tas&sachapman@aep.com&mfusage&6.0.0&&&mfusage&&&bcarico@actscorp.com&N.Y.N.N.Y.N.N
1.1.o.o.o.N.0&mfusage.001&mfusage.002&mfusage.003&mfusage.004&mfusage.005&mfusage.006
&::1.0:1:1:0&::2.0:1:1:0&::3.0:1:1:0&::4.0:1:1:0&::5.0:1:1:0&::undefined.0:1:1:0
&We have about 20,000 employees, about half of which have mainframe user ids, primarily due to the fact that one of our primary time-entry systems runs on the mainframe. We have about 5 million customers that we serve, which is of interest since the mainframe also hosts the customer system. While most of our customers dont directly interact with us every month, most do at least get a bill each month. :) Note that "customers" are really "accounts", which represent anything from a single-family home to a large industrial customer. So the number of people enjoying our primary product (electricity) is actually larger than 5 million.::&The easiest volumes are for the customer system, although certainly we have other important mainframe systems as well. But for the customer system, we process meter readings and produce bills for about a quarter of a million accounts per night. We run about 5 to 7 million CICS transactions per business day for the customer system. ::&Our mainframe systems handle customer accounting, trouble entry and reporting, materials management, work management, emissions allowance management, time reporting, and probably a few more things that were not even well aware of.::&We do not currently run IFLs or zAAPs. We have run Websphere for a long time. Since version 4 (currently we have 1.2, 3.5, 5.1, and 6.0 installed and running) it has become a major problem though in relation to the amount of work that its doing. Currently the version 5/6 WAS instances have a memory footprint greater than half of all our installed real memory. Maintenance/upgrade work for WAS has been very problematic and time-consuming at our site for some reason. It takes hundreds of CPU seconds (on a z900) just to start an instance.
We have a growing amount of work coming into DB2 from off-platform via DDF. This is coming from multiple software packages (e.g. Business Objects, QMF for Windows, SAS, etc.) and at times this workload can be significant. And at times it has caused problems for other workloads. While weve addressed most of the CPU contention issues via WLM, some potential exposures dont have such easy solutions, such as ad hoc users consuming DB2 resources such as buffer pool or RID pool space.::&At this time it does not seem likely that the mainframe will "go away" anytime soon. Primarily because the customer system can not be easily replaced on any platform--they investigated doing that a few years ago. In general (other than some DDF workloads), the applications that are there are getting good service and so in general they are happy for the time being. Having said that, we are under constant cost pressure and a compelling cost argument could change that. However, as people have started to better understand the costs of other platforms, it seems that in the past few years the mainframe has been looked down upon less than it was say 5 years ago. Unfortunately, it is not being promoted as a strategic direction for new workloads though. Weve had some small success getting some small new applications, and if some major purchased app came along that supported running on z/OS, I think we might have a chance to get that business. At least a better chance than we had a few years ago.::&You may contact me for clarification about anything if you wish, but I would prefer that the information remain confidential. Thanks.::
Not Scored by Definition