
Industry: Retail Clothing
Location: Europe
Ellos is a major European clothing retailer that sells clothing by mail, telephone, and on the Internet. Ellos is the leading direct marketer in Scandinavia and has more than 4 million customers. The company markets more than 30,000 items, and their staff of 400 telephone service representatives handles between 35,000 and 40,000 calls per day.
The Challenge
Ellos” business primarily runs on an IBM mainframe that runs OS/390”. Its CICS”
applications access IMS databases to service customers and fulfill product ordering
and tracking, and its e-commerce server is a packaged application that also runs on
the mainframe. The company’s systems programmers first became aware of a problem when
call center representatives began complaining of intermittent response time problems,
which impeded their ability to serve customers by telephone. Normally, the average
response time during order processing operations is 0.5 seconds. During short periods,
response time suddenly shot up to 10 seconds or more. Representatives that previously
had been able to handle 25 orders per hour were suddenly only able to handle 10. In
some cases, delays were so great that the representatives had to call the customers
back later in the day to complete the order. As suddenly as the problem appeared, it
would disappear, leaving everyone baffled.
John Enevoldson, systems programmer for Ellos, was called in to find a solution. He used standard administrative tools to provide a breakdown of the transaction time and was quickly able to determine that 98% of the response time was attributed to IMS/DBCTL. At one time, the performance and service levels of DL/I applications were easily monitored from within a CICS monitor. With the conversion to IMS/DBCTL, the CICS monitors lack visibility into IMS database calls, making it impossible to track the performance of DL/I CICS applications. Ellos was using ASG-TMON™ for CICS at the time and was very pleased with its performance, so it turned to ASG regarding an IMS/DBCTL monitoring solution.
The Solution
ASG-TMON™ for DBCTL, now part of the product
ASG-TMON™ for IMS, tracks thread detail
associated with IMS/DBCTL activity and size IMS buffers. This software helps systems
programmers and DBAs tune IMS applications by allowing them to view them as they
execute or selectively collect metrics for real-time and historical reporting and
proactively track service levels to ensure that applications meet business objectives.
“The first time I fired up TMON for DBCTL/IMS I could see exactly what was bogging down the application,” Enevoldson said. “It showed me the utilization of the different IMS buffer pools. I could clearly see that, during the time period when we were having the problem, several buffer pools were at 100 percent usage. We selectively increased the size of these buffer pools by 15 percent or so and looked again at the monitor for feedback. The utilization of the buffer pools dropped safely below 100 percent. That eliminated the response time problem. The productivity of our telephone service representatives immediately more than doubled. The whole problem took me less than 30 minutes to diagnose and solve. Call center management was delighted with the improvement, especially since the cost was close to zero. Since then we have run into several other IMS problems and have solved them nearly as quickly using ASG’s monitoring tool.”
Measurable Results
The TMON performance monitoring tools from ASG helped Ellos to identify CPU-intensive
transactions and provide clues as to how they could be tweaked to improve performance,
saving 25 percent of mainframe CPU cycles and avoiding the need for an expensive
processor upgrade.
“All in all, with ASG’s help the switch to more intensive monitoring of all aspects of our system, database and application has resulted in dramatic improvements,” Enevoldson said. “We have postponed the hardware upgrade expenditures for a considerable period of time and the response time improvements have increased the productivity of our telephone service representatives. Most important of all, the improvements that we have made have taught us the importance of comprehensive monitoring. We are now engaged in a ongoing process of system and application tuning that will provide continuing improvements for many years to come.”