, and with improved application response times as batch processing requirements are completed in less time.
- Obtain performance improvements without JCL or application changes. Manual performance tuning requires significant JCL or application code changes, while VR optimizes performance without alteration to either.
- Eliminate the monitoring of those JCL changes. Once manual changes are made to JCL to implement tuning for a specific data set, those changes must be monitored (as the personality of the data set and access to the data set may change over time). VR eliminates the need for this by dynamically performing it as part of its normal processing.
- Make dynamic changes in file processing. VR dynamically adapts to changes in file processing and helps with time-of-day memory management issues or other situations that could impact buffering requirements.
- Expand virtual storage. LSR and NSR buffers are built above the 16MB line to alleviate any virtual storage constraints. For LSR, expanded storage using hyperspace buffering is automatically and transparently used when the file size is larger than the LSR pool. This provides up to 2GB of virtual storage for buffers and gives ample room below the line to run larger applications.
VR is easily and transparently installed on any z/OS system. Once enabled, batch I/O requests are analyzed with the resulting buffering environment being modified to utilize the best possible accesses to each file. An immediate and dramatic difference in run times will be seen as buffers are allocated more efficiently, EXCPs are reduced, more storage is available for application programs, and file processing is optimized.
VR contains a proprietary database containing the rules and parameters that deliver the best I/O performance based on the current characteristics of the target data set and the type of access. Users can override these rules by keying in their own parameters, or by excluding a dataset or job name from VR's optimization process. Most users however, use the default parameters and achieve outstanding results from VR's I/O optimization. VR monitors the open and close of any data set and optimally buffers VSAM, QSAM, and BSAM files to significantly reduce the time required for batch processing.
VR customers routinely report that they can reduce VSAM batch processing times from 60 percent to 90 percent. QSAM and BSAM batch processing times can be reduced by up to 85 percent. VR saves up to 50% elapsed time for DB2 Unloads and Reloads and saves up to 40% elapsed time for IDMS sweeps.
VR runs on all z/OS platforms and requires 68kb of common storage (CSA) below the 16MB line. The user interface requires ISPF Version 3.3 or above.
Detailed statistics can be produced showing the type of buffering that was applied and the amount of data movement that occurred. Also available is the number and type of program I/O requests for VSAM files. These statistics may also be recorded in SMF records for later analysis.