Re: [dtn-interest] DTN BoF Proposal for IETF90

> If I recall correctly though there was something
> about your http-dtn stuff that I thought just didn't work for
> some use-cases, but I've not re-read it today, so I think your
> stuff might or might not be a viable starting point when looked
> at in detail

As always, good to get a detailed technical critique from you, Stephen.

Lloyd Wood
http://about.me/lloydwood
________________________________________
From: Stephen Farrell [stephen.farrell@cs.tcd.ie]
Sent: 27 April 2014 18:21
To: Wood L  Dr (Electronic Eng); jo@netlab.tkk.fi; Fred.L.Templin@boeing.com; dtn-interest@irtf.org
Subject: Re: [dtn-interest] DTN BoF Proposal for IETF90

On 27/04/14 14:40, l.wood@surrey.ac.uk wrote:
> if you're  enabling a web server for DTN,
> why use the bundle protocol?
>
> http://sat-net.com/L.Wood/dtn/http-dtn.html

Yes, starting by looking at approaches like that (though maybe
considering HTTP/2.0 now as well) is IMO well worth considering
if we want an eventual IETF PS RFC for DTN to be (able to be)
widely used. If I recall correctly though there was something
about your http-dtn stuff that I thought just didn't work for
some use-cases, but I've not re-read it today, so I think your
stuff might or might not be a viable starting point when looked
at in detail, But its defo something to look at I'd say if
we're really prioritising terrestrial deployment.

I think you and I Lloyd might agree on the above but disagree
about what to call the result - I'd like that to be done as
a revision of the BP so as to maintain easy gatewaying with
5050 implementations. I suspect you might not be so keen on
that:-) But I think that compatibility point ought also be
up for discussion if an IETF WG is being formed.

All that said, for an IETF WG, it'd matter a lot whether or not
there's a set of interested enough parties and I could imagine
that some people would like taking this kind of approach whereas
others might hate it and want far fewer changes compared to
5050, so I think that means that resolving the high level
direction for a 5050bis is something that should be done before
forming a WG, so a WG doesn't stall on not being able to get
consensus on the high level approach. (And that in turn means
that more folks weighing in on this topic could be very
helpful.)

Cheers,
S.

>
> Lloyd Wood
> http://about.me/lloydwood
> ________________________________________
> From: dtn-interest [dtn-interest-bounces@irtf.org] On Behalf Of Stephen Farrell [stephen.farrell@cs.tcd.ie]
> Sent: 27 April 2014 14:00
> To: Joerg Ott; Templin, Fred L; dtn-interest@irtf.org
> Subject: Re: [dtn-interest] DTN BoF Proposal for IETF90
>
> Hiya,
>
> On 27/04/14 09:03, Joerg Ott wrote:
>> While I see Stephen's point on the tooling, I don't agree with
>> this since tool fashion changes (including the praise for text-based
>> encoding in the IETF at some layers).  We have built commercial
>> products for terrestrial use based upon RFC 5050 and, really,
>> the tool part isn't the problem.  The security part is where
>> we want to be more sensible towards protocol re-use.
>
> Fair point. OTOH, I think its a real fail that one can't
> e.g. just install an Apache web server module to DTN enable
> your web proxy or something similar. Having to build an
> entirely new overlay for 5050 that doesn't re-use anything
> already deployed makes it much less likely to be used IMO.
> Actually, that's not just opinion, its based on our
> experience trying to deploy stuff we've piloted that worked
> but always seems to be too much trouble for non-subsidised
> deployments.
>
> That is not something that will change just because the
> RFC is in the IETF stream. But I think we should be able to
> fix that with a 5050bis while also keeping the ability to
> gateway to 5050 DTNs.
>
> To be clear, I don't want that we aim to DTN-enable the web, I
> want us to consider terrestrial deployability and making that
> as easy as possible as first class requirements. If the means
> to get that were not related to HTTP (though HTTP/2.0 timing
> might just still be fortuitous here), but say related to
> some OpenWRT build for homenet, then that'd also be just
> dandy.
>
> S.
>
>
> _______________________________________________
> dtn-interest mailing list
> dtn-interest@irtf.org
> https://www.irtf.org/mailman/listinfo/dtn-interest
>