Thanks Dave for the reply.
<<which time-out did the vendor specify? >>
Connection time-out is what they specified...
This is what their report says:
It was discovered that one of the reasons w3wp.exe tends to max out a cpu is that it is spending too much of its processing time in Garbage Collection (GC). Using Windows 2003 built in Performance Monitor, it was determined that w3wp was indeed spending most of its time in GC – approximately 85%. One cause of this problem is doing too many String concatenations in loops within the code. We did not however observe any of these within the code. Another cause is the use of many data grids on a screen. Data grid memory is transferred around a lot and then discarded. The page in question does indeed have many data grids. However, Microsoft’s .Net Framework 1.1 does not handle such massive amounts of processing in a very efficient manner. As a result, these become extremely processor heavy transactions. This is why we are seeing the processors spike to 100%
Main Topics
Browse All Topics
by: Dave_DietzPosted on 2007-02-28 at 14:36:49ID: 18629601
Not sure at all what is meant by ".Net Framework 1.1 does not handle massive data processing efficiently" but it could easily be doing a lot of garbage collection depending on the code in question.
.Net will create a heap per processor on multi-proc boxes and here we would be seeing 2 heaps per processor due to the HT. If there is something causing GC to occur more often than normal it will cause GC to happen on *all* the heaps at the same time which actually means twice per processor here. GC isn't a problem in and of itself but it certainly can be part of a problem if the code is doing things to cause GC to happen too often or is handling data in odd ways that make GC more difficult.
Also, which time-out did the vendor specify? There are several and no single one is really the 'default'.....
Dave Dietz