[ppml] Provider Independence???

[marcelo bagnulo braun]

Hi Michael,

I guess that the problem that Owen is pointing out is that when there 
are multiple ISP providing service to users within the same 
geographical region, all these providers will have to announce the 
aggregate address block that corresponds to the complete region. Since 
all the isps are announcing the whole geo block, this means that they 
will receive packets for all the users within the region, some of which 
won't be their direct costumers. At the end of the day, this means that 
ISPA will be carrying traffic for some of the customers of ISPB, which 
i guess lacks of a business case.

Regards, marcelo

El 08/12/2004, a las 13:17, Michael.Dillon at radianz.com escribió:

>> The fundamental flaw with the theory of geographical aggregation is 
>> that
>> it would need a form of provider transit love-fest to become feasible
> that
>> simply isn't the reality of today's market conditions.
>
> That's a big mouthful.
>
> Love-fest is not a network technical term so I have
> no idea what you are driving at. Clearly, providers
> do currently interconnect through private peering and
> through exchange points. There is no single model for
> how these interconnects are organized and managed. I see
> no reason to assume that all providers would reject
> routing addresses that require them to have peering
> connections with their local competitors since many
> of them already have such peering connections to reduce
> their costs.
>
> Secondly, the reality of today's market conditions is
> not equivalent in all regions of North America or in all
> regions of the world. One could reasonably assume that at
> any given point of time, some region somewhere is in
> recession. This is not a good reason to not do geographic
> addressing.
>
>> Today, you would
>> need to get all the major providers in every region to agree to 
>> provide
>> some form of transit for lots of people that aren't paying them to 
>> talk
>> to lots of other people that aren't paying them.  I just don't see 
>> that
>> as being likely to happen.
>
> I have no idea what you are talking about. This sounds like
> you are taking the position of the big five networks circa 1994
> who were trying to prevent their customers from selling connectivity
> to downstream ISPs. The market has worked out this problem.
> We have lots of paths with greater than 2 ASNs in them.
> Geographic addressing doesn't change this in any way.
>
> --Michael Dillon
>