[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Nodes and Users



> From: rivest@theory.lcs.mit.edu (Ron Rivest)
> IPSEC needs to provide EXPLICT answers to these questions.  The Photuris
> spec needs to be based on these answers.  This cannot be swept under the
> rug any longer.
>
No rug sweeping has been attempted, we have discussed this at length in
the Security Architecture.


> * On page 2 of the Photuris draft, it is said that the Security Association
>   is established between "two nodes".  What is a node??? (A machine with
>   an IP address??)
>
A node is a standard well-known IP term for an entity with one or more
interfaces.  Interfaces have IP addresses.


> * On page 45 of the Photuris draft, it is specified that the Certificate
>   (when MD5 is the `Signature Choice') is an "email address".  Are we
>   envisioning that the Security Association may link users?  (I imagine
>   that some would like the answer to be "yes".)  Or was this intended
>   merely to be an IP address?
>
User oriented keying is a fundamental goal of IPSec.  Again, RFC-1525.


> The problem:
>
>   If I try to establish a Security Association with a specific user, or
>   a specific process, on another machine, then I need some way to specify
>   who I wish to talk to in my initial Key_Request.
>
Nope.  Process selection is done with the headers _inside_ AH or ESP.


>   With Photuris as it is, the Responder is not identified by the Photuris
>   initiator to any finer granularity than the IP address to which the
>   Photuris packets are sent.
>
Correct.

Right up until the Signature exchange.  But I already have language to
this effect:

    To provide user-oriented keying, or create multiple Security
    Associations with different parameters, the sender can either
    initiate multiple Photuris exchanges, or send a Change_Message.

    The Destination MUST be capable of maintaining multiple Security
    Associations (SPI values) for each Source.

    It is the responsibility of the Source to internally segregate the
    different session-keys provided by the Destination.


>   If the intent is to support Security Associations between entities at
>   some other level of granularity, then the initiator needs to be able
>   to name who (or which process, whatever) at that IP address he desires
>   to communicate with.
>
We've talked about this, but currently the protocol only specifies who
the SA is _from_.

The initiator decides to start a Photuris exchange when:

 1) it has a datagram which it wishes to send with privacy.

 2) it has received an ICMP message from a destination which indicates a
    requirement for authentication.

Other needs to initiate a Photuris exchange are likely to be a matter
for considerable debate.  (and thus have been left out)


> As it stands, Photuris apparently only supports the establishment of
> Security Associations between entities named by IP addresses. Is this
> what is intended?
>
It establishes shared-secrets between a user and an IP address (in the
case of a server).

Of course, the two most likely scenarios are machine to machine
(firewalls making virtual private networks), or user to user (one person,
at least one CPU)!

Bill.Simpson@um.cc.umich.edu
          Key fingerprint =  2E 07 23 03 C5 62 70 D3  59 B1 4F 5E 1D C2 C1 A2


Follow-Ups: