Hello all,<BR>I am developing an extranet application where data is partitioned based on the user. I would appreciate any feedback on the ideas below.<BR><BR>The story so far...<BR>All list pages need to have paging. Instead of flowing the start and count parameters for each list page into the sql server I am thinking about caching the entire resultset ( a typed collection) when the user hits the list page for the first time. All subsequent requests for different pages will be taken from the cache.<BR>The data will be stored using the users session id as a key. ( One data cache per user ).<BR>The cache will expire after 1 minutes or so and if the cache fills up I will dump the entire cache. I could implement an LRU scheme which would further improve efficiency.<BR><BR>Only shallow copies would be cached. A worst case would be 100 concurrent users, 10000 object resources, each record 10K approx. This would work out at 10GB. This situation would be extremely unlikely.<BR><BR>What are your thoughts on this ? Is flowing start and count parameters into the sql server fine or is their some other cool way of doing this that will make me fall off my seat ? Have I said cache enough ?