Re: Review request for "Two Dimensional IP Routing Architecture"

[Fred Baker <[e-mail]>]

On Mar 10, 2012, at 3:25 AM, 杨术 wrote:

> Hi, Fred Baker, 
> 
> Thanks for your advices.
> 
> We have discussed your draft in our group meeting, because your 
> draft is strongly related with two dimensional routing. Your draft 
> illustrates detaiedly that how to devise a new protocol, that can make 
> forwarding decisions based on destination and source addresses. 

It also explicitly mentions DSCP as an option, and in homenet discussion (might have been offline, I don't remember), Brian Carpenter suggested flow label (although I think that would differ from the random number his current document puts into it). Yes, it has overlaps.

> Our draft differs from yours in that:
> 1. We try to illustrate the huge benefits from deploying two dimensional 
> routing, that makes forwarding decisions based on both destination and 
> source addresses. The network will be more flexible if routers can divert
>  traffic based on source address. Thus, policy routing, traffic engineering,
>  path/link protection, multi-path, multi-homing can be achieved more easily.
>  For example, with two dimensional routing, we can express "deliver traffic 
> from source A towards destination B to router C" explicitly and easily. 

> 2. We focus on the architecture of two dimensional routing. We try to properly 
> divide the whole routing system into several components, and point out how 
> to devise each component to achieve efficiency and consistency.
> 
> 3. We have designed a new forwarding table structure called FIST, and we are
>  developing it based on a commercial router. With one more address to lookup
>  during routing, we believe that the FIB is a key component considering scalability
>  issues. FIST is different with previous FIB structure in that, 1) it has two TCAMs,
>  one stores destination prefixes, the other stores source prefixes; 2) in SRAM,
>  there is a two dimensional table that stores the next hop information. Such that
>  we can achieve fast lookup speed, and avoid explosion problem in TCAM.

That's one implementation approach, and it sounds like it has value. While most routers have some concept of a forwarding information base, the IETF has never particularly commented on how the FIB was structured. That has been viewed as a competitive angle - different implementations might do it different ways with different effects. The last discussion I recall about FIB design was a BOF at IETF 14 or 15 led by Craig Partridge.

When I started working to regularize this, folks back at *my* ranch told me that I was re-inventing Multi-Topology Routing. 

> Shu Yang
>   
> 
> 
> On Fri, Mar 9, 2012 at 10:33 PM, Fred Baker <[e-mail]> wrote:
> 
> On Mar 8, 2012, at 12:02 AM, 杨术 wrote:
> 
>> Dear all,
>>  
>> We are looking for your comments on the new draft "Two Dimensional IP Routing Architecture".
>>  
>> 	This document describes Two Dimensional IP (TwoD-IP) routing, a new
>> 	Internet routing architecture which makes forwarding decisions based
>> 	on both source address and destination address. This presents a
>> 	fundamental extension from the current Internet, which makes
>> 	forwarding decisions based on the destination address, and provides
>> 	shortest single-path routing towards destination. Such extension
>> 	provides rooms to solve fundamental problems of the past and foster
>> 	great innovations in the future.
>> 	We present the TwoD-IP routing framework and its two underpinning
>> 	schemes. The first is a new hardware-based forwarding table
>> 	structure for TwoD-IP, FIST, which achieves line-speed lookup with
>> 	acceptable storage space. The second is a policy routing protocol
>> 	that flexibly diverts traffic.
>>  
>> 	We plan to give a presentation on this in the upcoming IETF83. The
>> 	draft can be found at http://tools.ietf.org/html/draft-xu-rtgwg-twod-ip-routing-00.
>>  
>> 	We would really appreciate any comments and questions about the document.
> 
> My first suggestion would be to compare/contrast with 
>     http://tools.ietf.org/html/draft-baker-fun-routing-class
>