My impression is that HashTable will support a lot of data without "going geometric." So one alternative could be to have a single machine with 4gb RAM, use the /3GB option at startup (this gives you 3gb of user address space and 1gb of system address space, instead of the default 2gb each; but I've never seen an assertion that .NET will [or won't] work in that environment), and keep all the data in a "normal" HashTable. Use this same machine as a remoting or web services host to do the querying against data in RAM.
If words are involved, and there's no punctuation involved, replacing the strings with a series of "word code" values could make word-based searches much faster. Keeping the data in RAM as much as possible would definitely make things go faster. Sounds like a very interesting project! Good luck. At 04:56 PM 1/18/2005, Pradeep Tapadiya wrote >Thank you for your help. > >In a way, we are implementing our own file-based database, optimized for >just fast searches. Support Regex is not a must but a nice to have feature. > >I was hoping I could leverage classes offered by .NET (such as Hashtable) >and just extend them to deal with huge amount of data. Looks like we would >need to roll our own mechanism. > >Thank you once again for your help. > >Pradeep >[snip] J. Merrill / Analytical Software Corp =================================== This list is hosted by DevelopMentorŪ http://www.develop.com View archives and manage your subscription(s) at http://discuss.develop.com
