Re: [gaia] Remote Peering Discussion at GAIA 117

[Kurtis Heimerl <[e-mail]>]

I don't know much about VRFs, do systems using it pass along AS hop info
normally?

To me, the concern is the long-haul lines obfuscating the L2 network and
making it hard to know whats "close". Clearly theres some utility but it
feels more like a hack to me to get these centralization and pricing
benefits that we could engineer in a better way. I dunno.

On Thu, Jul 27, 2023 at 2:16 PM Steven G. Huter <[e-mail]> wrote:

>
> On 7/27/23 1:54 PM, Kurtis Heimerl wrote:
>
> Great link! I was struck by this: " logic is the same; remote peering
> removes the colocation, equipment and circuit cost from the equation and
> replaces those costs with a metered transport alternative."
>
> I assume we can develop an extension to BGP that does that without
> obfuscating the important routing information, right?
>
> Again with the caveat of "we're sorta already doing this" that underlies
> so much of my confusion in this space.
>
>
> Hi Kurtis
>
> How do you tell the difference (in network terms) between an ISP in South
> Asia showing up at DE-CIX Frankfurt on their own (buying an IPLC), versus,
> an ISP connecting to DE-CIX Johor Baru in Malaysia and then paying DE-CIX
> for a VRF to Frankfurt?
>
> Both of these examples are remote peering, but two different
> implementations. For the former, the ISP knows what occurs with their
> 10Gbps of capacity. With the latter, how do you know the total number of
> others with whom you are sharing the capacity? What happens when the
> underlying capacity is oversubscribed and you don't know it?
>
> Why is peering in another continent a Good Thing when 80% of Internet
> traffic is now from the hyperscalers who are increasingly localized for
> their content delivery?
>
> Lots of other potential research questions.
>
> This is certainly an interesting public conversation to have and I'm glad
> it came up in the GAIA meeting.
>
> Steve
>