Wouldn't it be better to index those files on the file server, and query its "indexing server" over the network? Having 100s or 1000s of PCs in a company each crawl the multi-terrabyte (or probably even bigger) NFS server to make their own private index of it doesn't sound like a very optimal solution to me…