How To Improve Mainframe Performance

A Guide to Optimizing Your Mainframe’s Overall Performance

Most likely, your company still uses mainframes for mission-critical IT tasks, just like the vast majority of the Fortune 500. As a result of the shift to higher performance requirements across industries in the wake of the 2020 pandemic, businesses are looking for ways to optimize the efficiency and cost-effectiveness of their mainframes.

Users of mainframes have to work hard to keep up with the growing amount of data from distributed applications that depend on the transaction processing power of mainframes. In order to save money, businesses are constantly under pressure to get more out of their current mainframes.

Here are eight recommendations for optimizing your mainframe’s overall performance:

  1. Utilize mainframe monitoring software
    Before you can improve how well your mainframe works, you need to know what’s going on inside:
    Which tasks put the most strain on the system’s resources?
    What specific operational bottlenecks and processing inefficiencies need to be addressed?
    How do the performance baselines compare before and after optimization efforts?

    To get the answers you need, you’ll need monitoring tools that can track your mainframe processing in real-time and give you a complete picture. Our z/OS solutions, zWorkload Reporter and zGuard, pull important operational and analytics data from the mainframe and put it in a cloud reporting portal and a real-time Grafana monitor dashboard while automating tasks to help reduce your workload costs.
  1. Reduce costs with sustainability
    Large savings and a healthy ROI are the results of a mainframe that has been fine-tuned. When your mainframe is operating at peak efficiency, it consumes less energy. Reduce your carbon footprint and save money on energy costs by prioritizing sustainability at each stage of the product life cycle.

  2. Protect against threats before they happen
    Cybercrime continues to be a major concern for business. One company will likely be hit by a cyberattack every 11 seconds, with a total cost of over $6 trillion USD in 2021. Data breaches still happen, even with the best security, whether by attack or accident. In the case of cyberattacks, the question is not if they will happen but when. For secure and reliable end-to-end mainframe-driven environments, you need to carefully consider the compute platform, storage arrays, and network as a whole.

  3. Find ways to streamline the code
    Slow performance is often the result of poorly written COBOL or Db2 SQL code. COBOL is notably inefficient when it comes to tasks like sorting and rounding numbers. Rewriting critical paths or heavily used legacy code could improve performance in a big way.
  1. Fix the code compilers
    IBM keeps backward compatibility so that application programs don’t have to be recompiled every time the operating system, middleware, or hardware of a mainframe is upgraded. However, you may need to recompile some COBOL or PL/1 programs. 

    Because compiler technology is always getting better, just recompiling source code that hasn’t changed can make a big difference in performance. IBM says that “aggressively” using the latest compiler technology can cut CPU time by up to 17%.
  1. Reduce the batch processing window
    Even though mainframes are widely used for online transaction processing (OLTP) in real time, batch processing is still an important part. Regular batch runs are an important part of the IT workflow for many businesses, whether they are used to make daily operational reports, produce customer account statements, or process payroll so that employees can be paid.

    In certain respects, batch processing is significantly more efficient than online transaction processing (OLTP). In contrast to OLTP, which may require multiple time-consuming database queries for each transaction, a batch process may only access data once, storing it in memory for the duration of the program’s execution.

    When both batch and OLTP applications run at the same time and try to use the same CPU or storage resources, OLTP must take priority because it is real-time. If you want to make the most of your resources and spend the least amount of time and energy on batch processing, you should change the JCL that controls your batch jobs.
  1. Offload processing to zIIP
    For certain kinds of workloads, specialized hardware like IBM’s System z Integrated Information Processor (zIIP) can greatly lessen the CPU processing burden (and cost) on your mainframe.

    By using zGuard, you can cut the time it takes to do certain tasks by up to 25%. This is because up to 90% of the CPU cycles needed to do tasks like copy, SMS compression, and sort can be sent to zIIP instead. And here’s a great bonus: the only cost for using zIIP is the initial hardware purchase; there are no software license charges.
  1. The benefits of optimizing mainframe performance are substantial
    The time and effort spent on optimizing your mainframe’s performance will be well worth it in the long run if it means you can keep up with rising data processing demands without spending more money on new hardware.

If your company’s primary or dev/test CPU could use some breathing room so it can focus on mission-critical tasks, you should check out our zOptimization Platform to enable full offloading, acceleration, and cost savings.

Transform your mainframe for the digital era →