On Tue, 15 Aug 2006 08:28:47 -0600, Lee Stewart <[EMAIL PROTECTED]> wrote:
>This reminds me of my first week or so working for IBM as a PSR (those >were kind of software CEs for those of you too young to know). I was >working in Dayton, OH and they sent me out to Wright Patterson air base >to pick up a dump. It turned out it was at FTD (the Foreign Technology >Division) and a very secretive bunch. I got to the lobby of the bank >vault like building and waited for them to bring me the dump. When >they did, it was like swiss cheese. Someone had gone through it with >an Exacto and excised all the data or hints of data that were in it. >The big pile of paper only weighed about half of what it normally would. . > >Lee (much happier with electronic dumps) Stewart > On a somewhat related topic, in the early 1990s when I was at IBM and working on VM performance related studies with the WSC, we needed CP moni tor data from customers who considered the data confidential because of useri ds and dasd volsers it contained. Because we needed the raw monitor data for the studies, I wrote two programs (one for VM/SP HPO data, and one for VM/ESA ESA feature data) that would sanitize the monitor data consistentl y throughout by changing all userids and DASD volsers contained in the moni tor records to generic USERnnnn and Vnnnnn values. The customer was left wit h a listing to map the edited values to the real values. We could call them u p and ask "Who the heck was USER0081? It was driving the system nuts at 9am ." They could look at the listing corresponding to the data we were analysi ng and know "Oh. That was MIDEAST or COMMIE1" and decide what (if anything) they might tell us. I recall I included a tailorable exclude list so that if they did not care about certain names (VTAM, RACF, OPERATOR, etc) they co uld list them in the exclude file and they would not be edited in the monito r data. That feature could make for a more meaningful study with less confusion on our part and less need to ask questions. The programs were written in VS/PASCAL because we needed to be able to provide one module to the customers to edit their data with no need for additional run time libraries. Pipelines were not an option at the time a s it was not a standard part of the VM products customers were running. It had to be able to read both tapes and disk files where the monitor data was recorded, too.