Re: [radext] Summary of Where we are with Fragmentation

It is unclear to me how much looking at this will help.  My understand
(possibly wrong) is that the EAP fragmentation issue has as much to do with
the MTU size on the network between the client and NAS as anything else.
This is going to be frequently smaller than the MTU size between the NAS and
the server and thus becomes a huge gating factor that has nothing to do with
internet congestion itself.  The choice of making EAP do the fragmentation
rather than having the NAS do it is a trade off in terms of where the
complexity lies (i.e. just with the people who talk EAP) rather than having
it occur on every protocol that is between clients and NAS boxes.

Jim

> -----Original Message-----
> From: radext-bounces@ietf.org [mailto:radext-bounces@ietf.org] On Behalf
Of
> Sam Hartman
> Sent: Monday, March 18, 2013 5:08 PM
> To: Peter Deacon
> Cc: radext@ietf.org; Alan DeKok
> Subject: Re: [radext] Summary of Where we are with Fragmentation
> 
> >>>>> "Peter" == Peter Deacon <peterd@iea-software.com> writes:
> 
>     Peter> To the extent comments about the properties of Alex's
>     Peter> fragmentation scheme are being discussed specifically the
>     Peter> following concerns expressed by Sam:
> 
>     Peter> "draft will require significant transport review because they
>     Peter> both propose protocols with complex
>     Peter> MTU/congestion/fragmentation/performance characteristics."
> 
>     Peter> "Alex's draft and the MTU issue, I stand behind my claim that
>     Peter> it will tend to involve fragmentation at multiple layers.
>     Peter> There's no way to learn the MTU that is currently being
>     Peter> proposed."
> 
>     Peter> My observation or question is could we look toward current
>     Peter> widespread deployment of RADIUS/EAP to help understand these
>     Peter> concerns with respect to Alex's fragmentation scheme only?
> 
> Yes.
> I think that will help us.
> _______________________________________________
> radext mailing list
> radext@ietf.org
> https://www.ietf.org/mailman/listinfo/radext