Dear Dr Peering,

I want to build a peering exchange in my local city; do you think that this is a good idea?

Hugs and Kisses

Hope Phule

Bedroom ISP LLP

--

Dear Hope,

Whether you build an exchange or not depends on a large number of variables, but before we go onto those topics have you ruled out the following two:

1. Join an existing exchange in your city, most Internet Exchanges have been around a few years and have built up both a considerable amount of experience and (probably more importantly) a wide range of existing customers. This would give a fast route to establishing peering as other networks will have already built to this location and have infrastructure in place.

2. Join an existing exchange but not in your city (or even country) - see ep.net for a list of IXes and the Ask DrPeering article “Value of an Internet Exchange” for a guide to calculating the value of various exchanges. Remember that you will need to change the relative variable to suit your network type and properties

If neither of these is cost effective then you have other options before hitting the DIY route:

3. Talk to an existing IX operator about extending their exchange to your local datacenter (as LINX [http://www.linx.net/] have recently done in Slough [ http://www.reuters.com/article/pressRelease/idUS166839+29-Jul-2008+BW20080729 ]). For this to be cost effective you will need to be able to provide a list of local networks that are willing to join the exchange to justify the cost of the extension.

4. Talk to an existing IX operator about running an exchange on your behalf (as AMS-IX [http://www.ams-ix.net/] do for CAR-IX[http://www.car-ix.net/]). Again you will need to get the business case for the build but you have removed the technical and majority of the political issues around the operation of the exchange in a single step. Depending on the location of the exchange you might be able to get local/regional/national/European government funding for the setup costs.

If neither of these works then it’s up to you as to what to do next:

5. Create a community run exchange owned and operated by the participants is a fairly straight forward option as all you really need is a switch and a couple of volunteers. An exchange of this type is unlikely to attract the larger networks until it reaches a critical mass but if most of your traffic is local to the local area its likely to be worth the investment of time and effort. This option might also be able to attract funding but that would require that the project is formalized into a non-profit making organisation with a proper company structure (this is how LINX and AMS-IX operate today)

6. Starting a IX business with a formal business plan and funding (either from a group of individual companies or private investment) is the hardest of these options but is still achievable (even in difficult financial times businesses that have a solid foundation will survive and grow). The key to this is researching the potential of your location before creating a business plan and getting networks to commit in principal BEFORE you start spending any money on hardware or facilities

7. The 'Ray Kinsella' [http://en.wikipedia.org/wiki/Field_of_Dreams] approach to building an Internet Exchange is not something that Dr Peering would recommend....

If you do decide that building an exchange is for you then have a look at [http://www.euro-ix.net/ixp/startingixp/] and make sure that you talk to members of the Internet community who are usually happy to provide helpful advice and information (frequently in exchange for beer, wine, tequila and/or scotch).

Dr Blessing

Dr Peering Clinic

--

James Blessing is an Independent Consultant providing network strategy and design services for ISP’s, Datacenters and CDNs via Garou Ltd [http://www.despres.co.uk/]