[5645] in Kerberos
Krb 5.5 Encrypted Login Sessions
daemon@ATHENA.MIT.EDU (Joe Ramus)
Wed Aug 9 20:45:46 1995
Date: Wed, 9 Aug 95 17:36:43 PDT
From: ramus@nersc.gov (Joe Ramus)
To: Rita.Bettenhausen@quickmail.llnl.gov, hartmans@MIT.EDU
Cc: kerberos@MIT.EDU
>>From: Sam Hartman <hartmans@MIT.EDU>
>> >>>>> "Rita" == Rita Bettenhausen <Rita.Bettenhausen@quickmail.llnl.gov> writes:
>>
>> RE> Encryption in telnet for K5.5 ? 8/9/95
>> Rita> According to the patches submitted by P-Pomes@uiuc.edu, in
>> Rita> the Makefile for libtelnet, DES_ENCRYPTION must also be
>> Rita> defined. I have not yet tested it.
>>
>> I don't know if this has been pointed out in the past, but it
>> was the intent of the Kerberos team to leave encryption compiled out
>> of Beta5 telnetd. There is no standard, and among other things, we'd
>> rather wait until this is standardized and all man-in-the-middle
>> attacks are eleminated before enabling it.
>>
>> --Sam
I compiled telnet & telnetd with both ENCRYPTION & DES_ENCRYPTION defined.
It now trys to encrypt all of the traffic on the net connection.
But there seems to be a Bug because my display shows only a lot of strange
random characters. Conclusion:
The data stream is encrypted but the decryption fails.
Then I tried the encryption option with krlogin & krlogind.
That works using DES encryption.
I find the attitude of the "MIT Kerberos team" as expressed by Sam
to be unreasonable. I hope there are others on the team who are
more willing to use the tools that we have today to improve network
security. In particular:
"There is no standard"
But we can start with DES for now. It works as shown by krlogin.
Yes, there are all of the export issues. But that is a different matter.
"wait until this is standardized"
That could be a long wait. It would be better to use what we have
now and adopt standards when they exist.
"man-in-the-middle attacks are eliminated"
This might be a real problem. But we must face reality here.
If I do not use encryption, a man-in-the-middle attack is far easier.
They just need to read the traffic as it goes by on the net. Hijacking
a connection takes more trickery. But if all data is encrypted on
the net, it will require a far more sophisticated attacker to mount
a successful attack.
I would not use a DES encrypted session to send Top Secret data or
Extremely Sensitive data. But it is a reasonable choice for something
like System Administration over the net. Many people now are doing
System Administration over the net with clear text sessions. So using
an encrypted session is at least a big improvement.
----------------------------------------------------------------
| Joe Ramus NERSC Livermore (510) 423-8917 ramus@nersc.gov |
----------------------------------------------------------------
