Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!iuvax!cica!tut.cis.ohio-state.edu!pt.cs.cmu.edu!andrew.cmu.edu!jb7m+ From: jb7m+@andrew.cmu.edu (Jon C. R. Bennett) Newsgroups: comp.protocols.kerberos Subject: Re: Dictionary attacks Message-ID: Date: 15 Jun 90 22:08:57 GMT References: , <1990Jun15.164640.5090@terminator.cc.umich.edu> Organization: Carnegie Institute of Technology, Carnegie Mellon, Pittsburgh, PA Lines: 23 In-Reply-To: <1990Jun15.164640.5090@terminator.cc.umich.edu> > In article Ted_Anderson@TRANSARC.COM writes: > > I believe that the confounder was introduced to (surprisingly) confound > this attack. The confounder is a random number at the beginning of the > encrypted packet, thus removing the possibility for the attack above. > > wes you better make sure that the (psudo)random number generator you are using is REALY good, if not, all you are doing is effectivly give the attacker more cleartext to play with! in general having the server send the ticket back encrypted in the users key in reponse to any request is bad simply because it gives away information with can be used in an attack, forcing the client to send the request encrpted in the users key (where the request can be time stamped to prevent reuse by someone with access to the subnet and have a random number appended to the front to hinder stream encrypted DES attacks) prevents the release of more information then necessary. jon