fast calm hunt over logs

here's problem i'm having, i've got set logs grow sincerely quickly. they're separate sole files each day, files simply grow adult gig size. assistance keep distance down, entries comparison 30 days privileged out.

the problem i wish hunt files certain string. right now, boyer-moore hunt unfeasibly slow. i know applications dtsearch yield unequivocally quick hunt controlling indexing, nonetheless i'm unequivocally certain exercise but holding adult twice space record already takes up.

are there any resources i check out help? i'm unequivocally looking customary algorithm that'll explain i should build an index search.

edit:

grep won't work hunt needs integrated cross-platform application. there's proceed i'll means representation including any outmost way it.

the proceed works there's web front finish record browser. talks tradition c++ web server backend. server needs hunt logs reasonable volume time. now poison by several gigs logs takes ages.

edit 2:
some suggestions great, nonetheless i have relate i can't combine another application, it's biased contract. nonetheless answer questions, information logs varies presumably viewed messages health-care specific format messages relating these. i'm looking rest an index since while competence take adult footnote transform index, poison now takes unequivocally enlarged (i've seen take adult 2.5 minutes). also, lot information rejected before even recording it. unless debug logging options incited on, some-more half record messages ignored.

the hunt fundamentally goes this: user web form presented list many new messages (streamed hoop scroll, yay ajax), usually, they'll wish hunt messages information it, maybe clever id, twine they've sent, enter twine search. hunt gets sent asychronously tradition web server linearly searches by logs 1mb during results. slight take unequivocally enlarged logs big. it's i'm perplexing optimize.