Page 2 of 2   « previous page

Batch Optimization
Even with a gifted team of performance analysts, optimally managing the complex batch environment and a myriad of CICS regions is a daunting task. In the 24/7 world, the static settings that govern batch jobs and the static tables that manage CICS usually fail to deliver optimal performance. Yet it is impossible for anyone to continually tune the settings around the clock. There exist solutions that optimize batch and CICS processing to ensure good performance at low resource cost.

Batch processing continues to grow, and the batch windows continue to shrink. Continuous applications processing and changing business models place more demand on existing jobs. Optimizing batch performance is no longer an option - it is now a critical necessity. Batch optimizing automates the complex task of batch performance tuning by increasing parallelism, optimizing data access, and keeping as much data in memory as possible during processing.

Three areas offer opportunity for batch optimization.

Data Optimization
A common myth that is associated with file buffering is that more is better. Data optimizing helps evaluate the file type and determines the access method of each file to dynamically optimize buffer resource allocations. Optimizing I/O access can translate to huge reductions in the elapsed time for batch processing because data stores on DASD is still significantly slower than data in memory.

Job Step and Job Parallelization
Combining data and job optimization strengthens the impact of an optimized batch workload. The batch cycle comprises job steps and jobs that are both dependent and independent of a previous step's processing. When independent jobs and steps are run in parallel, the resources complete sooner, making resources available for other sources. As more jobs complete sooner, less batch processing time is required by each workload. This process shortens the batch cycle for jobs within the critical path.

Job Piping
Batch optimizing improves parallelism by allowing jobs to pipe data into each other, causing a job stream to complete faster. The technology also extends to the piping capability across LPARs within a sysplex. By piping data across two or more LPARs, jobs can run in parallel on alternate partitions or processors to further increase parallelism, offering significant reductions in batch processing times and helping widen the ever shrinking batch window.

CICS Optimization
CICS has always been a core component of any large business, and now this subsystem has expanded its role by becoming a cornerstone of business strategy. Web-related traffic transactions often differ from legacy application traffic. The dawn of the Internet age, increased distributed processing, and increased enterprise database access have increased the importance of CICS. The number of regions continues to grow, and the use of a single application owning region (AOR) is no longer limited to a single application. CICS environments are becoming increasingly complex, forcing IT personnel to find new and innovative ways of proactively managing performance. The service level agreements that companies have in place must be met, or client relations and business suffer. Simply put, IT managers need a way to correct problems with minimal impact to users and other crucial workloads.

Automation is crucial in the complex CICS environment. IT staff should look to implement solutions that dynamically optimize CICS performance in response to workload peaks and valleys, as well as dynamically optimize resources, often before a performance monitor would have found it.

In addition, since the primary CPU cost of CICS relates to the large number of MVS waits that CICS issues as it processes the transaction workload, managing these calls to ensure that critical work completes faster with an optimized environment is essential to business flow. The workloads use fewer resources, so more are available to handle other workloads.

The goal of every IT organization is to reduce the cost of running mainframe applications while meeting the demands of the customers and its business objectives. The traditional approach of moving workloads or simply adding more hardware is no longer the best option. The goal of a business is to keep the applications running more responsively while consuming fewer resources, allowing them to defer expensive upgrades while providing customers with the service that they require.

Page 2 of 2   « previous page