[NNagain] The Whys of the Wichita IXP Project

[Bill Woodcock]

On Wed, Feb 21, 2024 at 8:02 PM Brent Legg wrote:
> Saying an IXP can be built for $8k is enormously confusing to many policymakers

Only if someone tries to confuse them by referring to a datacenter as an IXP.  So, please don’t do that.  $8k is a reasonably-up-to-date global average cost for new IXP formation.

> Does it need to be a facility that networks can rely on to remain “up” in the wake of adverse events?  Yes.

Why?

> Resilient from power outages?  Yes.

Why?

> Resilient from cooling equipment failures? Yes.

Why?

> Resistant to wind damage?  Yes.

Why?

> Is “best effort” good enough?  No.

Why?  And what do you think is better than “best?"

> Then does it need to be professionally managed?  Yes.

Why?  And this is really a big one.  I know why Hunter needs it to be managed by someone else to fit his model, but how would that advantage anyone in Wichita?  It’s perfectly ok for different people to have fill different niches and have different business models.  Hunter likes neutral datacenters with a lot of interconnection, and I do too.  Having an IXP in such a datacenter is enormously advantageous to the datacenter, and makes its financial outlook much better.  Hunter is trying to move fast and cover a lot of ground.  Which is great.  Solving problems at scale is great.  But he’s solving a datacenter problem, not an IXP problem.  The IXPs are simply a way of making it more likely that the datacenters will thrive.  Which is great.  For the datacenters.  But if you’re trying to drop a hundred tiny datacenters off the back of trucks, and you’re moving fast, and you want an IXP in each one as soon as possible, doing the four to eight months of work typically necessary to organically organize an IXP in each location simply doesn’t scale.  So, outsourcing this to something like DE-CIX makes sense for Hunter.  But it doesn’t particularly make sense for DE-CIX, and it doesn’t particularly make sense for Wichita, or any other specific community of network operators.  Will it work?  In some places, sure.  Roll the dice enough times and you’ll win some of the time.  But being one of many bets, some of which will fail, isn’t particularly reassuring to any specific community.  As long as Hunter doesn’t make an _exclusive_ agreement with DE-CIX, I don’t see this as particularly problematic.  But I’m not sure you appreciate what a bad thing “professional management” is for APBDC.  Just look at Manchester.

> Where should it be built?  Where a concentration of eyeball traffic already exists that can grow a peering ecosystem faster than it might otherwise, and that is also proximate to existing fiber plant, and where diverse manholes can be placed on the edge of public right-of-way.
> In the case of Wichita, that’s at Wichita State University.

That’s one possible location, and again, it’s optimal for Hunter’s specific datacenter construction model, but it’s not optimal for an IXP site.

> Creating a secure, neutral, resilient interconnection facility with proper cooling, power systems, lockable cabinet space, diverse manholes and POE isn’t cheap.  The whole project is actually more than the $5M grant we received.  We’re putting in over $800k in cash, plus additional in-kind match.

No argument there, and it seems quite reasonable to build this sort of very-small purpose-built datacenter in places that don’t already have one.  And I have no problem with the use of public funds to make it happen.  I do have a problem with people mis-labeling it as an IXP, because that, as you so rightly pointed out, confuses policy-makers.  Who might mistakenly think that by funding the construction of datacenters, they’d helped the IXP situation somehow.  There’s no problem with helping datacenters, but datacenters and IXPs are radically different things, with different and not-aligned purposes and utterly unrelated business models.  Confusing policy-makers and leading them to think that IXPs can be helped by throwing money at them causes a lot of problems.  Which I wind up having to straighten out, so it makes work for me, and my time can be better spent on productive things.

So, good on you for getting more small datacenters into third-tier markets!  But please stop confusing policy-makers by calling them IXPs, or implying that they have something to do with IXPs.  There’s a possibility that some of them will be located in places that aren’t sub-optimal for IXPs, and that would be nice, but those would be generally be sub-optimal for datacenters, so I don’t recommend that you try to do that.

                                -Bill

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: Message signed with OpenPGP
URL: <https://lists.bufferbloat.net/pipermail/nnagain/attachments/20240223/de528890/attachment.sig>