On Thu, Jun 30, 2005 at 05:21:40PM -0400, Ken Hornstein wrote: > >I created a presentation PDF a while back that I've placed on the Web > >which goes into detail on Kerberos enctypes in terms of how they are > >used, negotiated and controlled via *.conf parameters. It can be > >downloaded via my blog: > > > >http://blogs.sun.com/roller/page/wfiveash?entry=everything_you_wanted_to_know > > This is a good presentation. I have two comments: > > - In my experience, encryption type settings are the herpes of the Kerberos > world - once they get out "into the wild", they spread magically to > other systems and it's damn hard to get rid of them. If you have > your applicatation server enctypes set correctly, you should almost > never need them. I'd stress that setting these enctype settings on > the client should only be used rarely (say, you're using MIT Kerberos > that supports AES, but one of your developers uses a Java Kerberos > implementation that only supports single-DES). I know you mention this > in your last slide, but I'd put something stronger in there.
Yeah, I'll stress doing the "right thing" more as this is one of the reasons I created the presentation (helping admins understand the entype knobs to get it right or at least leave well enough alone). > - I know you know this, but on slide 8 you imply with the diagrams that > the ticket in the AS_REP is double-encrypted, and of course it's not; > only the session key and a few other bits are encrypted by the user's > long-term key. A minor nit, but I only wanted to point it out for > accuracy's sake. Thanks for the feedback. I'll tweak the presentation to make it more accurate. -- Will Fiveash Sun Microsystems Inc. Austin, TX, USA (TZ=CST6CDT) ________________________________________________ Kerberos mailing list [email protected] https://mailman.mit.edu/mailman/listinfo/kerberos
