[e2e] architecturally speaking

[David P. Reed]

I would love to understand more the linkage between "delay tolerant 
internetworking"  and "real-time collaboraion focused 
internetworking".   Need they be conceptually different?

We're working on scalable wireless internetworking technologies that look 
like they can achieve speed-of-light message latencies (because they are 
very wideband) over a large number of hops.  This is useful for those of us 
who want to create, for example, distributed virtual multiuser 
environments, where low latency is crucial.  We tolerate delay by ignoring 
excessively delayed information, assuming instead that the resource 
(spectrum) can always be scaled proportional to demand (as measured by 
money users are willing to spend), so it doesn't approach congestive failure.

Delay tolerance is mostly an application problem (as becomes obvious when 
you consider the true speed-of-light issues).   It only becomes a 
networking problem when your goal is to underprovision a network compared 
to demand.

At 10:23 PM 3/28/2003 +0000, Jon Crowcroft wrote:
>In missive <200303282047.h2SKlTSs023198 at dwight.CS.Berkeley.EDU>, Kevin 
>Fall typ
>ed:
>
>  >>to what sort of environments it is targeted for and the justification for
>  >>some of the decision(s).  In particular:
>
>  >>     .. this paper concentrates on naming and addressing issues for
>  >>     establishing connectivity between radically heterogeneous networks,
>  >>     a problem that the Internet Protocol only partially solved.  In 
> this paper,
>  >>     we are literally concerned only with 'inter-networking,' and not 
> with any
>  >>     of the many other networking issues such as the units, timeliness or
>  >>     guarantee of resource allocation, security or auditing.  A concrete
>  >>     realization of our framework must address these issues, but 
> within the contexts
>  >>     of the particular networks being connected:  we do not believe it is
>  >>     sensible to address them through a single unifying overlay 
> network protocol.
>
>
>gosh - thats a quick read - we spent a while reading the dtnrg stuff -
>i suggest people read before writing:-)
>
>plutarch is not about naming and addressing much - if you want a good
>solutuon to that problem i would refer you to paul francis work
>(either his thesis or more recent work or both) - it is more about
>what it says so please read it:-)
>
>the DTNRG stuff , which we read is about disconnection.
>
>  >>Given that DTNRG is somehow focused precisely on an overlay network 
> protocol
>  >>approach (but with a particular service definition of non-interactive 
> delivery in the
>  >>worst cases), I'm particularly interested in why this is apparently 
> rejected by the
>  >>authors.  Further, in environments where long periods of disconnection 
> and/or high
>  >>delays may be encountered, my hunch is that some of the mechanisms 
> suggested
>  >>for supporting dynamic mappings within contexts (using soft-state, 
> using some form of
>  >>request/response name binding) might be difficult.
>  >>
>
>overlays are very f ine - overlays for disconnection have to do state
>(context) management very well - we skate aroud that coz we only need
>to do it "on the order" of as well as the internet or so...
>
>
>good luck - your parsec-age may vary...
>
>btwe,m i really like the DTNRG stuff , but it is orthogonal
>  >>
>  >>>
>  >>> From:  Lloyd Wood <l.wood at eim.surrey.ac.uk>
>  >>> To:    Jon Crowcroft <Jon.Crowcroft at cl.cam.ac.uk>
>  >>> cc:    end2end-interest at postel.org
>  >>> Subject: Re: [e2e] architecturally speaking
>  >>> Date:  Thu, 27 Mar 2003 19:18:32 GMT
>  >>>
>  >>> On Wed, 2003-03-26 at 15:09, Jon Crowcroft wrote:
>  >>> > fyi, we just sent this you know where -
>  >>>
>  >>> where? Are we supposed to read minds, or what?
>  >>> lamentably out-of-date/incomplete list on:
>  >>> http://www.cl.cam.ac.uk/%7Ejac22/out/
>  >>> no help. Still, at least the paper's online. Asking for accurate
>  >>> metadata/context as well possibly too much, really; fortunately,
>  >>> academics have journals simply to organise their output meaningfully.
>  >>>
>  >>> > http://www.cl.cam.ac.uk/~jac22/out/plutarch.pdf
>  >>>
>  >>> Plutarch: An Argument for Network Pluralism
>  >>> Jon Crowcroft, Steven Hand, Richard Mortier, Timothy Roscoe, Andrew
>  >>> Warfield, 24 March 2003.
>  >>>
>  >>> Hmmm, heterogenity, late-binding of names and contexts; a very similar
>  >>> approach to the Interplanetary Internet 'bundle' idea...
>  >>>
>  >>> ..which is currently being expressed in some recently-submitted
>  >>> delay-tolerant networking drafts:
>  >>> http://www.ietf.org/internet-drafts/draft-irtf-dtnrg-arch-00.txt
>  >>> 
> http://www.ietf.org/internet-drafts/draft-irtf-dtnrg-ipn-bundle-xfer-00.txt
>  >>> http://www.ietf.org/internet-drafts/draft-irtf-dtnrg-bundle-spec-00.txt
>  >>
>  >>>
>  >>> Bundling protocol software shortly available from:
>  >>> http://www.dtnrg.org/
>  >>> and very possibly by this weekend.
>  >>>
>  >>> L.
>  >>>
>  >>> <http://www.ee.surrey.ac.uk/Personal/L.Wood/><L.Wood at ee.surrey.ac.uk>
>  >>>
>
>  cheers
>
>    jon