Operations Must Overcome the Irrational 'Fear' of Applications
It’s now generally accepted that you have to monitor and tune your computer applications to ensure good performance and control costs. But IT operations teams still need to accept the fact that they have an important role to play in this.
The idea that operations professionals should have nothing to do “application-wise” with the systems they run on a day-to-day basis to service their company’s business is old fashioned and an extremely limiting factor in times when we all need to be more agile.
Application tuning and monitoring, which falls under the remit of application performance management (APM) within the IT infrastructure library (ITIL), involves profiling the way applications use CPU (and other) resources as they run—then identifying and correcting inefficiencies to optimize the use of processing power.
With most mainframe sites now subject to variable workload-license charging, which rewards those who use less processing power with lower software bills, APM can help reduce operational running costs as well as maintain better performance.
But the simplistic picture that IT operations should highlight a performance problem (“There’s something wrong with this!”) and that the developers—those people who originally wrote the application and are still around to maintain it—will then pick up the gauntlet and fix the problem, just doesn’t work.
Effective APM needs cooperation between operations and development to achieve success, and in no small part the operations people need to take the first steps in making this happen.
The days when application developers wrote systems and were still around to maintain them years (sometimes many, many years) later are long gone. Developers are allocated to other projects, move to other companies or retire (a particular problem in the case of mainframe applications). In other cases, application support is outsourced to an external company that might lack the specialist knowledge to fix problems easily when they arise.
That leaves operations teams—which are at the sharp end and therefore experience, first hand, the problems caused by slow and inefficient applications—to take the initiative. They need to accept their roles in looking at application performance, isolating problems and actively tuning applications to increase efficiency as well as reducing the amount of resources that are consumed.
Step one is to change the organizational culture to recognize the proactive role that operations teams need to play if they are going to implement a successful APM program. Step two is to make sure they’re provided with the right tools for the job.
Automated APM software enables operations teams to profile applications as they run, identify and highlight the badly performing database calls, the high CPU-using program code, and other application performance problems—all without the need for an applications background or detailed knowledge of individual applications.
Step three is for operations teams to manage the problem through to a successful conclusion. In many cases, problem resolution will be obvious—as the performance management application can pinpoint the source of the error. In other cases, it could be something for the database analysts to have a closer look at (if it is a database-related issue, as many are).
Otherwise the task of providing a fix can go back to the development staff, along with the specific information now on hand that pinpoints the exact location and nature of the problem, thus removing the need for as much specialist knowledge.
The upshot is that through cooperative working such as this, IT teams will have a better chance of achieving the goal of improving application performance and saving their company money by consuming fewer resources. They will also be making better use of the operations teams’ skills, which are so well placed to address APM issues.
Philip Mann is principal consultant at mainframe performance management expert, Macro 4. He has been working with IBM mainframes for more than 30 years, including more than 10 with Macro 4, where application-performance tuning is one of his major interests.