[e2e] Discrete IP - retake

Jon Crowcroft Jon.Crowcroft at cl.cam.ac.uk
Tue Sep 18 03:30:41 PDT 2012


as i said, read the paper we published in sigcomm future network
architectures nearly 10 years ago
http://www.cs.ubc.ca/~andy/papers/plutarch-fdna.pdf

no one in the IETF says you can't run multiple versions of IP 
and build interworking points that copy payloads , if you so wish -
actually, there's a lot of this going on in middleboxes one way and another
already

you need to propose how you find the right place to do the translation of
headers - this requires some sort of overlay control plane and might ential
the use of a new overlay meta-addressing system or make  use of name
spaces as in IPNL or related work on I^3 
http://128.232.0.20/teaching/0910/R02/papers/ipnl.pdf
http://www.cs.rice.edu/Conferences/IPTPS02/166.pdf


many of the ways IPv4/IPv6 interworking have also tackled this, not just
using tunnels or tunnel brokers, but lots of other techniques 

there are LOTS and LOTS of papers in the future internet research
programmes around the world on how to do this ad solve other practical
problems - see work in the IETF on ILNP for example
http://ilnp.cs.st-andrews.ac.uk/

e2e is just one discussion place - the future internet research programmes
have moved a long way beyond the necessary and insufficienct business of
deploying IPv6 as fast and as widely as possible, and on to tackling a
bunch of new problems (Information centric networking, for example, or
massive scale internet of things and sensors etc etc)

communities of interest for that include conferences such as ACM Sigcomm
and Usenix NSDI and IEEE Infocom and many others...

ideas like XIA (see recent FIA report
http://www.nets-fia.net/Meetings/May11/May%202011%20meeting%20report%203-1.pdf
amongst oterhs) go way beyond identifiers for end points and have entire
DAGs coded in packets (efficiently)

there's so much exciting new stuff out there....


on the other hand, practical barriers to deploying lots of different stuff
exist, not just in the slowness/ossified internet core IPv4 routers, but in
the many weird boxes nearer the edges - see the Trilogy project
folks' paper on
How Hard Can It Be? Designing and Implementing a Deployable Multipath TCP
at
https://www.usenix.org/conference/nsdi12/tech-schedule/technical-sessions

and also in same conference session, the cunning tricks the Yale folks had
to empoy to get small changes into TCP:
Fitting Square Pegs Through Round Pipes: Unordered Delivery Wire-Compatible
with TCP and TLS 

In missive <CACQuieYAU+O1bXYdM+ZJsknXE=8wPgzftOxKJ73Mshxu2Dtc6A at mail.gmail.com>, Par
s Mutaf typed:

 >>--047d7b4140c626654d04c9f67a1c
 >>Content-Type: text/plain; charset=ISO-8859-1
 >>
 >>Off-list messages that I received indicate to me that I was not taken
 >>seriously
 >>because I am too modest. I was called a troll and hobbyist. I have in fact
 >>a PhD in
 >>computer science from INRIA, France. I am now an asst prof. in Turkey.
 >>But I let go my PhD title because my thinking evolved beyond your
 >>imagination ;-).
 >>Check and support my project (off-list please this is off-topic):
 >>
 >>http://www.content-based-science.org/
 >>
 >>What is important is the content. Not the name.
 >>
 >>Now back to our topic. IPng was clearly designed the wrong way. Now read
 >>the
 >>message again, this is the result of 15 years thinking:
 >>-----
 >>
 >>Dear colleagues,
 >>
 >>I believe that the next step in IP's evolution would not be IPv6. It would
 >>be "Discrete IP" allowing any IP version.
 >>I concluded that Discrete IP better respects the end-to-end principles
 >>therefore it is economically more viable.
 >>
 >>***Do not touch the existing Internet, do not assume that IPv6 is the end of
 >>centuries of research.***
 >>
 >>-I propose that we do not touch the core Internet, i.e. enforce the
 >>modification of all Internet routers, this is what IPv6 does.
 >>-People should be free to choose the IP version that they wish because
 >>deciding for others is a technology blocker. IETF designs IPv6, IETF blocks
 >>its development. Because IETF does not give freedom of choice. This is not
 >>normal. Some entities may use IPv6 others IPv4 yet others IPv7 for unknown
 >>reasons. Everybody may agree on IPv6, or not. We do not know. We do not
 >>have to.
 >>-To give such freedom of choice, we need to change the end-nodes, for
 >>example TCP.
 >>-This is the end-to-end principle.
 >>
 >>Here is a picture (in this picture we have a network of Internets running
 >>random IP versions):
 >>http://htmlimg4.scribdassets.com/3798kx3chs1szfhj/images/4-ce35c39dd1.jpg
 >>The question is:
 >>***Would this be the ideal for the Internet? Please discuss this question
 >>without entering in design challenges.***
 >>
 >>For more information, see my unpublished paper:
 >>
 >>http://www.scribd.com/doc/105448105/Discrete-IP
 >>
 >>Cheers,
 >>Pars
 >>
 >>-- 
 >>http://www.content-based-science.org
 >>
 >>--047d7b4140c626654d04c9f67a1c
 >>Content-Type: text/html; charset=ISO-8859-1
 >>Content-Transfer-Encoding: quoted-printable
 >>
 >>Off-list messages that I received indicate to me that I was not taken serio=
 >>usly<br>because I am too modest. I was called a troll and hobbyist. I have =
 >>in fact a PhD in <br>computer science from INRIA, France. I am now an asst =
 >>prof. in Turkey. <br>
 >>But I let go my PhD title because my thinking evolved beyond your imaginati=
 >>on ;-). <br>Check and support my project (off-list please this is off-topic=
 >>):<br><br><a href=3D"http://www.content-based-science.org/">http://www.cont=
 >>ent-based-science.org/</a><br>
 >><br>What is important is the content. Not the name.<br><br>Now back to our =
 >>topic. IPng was clearly designed the wrong way. Now read the <br>message ag=
 >>ain, this is the result of 15 years thinking:<br>-----<br><br>Dear colleagu=
 >>es,<br>
 >><br>I believe that the next step in IP&#39;s evolution would not be IPv6. I=
 >>t would<br>be &quot;Discrete IP&quot; allowing any IP version.<br>I conclud=
 >>ed that Discrete IP better respects the end-to-end principles<br>therefore =
 >>it is economically more viable.<br>
 >><br>***Do not touch the existing Internet, do not assume that IPv6 is the e=
 >>nd of<br>centuries of research.***<br><br>-I propose that we do not touch t=
 >>he core Internet, i.e. enforce the<br>modification of all Internet routers,=
 >> this is what IPv6 does.<br>
 >>-People should be free to choose the IP version that they wish because<br>d=
 >>eciding for others is a technology blocker. IETF designs IPv6, IETF blocks<=
 >>br>its development. Because IETF does not give freedom of choice. This is n=
 >>ot<br>
 >>normal. Some entities may use IPv6 others IPv4 yet others IPv7 for unknown<=
 >>br>reasons. Everybody may agree on IPv6, or not. We do not know. We do not<=
 >>br>have to.<br>-To give such freedom of choice, we need to change the end-n=
 >>odes, for<br>
 >>example TCP.<br>-This is the end-to-end principle.<br><br>Here is a picture=
 >> (in this picture we have a network of Internets running<br>random IP versi=
 >>ons):<br><a href=3D"http://htmlimg4.scribdassets.com/3798kx3chs1szfhj/image=
 >>s/4-ce35c39dd1.jpg">http://htmlimg4.scribdassets.com/3798kx3chs1szfhj/image=
 >>s/4-ce35c39dd1.jpg</a><br>
 >>The question is:<br>***Would this be the ideal for the Internet? Please dis=
 >>cuss this question<br>without entering in design challenges.***<br><br>For =
 >>more information, see my unpublished paper:<br><br><a href=3D"http://www.sc=
 >>ribd.com/doc/105448105/Discrete-IP">http://www.scribd.com/doc/105448105/Dis=
 >>crete-IP</a><br>
 >><br>Cheers,<br>Pars<br clear=3D"all"><br>-- <br><a href=3D"http://www.conte=
 >>nt-based-science.org" target=3D"_blank">http://www.content-based-science.or=
 >>g</a><br><br>
 >>
 >>--047d7b4140c626654d04c9f67a1c--

 cheers

   jon



More information about the end2end-interest mailing list