Re: too many false alarms
Bennett Todd <[email protected]> Fri, 25 Jan 2008 00:46:55 +0000
| Newsgroups | gmane.comp.sysutils.loganalysis |
|---|---|
| Message-ID | <[email protected]> |
2008-01-24T23:04:43 Jon Stearley: > what false alarm rate do you tolerate for your current monitoring system? Monitoring what? How expensive is a false negative? How often do real alerts come in? Failed attacks on a successfully-hardened server, as long as you know the accuracy rates you can deduce the stats which are all care about. If missed alerts are enormously expensive and alerts come in no oftener that once a week, it could be that a 300% false positive rate would still be fine. If you're keeping a desk of 50 people racing full-time and are neglecting real alarms because you can't keep up, a 2% false positive might be worth a lot of engineering effort to tighten things down that much more. -Bennett _______________________________________________ LogAnalysis mailing list [email protected] http://www.loganalysis.org/mailman/listinfo/loganalysis
signature.asc
(application/pgp-signature, 189 B)
-----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) iD8DBQFHmTF+HZWg9mCTffwRAtUKAKDArmdBHnXGQdWGPNcJwrjkZAfipgCeMp8O 6YHCw0QRcHS8x/4SVL1GT1U= =V2vh -----END PGP SIGNATURE-----