[1370] in Kerberos_V5_Development
Re: kdc performance and rcache
daemon@ATHENA.MIT.EDU (Sam Hartman)
Tue Jul 2 20:58:35 1996
To: "Barry Jaspan" <bjaspan@MIT.EDU>
Cc: raeburn@cygnus.com, krbdev@MIT.EDU
From: Sam Hartman <hartmans@MIT.EDU>
Date: 02 Jul 1996 20:58:24 -0400
In-Reply-To: "Barry Jaspan"'s message of Mon, 1 Jul 1996 15:25:10 -0400
>>>>> ""Barry" == "Barry Jaspan" <bjaspan@MIT.EDU> writes:
"Barry> It seems pretty clear that the replay cache is a time sink
"Barry> and should be removed if doing so is safe. The most
"Barry> likely possible problem anyone has mentioned is the
"Barry> possibility of a known-plaintext attack if the replay
"Barry> cache is removed. Is it really feasible for this to
"Barry> present a threat even to single DES, given the krbtgt
"Barry> maximum lifetime and the KDC's maximum throughput?
I don't know but would guess not. One minor point: I'm think of AS requests not TGS requests, so it's the lifetime of a service key or password not the life time of the TGT. Also, I think it is the lookaside cache that prevents this attack not the replay cache; I see no reason not to toast the replay cache.
--Sam
P.S. I'm trying to remember who brought up the issue of the lookaside
cache and plain-text attacks. It was in an article on c.p.kerberos a
few months ago, where a user asked what happened when a client did a
UDP transmit.
