Re[2]: Long-running queries

______________________________ Reply Separator _________________________________ Subject: Re: Long-running queries Author: Nils Myklebust at IDGNOR Date: 27.07.95 12:52 > aus@vax.rz.uni-wuerzburg.d400.de (Hans M. Aus) writes: > > How many ad hoc and or long running queries need today`s data ? Weekly, > monthly and quarterly reports and billing probably don't. > >>>> No, but 5 minutes ago someone just finished loading 200.000 adresses into our database. 150.000 of them got connected to existing adresses where lots of additional information is available (20 tables with upto several hundred records per adress). Now the marketing department is going to deside which 4000 of these shall have a free issue of a magazine and the distribution data must be finished within two hours. The analysis envolves reading at least 500.000 to 1 mill. records doing some groupings, sum, avg ++ The data must be inserted into the OLTP database as the customer service department needs it, both for genearal maintenance, and when someone calls in. Of course all this must be done while OLTP type operations are running. The above didn't happen right now, but it is a very real senario for us. There are lots of cases like this. (Even billing we would *not* like to do on anything but todays (or rater last minute) data, and that sometimes means reading and updating 100.000 records or more and inserting as many.) There is more to modern use of databases then the old fashined weekly/ monthly report way of doing bussines. So much more. Nils.Myklebust@ccmail.telemax.no NM-data, Dalsbergstien 7, N-0170 Oslo, Norway My opinions are those of my company