How to Make Your Mainframe Work Harder in 2020

With the advent of distributed processing and the cloud, many predicted that the mainframe would eventually fade away. In fact, just the opposite is happening. According to a recent survey, the widespread use of mobile devices has alone led to an average 41% increase in mainframe MIPS (million instructions per second) consumption. Today’s mainframes are working harder than ever.

But IT budgets aren’t keeping pace. Even as mainframes are being tasked with doing more, the financial resources needed to support that expanded role are often increasing slowly if at all. In other words, mainframes today are expected to do more with less.

If that’s the case in your organization, here are some mainframe optimization steps you can take to make your it more productive without significantly increasing costs.

Take advantage of zIIP

Companies typically devote upwards of 30% of their mainframe budgets to MLC (IBM’s Monthly License Charge). Because MLC is directly tied to CPU usage, reducing MIPS consumption is vital for getting more out of your mainframe without increasing costs. And that’s exactly what IBM’s System z Integrated Information Processor (zIIP) is designed to do.

With zIIP, you can offload some workloads from your CPU to hardware. For example, with Syncsort ZPSaver and Syncsort MFX working together, you can transparently transfer as much as 90% of the CPU cycles used for Copy, Sort, and SMS Compression operations to the zIIP engine, cutting MLC costs and reducing elapsed time for those operations by up to 25%.

Optimize databases

Db2 is a highly efficient relational database system. But some common mistakes in query coding can slow it down drastically. For example, a common error is using SELECT * when all columns are not actually needed. It’s much better to specifically identify the required columns in the SELECT statement. In one test, misuse of SELECT * resulted in a CPU time penalty of 70%, while elapsed time increased by 17%. Syncsort Optimize DB2 can help you to identify and correct problematic SQL coding quickly.

Also, if your shop is running both Db2 and IMS, you might consider migrating from IMS to Db2. IMS remains an excellent database system, but by standardizing on Db2, you can avoid duplicating staffing and licensing costs. Syncsort Optimize IMS can make the process relatively painless.

Reduce application code inefficiency

As with database code, updating poorly written COBOL or PL/1 application code can yield big performance dividends. For example, one application employed a file write subroutine that opened and closed the file every time the subroutine was called. Moving the file open and close statements outside the subroutine substantially increased the speed of the entire application.

Optimize the batch environment

Batch processing still accounts for more than 50% of data processing workloads, so optimizing these operations can yield substantial efficiency improvements. One industry study reveals that 5% of batch processing isn’t even necessary, and that properly tuning the batch environment can reduce processing times by up to 30%.

Optimize I/O

Batch jobs spend most of their time (60-70% on average) doing I/O. So, improving I/O efficiency is key.

A good first step is making use of PAV (Parallel Access Volumes). This z/OS optional feature allows multiple tasks to access a disk concurrently rather than sequentially, thus significantly improving I/O throughput.

Upgrade compilers

Because of ongoing advances in compiler technology, a modern compiler can significantly increase the performance of unmodified legacy COBOL or PL/1 code. According to IBM, aggressively recompiling legacy applications using the latest compiler technology is the single most powerful tool available for improving application performance.

Mainframe optimization is key to efficiency

By being intentional about mainframe optimization, you can make your hard-working mainframe even more productive without busting your IT budget.

Learn how to offload, accelerate and lower cost while leaving the primary CPU with more headroom for the organization’s core business applications — read our white paper: Getting the Most Out of Your Mainframe