Mainframe Optimization with MIPS Reduction

MIPS Reduction for Production Control Analysts

1/5/20242 min read

Did you know:

  • 60-65% of mainframe MIPS are used for batch processing

  • Mainframe optimization can reduce MIPS 5-10%

  • Mainframe modernization can reduce cost 60-80%

Batch processing is a key attribute of mainframe operations, and the daily window for batch processing is always being improved upon in order to maximize cost and time efficiency. Because there are many new analytic and operations processes that continue to add demand, multiple daily batch windows are an increasingly common practice.

But the reality is that batch processing on the mainframe will always be expensive and time consuming. And batch processing during normal operation—busy hours—will impact the performance of mission-critical on-line computing applications, such as transaction processing. The failure or push-out of a batch window can cascade through the daily job schedule. So the opportunity to offload batch workloads from the mainframe helps business agility, saves costly MIPS and avoids the risk associated with missed SLAs.

vStorm Enterprise is a scalable, high-speed solution to offload mainframe data to Hadoop for batch processing and thus dramatically reduced run time and expense. Batch processing in Hadoop unlocks affordable scale-out performance, lower cost storage options, and better analytics capabilities leveraging modern technologies including Java, HBase (a no-SQL database), and Hive (SQL queries), and ElasticSearch.

On average, a mid-sized enterprise with an average of 500 MIPS can save $500K per year through a MIPS reduction strategy by offloading automated transaction processing and bulk database update oriented batch workloads to Hadoop. Additional strategic advantages include:

  1. The ability to evaluate new offerings with multiple daily batch windows

  2. Data storage savings from EDW offload (Teradata, for instance)

  3. Savings in Audit from raw granular data retention

  4. Mainframe maintenance cost reduces to fraction of one tenth for using PIGS on Hadoop instead of COBOL for scripting.

A Case Study

Very often, for growing volumes of data the current SLA just doesn’t work. In one case, pricing data kept on DB2 grew to over 500M records and could not be processed within the mainframe’s 3.5 hour batch window. Offloading the job to Hadoop enabled the job to complete in 1.5 hours and saved over $100K.

The vStorm Enterprise software is a design-time platform (point and click based GUI) that has batch capabilities to stream z/OS (relational and non-relational) data into Hadoop and other business analytic platforms. It does the complex work of data conversion for DB2, VSAM, QSAM, SMF/RMF and other formats without consuming MIPS or requiring any staging. It supports offload to Hadoop on IFLs, Power Systems, or x86 platforms easy and efficient.

Users will see MIPS savings because this product is different from traditional ETL tools that typically consume MIPS while running large SQL queries, and custom COBOL programs that run on-platform.

Irrespective of the infrastructure environment, the connector will maintain existing RACF security and move the data over a secure channel as well, maintaining the important governance processes that typically cover transactional mainframe data.