Word came down from top brass that peering is pretty much finished in North America. What is the G2 on peering in Canada for example?
Anthony Peter Coleman
I gave a talk at the IXP Symposium in Toronto earlier this month, sponsored by the Ontario government. The presentations and discussions reminded me of the European IXP and ISP forums - everyone shared very openly what was working and what was not working. There is still some heavy lifting to do in some markets here.
Why is peering important for Canada?
The price of transit tends to drop for the localities around successful IXPs. Imagine a topology map with concentric rings indicating altitude. The center of the Transit Topology map would be the populated colocation centers where the price of transit is cheapest. As a generality, just like competition leads to lower prices, the further away you get from popular IXPs, the higher the price of transit gets.
The reason for this is intuitive. One can purchase transport to the network dense colocation center to pick up cheap transit from an open market, the price in the original region can be provided at a lower cost. Generally you will find prices 30% lower within a network dense colocation center than when compared with outside the building. Therefore others can do the same and transport those price points back to a nearby market. Then the prices drop proportional to the cost of the transport to the nearest network dense IXP.
One participant was singled out Hurricane Electric as coming in and providing wonderfully disruptive pricing for Canadians, much better than half the market price.
Is peering the answer?
Is focusing on building a broader open market of transit providers the right prescription for Canadian colocation providers and IXPs? Is there sufficient peering population in the various provinces to justify an IXP in each? Perhaps colocation and an open transit market is sufficient?
For much of Canada outside of Toronto the peering populations are small by global standards. The addressable market is small, which means introducing another low cost provider into the market means more players are fighting over a small market. It is not clear there is sufficient population to justify establishing an open transit market within colocation centers there. We will see.
The other option of course involves focusing on growing the peering population across Canada by increasing the gravitational pull into the colo centers with the three services:
1) open market for transit – lure a couple more providers from the U.S. to the colo. Get gov’t to establish presences in colo, and award grants that require delivery of services into these colos.
2) peering away local traffic locally will grow as a natural side effect, but you will need a peering liaison for each colo. This full-time person will evangalize peering across Canada and specifically into their colo and IXPs.
tethering from the colocation sites across Canada to the other exchange points. This increases the reachable peer count at the colos. Point-to-point circuits into the IXP are expensive, but colos could turn to turn key remote peering providers (such as IIX, my employer) for one pipe delivering multiple VLANs to many IXPs. This makes the colocation center a regional peering point, not just local, and should drive more participation.
Peering population growth can also happen organically given enough time, care, and focus. But I like these three things to accelerate the path to make the colocation centers (the ones with IXPs hopefully) across Canada the place to get directly connected to higher performance lower cost Internet Transit and Peering products. Then, the population will grow, the peering population will grow, and the breadth of services provided in other parts of the world will work juts as well in Canada.
PS - If you ever get a chance to participate in one of these Canadian peering events, go for it. The food was great, the people from government, R&E, IXPs and ISPs were wonderful, the discussions productive, positive, and forward looking. We toured the TORIX and saw both the Equinix and Cologix data centers there at 151 Front. The hard part isn’t the technology, it is the market, and if the government wants to play a guiding hand, it will have to roll up its sleeves and get a little bit deeper into and more involved directly with the Internet Peering Community across Canada.