Disclaimer - I do work in the INT area, but have not been involved in
the PANA WG.
When this work was chartered, I failed to understand its need and the
deployment use cases. Subsequently, about 3 years ago, I recommended
against the use of PANA for the needs of my ex-employer. More recently,
I have been advocating against the use of PANA in 3gpp2. With that
background, some high level confusions that I have about the use of
1. Reading the framework document, it is simply not clear to me as to
why PANA would be useful for any type of network. Actually, even after
reading some other PANA documents, this is not clear to me. It seems to
have become this complex suite of protocols that aren't specifically
buying anything. Why would I want to go through IP address acquisition,
for one, if I had a link layer that carried EAP, particularly for link
2. Link layer agnostic approach has often been cited as a reason to use
PANA. I must say that I don't understand how it is link layer agnostic,
when it depends on 802.11i 4-way handshake or the like for the secure
association protocol. In fact, I am not sure how this even works with
the currently defined secure association protocols in lower layers. None
of the documents have provided any clarity to me on this. The use of
IPsec/IKE for this is even more confusing to me - in this case, why
would PANA be used instead of EAP-in-IKEv2?
3. Statements like "...whereas it allows only limited type of traffic
(e.g., PANA, DHCP, router discovery, etc.) for the authorized PaCs" do
not exactly provide a clear picture of port control - yet another reason
IMO to leave the EAP lower layer where it belongs (at the link layer).
There are many other such confusions in my mind arising from the PANA
documents that I won't get into here.
4. The choices that are left to deployment while using PANA constitute a
long list for one protocol! Security before/after the PANA run, global
vs local PRPA, DHCP vs 3927 vs 2131 for IPv4 PRPA, dual-stack handling
behavior, secure association protocol choice and what it takes to make
it work with PANA, need for POPA and "unconfiguring" PRPA, PAA/EP
location and choice of protocol between those entities, IKEv1 vs IKEv2
etc. - I am sure I've missed more choices here; but, complex enough, to
say the least!
Given all of this, publishing these documents in the current stage as
proposed standard RFCs concerns me.
From: Sam Hartman [mailto:hartmans-ietf(_at_)mit(_dot_)edu]
Sent: Wednesday, May 24, 2006 8:12 AM
Subject: The Emperor Has No Clothes: Is PANA actually useful?
Hi. Speaking as an individual, I'd like to make an explicit
call for members of the IETF community not involved in the
PANA working group to review draft-ietf-pana-framework.
Please speak up if you have done such a review or attempted
such a review and been unsuccessful. Let us know what you
think PANA is intended to be useful for and whether you think
it is actually useful.
My strong hunch is that we've chartered work for some reason,
and now that the working group is nearing the end of its
charter, we still don't understand why we want this thing
we've built and whether it's a good idea. People aren't
screaming not so much because they are happy with results but
because no one actually understands PANA.
I understand that there's a strong presumption that once
chartered, work is useful. I'd like to challenge this
presumption enough to get people to actually read the
document. If people not involved in the effort sit down,
read the document and understand what it's all about, my
concern is satisfied. But if enough people try to read the
document, try to understand and fail, we're not done yet. We
certainly cannot have consensus to publish something we've
tried and failed to understand.
It's not just me. I've been trying to find people outside of
PANA who claim to understand the effort and what it's good
for and why link-layer solutions are not better. When the
first discussion of PANA hit the IESG, I asked other IESG
members why PANA was a good idea and what problem it solved.
"Don't go there," was the advice I got from the responsible AD.
At that time (a year and a half ago) there was no one on the
IESG who claimed to understand PANA or to think it was a good idea.
I'm fairly sure that with the possible exception of Jari (who
is a technical advisor to PANA), that's still true.
The security community has been trying to understand PANA.
I've sent multiple security reviewers at the PANA document.s
They always come back fundamentally confused about what PANA
is trying to do or about whether it is a good idea. They end
up focusing on some detail or another and asking for some
minor part of the system to be fixed. But I don't get the
impression from the reviews they understand the overall
picture; explicit discussion of this also indicates that they
are not confident in their understanding nor do they know
whether it is a good idea.
We keep running back over the same ground, still confused and
still trying to muddle through to no real effect.
I've tried to understand it myself. I tried to understand in the BOF.
It was very clear to me leaving the original PANA BOF that
something was very confused. Every year or so since I've
tried to go back and figure out what I missed. Eventually
though I've started wondering whether the problem wasn't me,
but was an actual lack of clarity.
So, folks can you please help us all out. Especially if the
internet area is not your primary focus, especially if you've
never heard of PANA before, take a look at the framework
document and all their other documents. Do you get it? Is
it a good idea?
Thanks for your time.
P.S. Again, this is me speaking as an individual. At this
late stage, it would be entirely inappropriate for me to take
actions as an AD claiming that we didn't understand a problem
without a strong community consensus.
Ietf mailing list
Ietf mailing list