Re: [NNagain] The Whys of the Wichita IXP Project

[Bill Woodcock <woody@pch.net>]

[-- Attachment #1: Type: text/plain, Size: 1496 bytes --]



> On Feb 22, 2024, at 19:58, rjmcmahon via Nnagain <nnagain@lists.bufferbloat.net> wrote:
> Boston University spent $305M on this and it doesn't have an IXP.
> https://www.bu.edu/articles/2022/center-for-computing-and-data-sciences-photo-essay/
> It's like building a magnificent train station w/o any tracks to/fro the station.

Most datacenters don’t contain IXPs, and most IXPs aren’t located in datacenters.  It’s very financially advantageous for a neutral multi-tenant datacenter to contain an IXP, but generally much less advantageous for an IXP to be located in a datacenter.  Datacenters tend to concentrate content, but that content can be transported to an IXP over just a few strands of fiber.  Whereas eyeballs have to be physically aggregated, and that’s over thousands of strands, so the average distance to eyeballs matters, whereas the average distance to content just doesn’t have a significant multiplier on it, and the content is portable anyway.  The optimum location for an IXP is in a city center, whereas the optimum location for a datacenter (all political, zoning, and real-estate factors considered) is typically in an industrial park well outside the city core.

Lots of organizations need a datacenter for their own use, and universities are typical in that.  It doesn’t mean that they’d make sense as locations for an IXP, unless they’re also aggregating a lot of eyeball fiber for some reason.

                                -Bill


[-- Attachment #2: Message signed with OpenPGP --]
[-- Type: application/pgp-signature, Size: 833 bytes --]