SYSCO Corporation Solves Mainframe Performance Problems Using ASG-TMON™ Products
|Industry||Food Marketing and Distibution|
SYSCO Corporation, headquartered in Houston, TX, is the largest marketer and distributor of foodservice products in North America. Operating from 124 distribution facilities, SYSCO provides products and services to about 370,000 restaurants, hotels, schools, hospitals, retirement homes, and foodservice operations throughout the United States and portions of Canada. With sales of $21.8 billion for its fiscal year 2001, SYSCO distributes a wide variety of fresh and frozen meats, seafood, poultry, fruits and vegetables, bakery products, canned and dry foods, paper and disposable products, sanitation items, dairy foods, beverages, and kitchen and tabletop equipment, as well as medical and surgical supplies.
SYSCO has a leading-edge data processing center that includes a MVS mainframe, AS/400®, and RS/6000® midrange computers. Problems such as lockups, looping, storage shortages, poorly optimized queries, and network problems used to frequently require a long and convoluted troubleshooting process due to limitations in standard CICS monitoring functions.
“Recently, we had some issues where we were short on storage during peak periods,” said Eric Vadla, team leader of the mainframe group at SYSCO. “In the past, we probably would have added storage to all of the pools, which would have wasted space, or we might have experimented by adding storage to one pool at a time to see if the problem went away. That would have taken extra time.”
Information technology managers evaluated several different performance monitoring tools and selected ASG-TMON™ for CICS , ASG-TMON™ for MVS , and ASG-TMON™ for DB2® to quickly detect and diagnose most problems.
“Let’s say there was a problem an hour ago that just came to my attention,” Vadla said. “The system was locked up tighter than a drum for about 15 minutes. The first thing I do is login to TMON, tab to the history display, and scroll back to the time period in question. Then, I request a list of transactions that had more than a 20-second response time. This immediately gives me a list of the bad offenders – transactions that might have caused a temporary lockup. The details screen provides me with information such as what section of the program is executing, what resources are in use, and which databases have been called. This level of detail usually makes it possible to fairly easily determine, with pretty good precision, what type of problem I am facing.”
Vadla and other systems programmers also use the monitoring software to troubleshoot SQL queries. Poorly optimized SQL code can cause a major performance hit. The programmers are often able to detect problems by periodically launching the TMON task monitor and reviewing what is running.
Vadla said that the ASG monitoring software is typically the first tool that he turns to whenever a problem is reported. "Most problems are intermittent which, in the past, meant that they were difficult or impossible to solve because we had no way to recreate them,” he said. “Now, I can go back in time and determine exactly what the operating conditions were when the problem occurred. I begin by looking at system utilization and whether there were exceptions such as a high paging rate or heavy I/O activity. The program is interactive, which means I don’t have to wait for a paper report but can just pull the information right off the screen.”
“All in all, the switch to more intensive monitoring of our mainframe systems has helped us keep our users happy by solving performance problems faster than was ever possible in the past,” Vadla concluded. “The information provided by the ASG monitoring software in nearly every case helps us track down the source of the problem and, in some cases, even provides the platform for us to solve it in a single step.”