By now, we all know the importance of superior application performance. Applications that are fast, reliable and easy to use delight end-users and lead to greater adoption. But for mainframe applications, performance takes on a whole new level of importance. For these applications, even a few added milliseconds in application load or transaction time can lead to application abandonment and lost revenues.
Why? Surely, mainframes process so many transactions for so many people that even a slight improvement in processing time can have an impact on millions of end users. For example, a leading bank recently saw that a mainframe application was taking too long to make a database call, increasing from three milliseconds on average to five milliseconds. While this may seem like a trivial time increase, it caused more than three million transactions during a critical period to slow way down, or even time out. After identifying and fixing the problem, the bank was able to bring response time levels back to normal. Given the sheer number of transactions affected, the impact on customer satisfaction and the overall business was enormous.
But beyond the end-user experience, it is equally important to manage mainframe application performance from a resource efficiency perspective, since problems here can also result in huge costs to the business downstream. This is especially true as trends like cloud, mobile and analytics, and the availability of the new IBM z13, push increased workloads to the mainframe.
Mainframe transaction processing is very cost-effective – even more than commodity servers in many instances. This is because as many businesses experience massive increases in computational loads, the mainframe has decreased in unit cost enough to offset changes in volume – more so than commodity servers.
The mainframe is inherently more scalable than most commodity servers – a reason it has long been a platform of choice for critical transaction processing, along with superior reliability and security. But one potential danger of mainframes is how monthly license charges (MLC) for mainframe software often consume more than 30 percent of mainframe budgets. This can lead to costs spiraling out of control, especially as mobile apps become even more ubiquitous. Consider that a single mobile transaction often triggers a cascade of related events across systems, including such things as comparison to past purchases, business to business reconciliation (like those between banks), customer loyalty and rewards program updates, and many other examples – a phenomenon known as the “starburst effect.” Businesses using mainframes need to keep an eye on MLC costs – otherwise, the promise of low TCO may be endangered.
The global thirst for more computing capacity continues to grow. The mainframe has a very real place in this new paradigm. But the key to maximizing the significant cost-savings and overall business performance advantages the mainframe offers lies not just in managing application performance from an end-user experience perspective – but also from the critical perspective of resource utilization and consumption.
Spencer Hallman is Subject Matter Expert for Compuware.