Rob van der Heij wrote: >We recently looked at data from a similar configuration. The data was >sufficient to explain why their Linux server ran about 10 times as >slow as they anticipated. A factor of 3 could be helped by proper >configuration, the rest was overhead in V/SIE basically. >Linux is using things that are not implemented in V/SIE, so CP has to >do a lot of smoke and mirrors to maintain the illusion.
Well gee, so we were just ahead of our time! Glad to finally understand it. This was of course my guess, but without any data that wasn't useful. And at the time (late 2002) everybody who might have known (Barton, Bitner) had no data either; we all said "Gee, it should work fine". ...phsiii (One mystery of life down, 2,147,483,646 to go)
