Re: [EXTERNAL] Embedding IP information in an IPv6 address (OMNI)

[Bob Hinden <[e-mail]>]

Fred,

> On Oct 15, 2020, at 4:08 PM, Templin (US), Fred L <[e-mail]> wrote:
> 
>>> 
>> 
>> I think this is a bad idea, that is trying to make a set of networks with a whole range of characteristics (i.e., speed, delay, drop rates,
>> errors, cost, etc.) look like a single "big bridged campus LAN”.   Most people have stopped buildings “big bridged campus LANs” a long
>> while ago, what your are proposing is much harder given the different characteristics of the networks you mention.
> 
> I think you may have gotten a polar opposite read from what the OMNI technology
> is all about - OMNI is about diversity; not homogeneity. The OMNI "multilink" concept
> in particular is about accommodating all manners of data links having diverse properties
> and characteristics, while presenting the best possible performance profiles to mobile
> nodes in the face of dynamically changing conditions. The "bridged campus LAN"
> concept is not a physical thing; it is a virtual one that allows nodes to communicate as
> single-hop neighbors at the IP layer while their underlying data link properties may be
> varying dynamically. This discussion now starts to enter into the realm of what we have
> been debating in the ICAO working groups for the past several years, and has motivated
> the design based on expert input from the aviation community. And, what is good for
> aviation also turns out to be good for the worldwide mobile Internet for any manners
> of mobile nodes.

I think I understand what you are trying to do.  I just don’t think this will work, some problems are best not solved at L3.   As I outlined below, I think the transport layer (and probably application layer) will also need to be involved.

Bob


> 
> Fred
> 
>> I think a better approach is to treat the networks separately and use transport protocols techniques to use them efficiently separately
>> and concurrently.   Likely the transport protocols and applications above them will need to be aware of the underlying networks.
>> 
>>> 
>>>>> About changes to IPv6, all that has been asked so far is for the OMNI
>>>>> interface to define its own link-local addresses - having that, none of the IPv6
>>>>> standards like RFC4861, RFC8200 etc. are changed.
>>>> 
>>>> The OMNI draft updated RFC1191, RFC3879, RFC4291, RFC4443, and RFC8201.
>>> 
>>> When I said "all that has been asked for *so far*", I meant in terms of what has
>>> been discussed on the list up to now. Up to now, the only aspects of OMNI we
>>> have touched on are the LLAs and SLAs, and it is only RFC3879 and RFC4291 that
>>> are in question. So, until we get around to discussing the OAL let's put the other
>>> RFCs aside for now.
>> 
>> I disagree, you are clearly proposing a lot of changes, discussing them separately doesn’t change that.
>> 
>> Bob
>> 
>> 
>