You are currently on IBM Systems Media’s archival website. Click here to view our new website.

MAINFRAME > Case Studies > Miscellaneous

An Educational Experience

The University of Georgia cuts costs via sub-capacity pricing

The University of Georgia cuts costs via sub-capacity pricing
Photography by Stan Kaady

Pay for Use

When the z890 was introduced, the school quickly jumped on the opportunity to move to a 64-bit environment and a much faster machine. “When we made the switch, we experienced increased processor speeds,” Marable recalls. “Our application developers and business users couldn’t believe how much faster things were running. They were very pleased with the results.”

At the same time, the school was facing increased budgetary belt tightening, with both the school and the state having fewer resources than they’d had in the past. This, of course, impacted many of the university’s operations, including its IT resources. It had to find a way to reduce back-office costs. Thankfully, the z890, which Mainline Information Systems helped size, offered a roadmap to meet that challenge.

In addition to increasing response-time speeds and allowing the school to move into a 64-bit operating environment, and therefore allowing it to use the latest versions of mission-critical software such as DB2, the upgrade also allowed the university to take advantage of the sub-capacity pricing of some of the IBM software products, in a “pay for what you use” model.

This was particularly important to the University of Georgia because its IT processing business model wasn’t typical when compared to many commercial enterprises, which might have steady, daily processing requirements. For example, the school has several periods during which user activity dramatically spikes, including in January and August, “when,” as Marable explains, “we have registration. During those times, we could have 35,000-plus users hitting the system.” The end of the fiscal year was also another time when system usage would surge.

The university bought a system that would accommodate those annual peak periods, but rely on sub-capacity software pricing. As a result, says Marable, “We pay for what we use and not the full capacity of the machine.”

As she further explains, “There are certain services, such as IMS and different toolsets that are sub-capacity eligible, which means that you’re only charged for the highest four-hour average of your machine use for a month. So once a month, I send a report to IBM telling them what our CPU utilization was, and they charge us only for that use on the eligible software and tools.”

The money the university has saved by moving to this sub-capacity pricing model (which, as mentioned previously, is in the thousands of dollars a month) has allowed EITS to use some of those savings for other purposes. For example, it purchased an additional 8 GB of memory to put into the mainframe, which, when added to the buffer pool, increases performance while decreasing CPU usage. Another purchase was IBM’s DB2 Query Monitor for z/OS, which allowed database administrators and application staff to quickly and easily pinpoint specific SQL calls for improvement. “We’re not only lowering our CPU, but we’re making processes faster and better for our clients and customers, which are our students, faculty and staff,” Marable notes.

Tuning and Tweaking

Of course, that’s the easy description of how the University of Georgia used sub-capacity pricing. To make sure it took full advantage of this service, the school had to monitor its workloads and move them as needed, which helped drive down peak usage periods. And as Marable explains, “It’s all about averages. We had to make sure we had some low spots near some high spots and that they weren’t overlapping and hitting some higher daytime averages.” Mainline was critical to this endeavor, too, running analyses to ensure the school was making the most of the sub-capacity pricing model.

Some processing, such as DB2 queries, however, couldn’t be moved around. They’re dynamic and happen throughout the day. The school addressed this, however, by improving queries, using available performance and analysis tools for DB2, and by adding part of that all-important 8 GB of memory to the DB2 buffer pools—not to mention calling on its database administrators to find any efficiencies they could. “We were able to decrease CPU utilization of that workload by a good 20 to 30 percent just by adding the memory and reevaluating the top 5 percent of the dynamic application calls for performance,” Marable says. Database administrators found other ways to improve database efficiencies and worked with the application-development teams on enhancements as well.

The organization even put alerts in place, using RMF, for example, to notify operators when CPU usage was approaching a predetermined threshold. When an alert is received, the administrators can analyze the situation and make any necessary changes to the workload. This process works well not just because of the alerts, but also because Marable made a concerted effort to educate everyone on faculty and staff who interacts with the system about what her team was doing.

“We put on a presentation for clients and staff so everyone knew about the types of goals we had in mind. I wanted to make sure we were all speaking the same language, and so when I talked about rolling four-hour averages, they would understand what that meant,” Marable says. “There were a lot of people in EITS and even clients who really picked up the ball. So when we would call and ask why a process was running a certain way, they would figure out the business rules and see if that’s the way the process needed to be run or if it could be tweaked.”

A Fine Example

The school continues to run yearly analyses to make sure it’s achieving or exceeding its goal of reducing peak-time processing. During one such recent analysis, Mainline told the school it was, according to Marable, “a poster child for sub-capacity tuning.” For example, those peak registration periods in January and August began to shift, with more and more students registering later than they had in the past. The school, of course, noticed this trend and began shifting workloads to accommodate the change.

Notably, Marable credits SHARE for many of the tips and techniques she has employed in her environment, saying, “Even this year, they were talking about memory and DB2 buffer tools to help performance.” She and the University of Georgia were used as an example of this, including how to improve mainframe performance while also cutting costs.

That the school won the 2008 SHARE Award for Excellence in Technology was somewhat of a surprise to Marable. In fact, the university hadn’t even considered applying for the award until Mainline’s poster-child comment. And even then, Marable wasn’t sure the university would even place let alone win, given the competition. But her cost-cutting story was enough to tip the scales in the school’s favor, a reminder that nearly every organization—public, private and commercial—is facing decreasing IT budgets—and that there are ways to alleviate them.

Jim Utsler, IBM Systems Magazine senior writer, has been covering the technology field for more than a decade. Jim can be reached at jjutsler@provide.net.



Advertisement

Advertisement

2019 Solutions Edition

A Comprehensive Online Buyer's Guide to Solutions, Services and Education.

A Lesson Learned

How Boston University is making the most of its IBM zSeries and TotalStorage environment

An Educational Experience

The University of Georgia cuts costs via sub-capacity pricing

Creating a Community

How Palm Beach Community College learned that consolidation and virtualization can result in more than just ease of administration and lower overhead costs.

IBM Systems Magazine Subscribe Box Read Now Link Subscribe Now Link iPad App Google Play Store
Mainframe News Sign Up Today! Past News Letters