Re: [DMM] Kathleen Moriarty's Discuss on draft-ietf-dmm-mag-multihoming-04: (with DISCUSS)

["Sri Gundavelli (sgundave)" <[e-mail]>]

Hi Kathleen,

Thank you so much. Please see inline.




Good morning,

On Wed, Aug 16, 2017 at 11:04 PM, Sri Gundavelli (sgundave)
<[e-mail]<mailto:[e-mail]>> wrote:
Hi Kathleen and Hilarie,

Thank you for reviewing the documents and the feedback. Please see
inline.



For security considerations, the authors predictably defer to RFC5213,
³Proxy Mobile IPv6², and assert "no impact on the protocol security".
However, there is one security issue that is mentioned in RFC5213 that
is
exacerbated by the current draft. I.e.,

  To address the threat related to a compromised mobile access gateway,
the
local mobility anchor, before accepting a Proxy Binding Update message
for a
given mobile node, may ensure that the mobile node is attached to the
mobile
access gateway that sent the Proxy Binding Update message.


A MAG is required to prove the presence of a mobile node¹s presence on
its
(ingress) access link. That assertion needs to be validated by the LMA.
But,
MAG using a single tunnel or multiple tunnels has no relation to this
issue.
The LMA identifies the MAG using the MAG-NAI and not using Care-of
Address;
a new option, MAG Identifier option is defined in section 4.2 for this
purpose.


Could you add some text to the security considerations to explain why
there is no concern then related to split tunneling/data leakage to
the incorrect tunnel?

I assume you are referring to the below discussion point.  We can add some text, but this IMHO is absolutely a non-existent scenario; not trying to ignore a valid issue. Out of order delivery with such delays, where one MAG releasing the address and another MAG obtaining the same IP address with the associated address configuration delays, MN attachment triggering a PBU at that juncture, and a PBU/PBA exchange,  a tunnel establishment and a packet from the previous MAG showing up at the new MAG is not a technical possibility, IMO. I am not even sure how to frame this.

But, I am happy to include any text, but IMHO it will be a text for non-existent scenario. Let me know.



Regards
Sri











Is there any reason to worry about reuse of CoAs? Could packets from
one
tunnel get a CoA that was recently used by another tunnel, and could
delayed
packets get routed through the wrong tunnel? Just asking.

The tunnels between LMA and MAG are dynamically established after
protocol
signaling. The idea of CoA re-use between MAG¹s and delayed packets
getting
delivered to a different MAG is impossible to realize even in lab
conditions. It is possible there are two MAG¹s in a given access
network and
one looses the CoA and the other MAG gets the name address. But, the
tunnels
comes up after PBU/PBA exchange which introduces some delay, and so the
possibility of packets from previous MAG-era getting showing up at the
new
MAG is nearly impossible and is not worth mentioning it, IMO.





Nits. On page 3 there is a paragraph beginning ³In the continuation of
c,
a Proxy Mobile IPv6 ..." There is no explanation of ³c". Is this a
remnant
of a list of items "a, b, c"?

Fixed in -04 version





On page 4 there is Figure 1 showing four flows and two tunnels. The

We just wanted to hint that the Flow-4 is based on per-packet load
balancing
using both the paths, whereas the other flows are routed based on
Per-flow
load balancing. But, I think the comment is still valid. We should show
the
use of a single forwarding mode and not mix both the modes. Minor nit,
will
fix it.

Thank you for fixing this.
Kathleen



Thank you for your review.


Regards
Sri







‹‹‹‹‹‹‹‹‹‹‹‹‹‹‹‹‹
Security review of
MAG Multipath Binding Option
draft-ietf-dmm-mag-multihoming-03.txt

Do not be alarmed. I have reviewed this document as part of the
security directorate's ongoing effort to review all IETF documents
being processed by the IESG. These comments were written primarily
for the benefit of the security area directors. Document editors and
WG chairs should treat these comments just like any other last call
comments.

If you've been frustrated by being limited to only one IP tunnel
between a mobile access gateway and a local mobility anchor, then
you'll be glad to know that this draft fixes the problem and enables
multiple care-of addresses and IP tunnels. Now mobile devices can be
assigned to any applicable tunnel between the MAG and the LMA.

For security considerations, the authors predictably defer to RFC5213,
"Proxy Mobile IPv6", and assert "no impact on the protocol security".
However, there is one security issue that is mentioned in RFC5213 that
is exacerbated by the current draft. I.e.,

To address the threat related to a compromised mobile access gateway,
the local mobility anchor, before accepting a Proxy Binding Update
message for a given mobile node, may ensure that the mobile node is
attached to the mobile access gateway that sent the Proxy Binding
Update message.

The RFC has no recommendation for a solution, but because there are
now multiple tunnels, this assurance may be more difficult to obtain.
For example, if the LMA expects to contact some kind of trusted entity
that is keeping track of the mobile devices that the MAG is sending on
a tunnel, then the MAG and LMA may now have to keep track of multiple
trusted entities, one for each tunnel. Whether or not this is a
realistic scenario is not something that I can judge because RFC5213
punts on what seems to be an important security issue.

Is there any reason to worry about reuse of CoAs? Could packets from
one tunnel get a CoA that was recently used by another tunnel, and
could delayed packets get routed through the wrong tunnel? Just asking.

Nits. On page 3 there is a paragraph beginning "In the continuation
of c, a Proxy Mobile IPv6 ..." There is no explanation of "c". Is
this a remnant of a list of items "a, b, c"?

On page 4 there is Figure 1 showing four flows and two tunnels. The
text immediately preceding that says that "Flow-1,2 and 3 are
distributed either on Tunnel-1 (over LTE) or Tunnel-2 (over WLAN)",
but the diagram shows Flow-1 on Tunnel-1 and Flow-2,3 on Tunnel-2.
I think the text should indicate that the first three flows are
each assigned to a single tunnel. The authors probably meant that
either Tunnel-1 or Tunnel-2 could have been assigned, but the choice
was to put Flow-1 on Tunnel-1 and the other flows on Tunnel-2.
I had to read over the text several times before I was sure of the
intent.

Hilarie

On 8/1/17, 1:13 PM, "Kathleen Moriarty"
<[e-mail]<mailto:[e-mail]>>
wrote:

Kathleen Moriarty has entered the following ballot position for
draft-ietf-dmm-mag-multihoming-04: Discuss

When responding, please keep the subject line intact and reply to all
email addresses included in the To and CC lines. (Feel free to cut this
introductory paragraph, however.)


Please refer to
https://www.ietf.org/iesg/statement/discuss-criteria.html
for more information about IESG DISCUSS and COMMENT positions.


The document, along with other ballot positions, can be found here:
https://datatracker.ietf.org/doc/draft-ietf-dmm-mag-multihoming/



----------------------------------------------------------------------
DISCUSS:
----------------------------------------------------------------------

Thanks for your work on this draft.  I had the same concern as the
SecDir
reviewer in reading the draft, the concern about leaking traffic as a
result
of
multiple tunnels is not addressed in the security considerations
section.
Hilary's writeup is quite helpful

https://www.ietf.org/mail-archive/web/secdir/current/msg07446.html