Well it's been about two days now without a single memory.exception.  I've got all my modules back up on the homepage except for the DNN Search Input module and the Search Results module.  The DNN Search Indexer is also disabled in the Scheduler.

What does this mean?  I don't know.  Traffic drops off during the weekend so maybe one of the other modules I've put back up is actually the culprit.

I'll know more in a couple days.

Any chance it is the DNN Scheduler though?  I would think that cat would be out of the bag by now if that was the case.  My guess is it's one of my custom modules I've put back up on the page but just hasn't gotten enough traffic hits to start being a problem.  We'll see.

I'm not a IIS guru, but their explanation of App Pool helping was pretty weak.  All this really did was help isolate your install as being a resource hog.  Unless they add more physical ram on the server, they are still apportioning the RAM across all AppPools to my knowledge.  I can't see how that can really help unless they dedicated a block of RAM to your pool alone.  The only real reason to do this other than if they would dedicate RAM to your pool is to have different performance settings of recycling and restarts of the app pool.  But they certainly singled you out once you paid that upcharge. 

Chris Patera or Cathal posted some info a month ago about hits to dnn - that he termed "being Sharepointed".  You might check those out and see if something (a bot or crawler) is indexing your site.  The post/blog was more in reference to bandwidth usage, but the server was busy working those requests. 

As for a memory leak - I'm trying to figure out how that would come into play.  If you're coding in C++.net perhaps, but I thought VB.net and C#.net did not fall victim to memory leaks the same as C adn C++ applications did.   I'd love to hear from some programming gurus here to explain further, as this would be a good learning point. 

Glad I saw your post too - I just signed up with PowerDNN for the free trial, and was close to trying these guys out first. 

Well - don't rule out the servers.  4GB is not a large amount of RAM and is the intro level for all servers we buy.  More so now with 64 bit OS versions.  My production SQL Server boxes have at least 8-16GB and app servers 4-12.  If they are putting many installs on the same server, 4GB can go away really quickly (I would think - but I have never hosted).  There are some specific startup switches they could be setting - but I'm not sure you'd be able to get that out of them.  The switch is the /3GB and /pae switch (for 32-bit OS only with >2GB and <= 16GB RAM) that tells the OS to use only 1GN for system addresses, and the user process space is increased from teh default 2GB to 3GB.  Without the /3GB switch - user processes including the worker processes for .net only have access to 2GB (according to MS).  If they add more RAM on a 32 bit machine, they should reference MS KB #283037 for the physical address extension.   The one for the 3GB switch is 291988, but I'm not 100% it applies to the .net app pools in IIS.  It does for SQL server. If they're running SQL and the IIS on the same box, then I'd wonder if that 4GB is not enough.  Maybe one of your modules is more resource intensive and causes that threshold to be hit with all the other sites on there. 

ASk if they have applied that switch and point them to the MS KB's.  It's worth a try if you can't figure it out, and never assume their engineers know about the RAM tuning.  Always ask if given the chance.