grep can trawl through gigs of files in a few seconds (and that's without indexes), so I don't see why would be an unhappy experience. In any case, there are alternatives; one would be to "recruit" a few other clients of the same organization for a distributed parallel search.