[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: DMTF SAAction restriction
Correct me if I am wrong. But why cannot the three tries be bundled into one
action with multiple proposals? My understanding is that the purpose of
having multiple proposals is exactly such that if the responder cannot
accept the first one (hence fails), it has the choice to choose another one
from the list.
Man Li
Nokia
5 Wayside Road, Burlington, MA 01803
man.m.li@xxxxxxxxx
phone 1-781-993-3923
GSM 1-781-492-2850
-----Original Message-----
From: ext Eric Vyncke [mailto:evyncke@xxxxxxxxx]
Sent: Tuesday, February 06, 2001 11:17 AM
To: Man.M.Li@xxxxxxxxx; ipsec-policy@xxxxxxx
Subject: RE: DMTF SAAction restriction
We talked about different things:
1) when ISAKMP is trying to build a SA, it sends multiple proposals
2) in the DMTF model, for one data traffic pattern, multiple actions are
tried
until success:
a) let's try ISAKMP with 3DES+RSA or 3DES-pre-shared with SG1
b) if it fails, let's try ISAKMP with 3DES+RSA or 3DES-pre-shared with
SG2
c) if it fails, let's try DES-pre-shared with SG3
Obviously, the DMTF model allows the configuration of a complex IKE proposal
(see IKEProposal class which is associated to SANegotiationAction).
Hope this helps
-eric
At 09:30 6/02/01 -0600, Man.M.Li@xxxxxxxxx wrote:
>Is the "try until one succeeds" being described in any RFCs or is it only
>ONE way of implementation? I recall ISAKMP allow initiator to propose
>multiple proposals and multiple transforms for responder to choose from.
But
>I do not recall seeing this "try until one succeeds".
>
>
>Man Li
>Nokia
>5 Wayside Road, Burlington, MA 01803
>man.m.li@xxxxxxxxx
>phone 1-781-993-3923
>GSM 1-781-492-2850
>
>-----Original Message-----
>From: ext Lee Rafalow [mailto:rafalow@xxxxxxxxxxxxxxx]
>Sent: Tuesday, February 06, 2001 9:48 AM
>To: ipsec-policy@xxxxxxx
>Subject: Re: DMTF SAAction restriction
>
>
>The approach that we've thought through so far would use your first example
>to set up the tunnel. We could explore your suggestion to add a sematic to
>the action list, but that complicates it significantly. We already have
>semantics for the initiator and responder that are variations on "try until
>one succeeds." Recall that for the initiator it's try the first one, if
>that fails try the second, etc. For the responder it's a little more
>complicated: respond from the first action that is appropriate to the
>request (this allows a single rule to deal with both aggressive and main
>mode requests).
>
>
>To add semantics for "do ordered list of actions" would require a fairly
>complicated grammar for grouping actions into execution strategy groups (do
>all, do first successful).
>
>A way out of this might be to use the structured rules being proposed in
the
>PCIM Extensions draft (currently being written) but that'll have to wait
>until we have a PCIMe draft to discuss.
>
>Lee
>----- Original Message -----
>From: <Man.M.Li@xxxxxxxxx>
>To: <HORMAN@xxxxxxxxxx>; <ipsec-policy@xxxxxxx>; <jamie.jason@xxxxxxxxx>;
><Man.M.Li@xxxxxxxxx>
>Sent: Monday, February 05, 2001 4:14 PM
>Subject: RE: DMTF SAAction restriction
>
>
>> Security per se may not be a problem. But if only a small percentage of
>> traffic needs multiple nested protection and we end up with giving all
the
>> traffic multiple protections, it would be a waste of resources (e.g.,
>> processing resource for encryption). Security comes with a price and I
>> question if service providers or corporate IT managers will be happy
about
>> giving the free ride at the expense of network resources.
>>
>> Man Li
>> Nokia
>> 5 Wayside Road, Burlington, MA 01803
>> man.m.li@xxxxxxxxx
>> phone 1-781-993-3923
>> GSM 1-781-492-2850
>>
>> -----Original Message-----
>> From: ext Hilarie Orman [mailto:HORMAN@xxxxxxxxxx]
>> Sent: Monday, February 05, 2001 3:10 PM
>> To: ipsec-policy@xxxxxxx; jamie.jason@xxxxxxxxx; Man.M.Li@xxxxxxxxx
>> Subject: RE: DMTF SAAction restriction
>>
>>
>> Why would it be erroneous to include a tunnel gratuitously?
>> Wrt to security, what is the problem?
>>
>> Hilarie
>>
>> >>> <Man.M.Li@xxxxxxxxx> 02/05/01 11:22AM >>>
>> What selectors are you looking for in your second search after the
>transport
>> rule? Is it a selector for the already encrypted packets (hence protocol
=
>> ESP, for example) or for the original packets (protocol = tcp)? I see
>> problems in both cases. In the first case, if there is a rule to set up a
>> tunnel for ESP packets, we may include some other traffic that are ESP
>> transport protected but should not be included in the tunnel. In other
>> words, those traffic should only be transport protected.
>>
>> In the second case, we cannot reach the second selector for tunnel set up
>> because first matches will stop the search.
>>
>> A simple solution would be to remove the restriction and let SAActions to
>> indicate an ordered list of actions to be executed.
>>
>> Man Li
>> Nokia
>> 5 Wayside Road, Burlington, MA 01803
>> man.m.li@xxxxxxxxx
>> phone 1-781-993-3923
>> GSM 1-781-492-2850
>>
>> -----Original Message-----
>> From: ext Jason, Jamie [mailto:jamie.jason@xxxxxxxxx]
>> Sent: Monday, February 05, 2001 11:01 AM
>> To: 'Man.M.Li@xxxxxxxxx'; ipsec-policy@xxxxxxx
>> Subject: RE: DMTF SAAction restriction
>>
>>
>> Lee can correct me if I am wrong on this one, but I believe we had this
>same
>> discussion during a conference call between Lee, Eric Vyncke, and myself.
>I
>> believe the solution we came up with was that policy evaluation would
>first
>> encounter the transport rule. Policy evaluation would then search to
>> determine if there was a tunnel rule that would apply (it would do this
>> recursively as you may have tunnels in tunnels). This update probably
>> hasn't hit the MOF file or whitepaper yet.
>>
>> Jamie
>>
>> > -----Original Message-----
>> > From: Man.M.Li@xxxxxxxxx [mailto:Man.M.Li@xxxxxxxxx]
>> > Sent: Monday, February 05, 2001 7:16 AM
>> > To: ipsec-policy@xxxxxxx
>> > Subject: DMTF SAAction restriction
>> >
>> >
>> > Hi,
>> >
>> > The DMTF policy model Section 3 first paragraph indicates
>> > "The IPsec model
>> > restricts the use of SAActions to an ordered choice rather
>> > than a list of
>> > actions to be executed." I am wondering how the following
>> > situation would be
>> > handled with this restriction. We had similar discussions
>> > among IPsec PIB
>> > authors.
>> >
>> > A (host)===========C(gateway)---B(host)
>> >
>> > A and C are connected to public Internet and B is connected
>> > to C. To protect
>> > TCP traffic between hosts A and B, an IPsec security association in
>> > transport mode needs to be established between hosts A and B.
>> > In addition,
>> > an IPsec security association in tunnel mode may be set up
>> > between host A
>> > and the gateway C that protects the LAN host B resides.
>> >
>> > In this case, A takes one action to set up an association
>> > between A and B.
>> > In addition, A should also set up a tunnel between A and C.
>> > From A's point
>> > of view, there are MULTIPLE actions to be taken in that order.
>> >
>> > How would you specify a policy to A if you are not allowed to
>> > specify a list
>> > of actions to be executed?
>> >
>> >
>> > Man Li
>> > Nokia
>> > 5 Wayside Road, Burlington, MA 01803
>> > man.m.li@xxxxxxxxx
>> > phone 1-781-993-3923
>> > GSM 1-781-492-2850
>> >
Eric Vyncke
Distinguished Engineer Cisco Systems EMEA
Phone: +32-2-778.4677 Fax: +32-2-778.4300
E-mail: evyncke@xxxxxxxxx Mobile: +32-475-312.458
PGP Key available on request MOBILE HAS CHANGED ON 11th November 2000