On 27/03/2014 13:20, "Will Hargrave" <[email protected]> wrote:
>
>Actually, it doesn¹t - as evidenced by the massive reseller growth at
>AMS-IX of smaller ASNs (they¹re up to 659 connected now) since
>introduction of the reseller programme.

They have done (as have many others) a great job to make it easy. I
believe though that its entirely possible using a traditional model and
generates a better overall return case.

> 
>
>I share many of your concerns - the 30,000ft view - about shared fates
>and the variety of other problems partner programmes and l2
>interconnections bring; but I think, having spent a decade running
>networks at large telcos, you don¹t appreciate the market other networks
>operate in. This model works very well for hundreds of networks. Today's
>peering networks are more diverse than ever.

I don¹t think I said that the model isn¹t working Will, I said its not a
good model from a design perspective or at least that¹s what I meant and
that its not the only cost effective solution. I also think it has
drawbacks as does any model. But if I may throw your assertion back at you
(:)) you don¹t appreciate the view that one gets given the knowledge of
unit costs of infrastructure and systems to support the type of
infrastructure required to build what I¹m positioning. You also don¹t
appreciate that in many cases and in the many large telcos I¹ve worked in
- we are often, if not always, competing with those types of companies,
notably in the small business sector and wholesale sectors - but do I take
seriously the cost of networks, absolutely I do, I also take massive note
of what I call ³design debt² where you make a choice that costs you a lot
to live with later on.

But that¹s the great thing about UKNOF, it allows us to share our views
and hopefully learn something from each other.

Regards,
Neil.


Reply via email to