The changing IP interconnection scene, observations.

What I am hoping to address in this post is the changing interconnection market, I live and work on the middle of this market in what is arguably one of the main IP interconnection markets in the world – Amsterdam.

First of all I am by no means a technical expert on interconnection, more of an interested observer. My views have been built up over the past 5 years watching the market and being involved from a commercial point of view. I am sure some of the technicians reading this will shoot it down in flames but that’s more than welcome.

Over the past years I was involved in the peering market as I was working at AMS-IX, had daily conversations about the benefits of peering on a public exchange (as opposed to private peering off an IX) but that was almost 3 years ago now, I was wondering if the picture has changed. The market is very dynamic and 3 years ago it was more of a simply discussion around peering v transit but times have changed and the market is unrecognizable! So here are my (non technical) thoughts on what is going on. I still have a lot of exposure to peering as the company I work for has 3 AMS-IX POPS (accounting for close to 50% of the traffic) as well as POP of NL-IX, ECIX and a couple of others, I actively promote peering to my customers.

There are a number of factors that I have noticed that are having an effect on the interconnection market and all these together add up to a change in direction ( the work game changer is being knocked around by some people) each element is small but together then all add up……..

First factor is the role of the IX and the way they work. First of all we are seeing IX’s become distributed (some more than others) and in a couple of cases they are connecting their nodes cross border – this allows people to peer cross border with no additional transit costs. This product could have a negative effect on a) carriers who have their revenue eroded and b) other IX’s who could become uncompetitive. None of the so-called ‘big 3’ offer distributed exchanges (either Europe of US) but it could be this will change before long.

The second factor is the continuing drop of the IP transit price (how low can it go), although on a pure cost per MB level the IX’s are still much cheaper (at 10GE levels) but you also need to factor in the additional ‘hidden’ costs such as local coloction, transport to reach the IX’s as well as the human cost of expertise to manage the peering connections and keep them going. I am seeing a move to fire and forget transit from multiple vendors. The transit market is evolving too with lots of consolidation (Level3 / Global Crossing – Hibernia / Atrato – GTT / TINET) meaning people have real global networks with capacity to shift. In some regions an ‘airline’ model is evolving where people will shift local capacity at low-cost in order to get people to commit to the more profitable ‘intercontinental routes’.

Thirdly we are seeing the good old buzz word of moving into the cloud! What blog would be complete without this topic. However joking aside this is one of the first real game changers that are emerging, this is a risk to many parts of the ecosystem with the benefits falling on a) the cloud providers and b) the end-user is they use the offering correctly. The shift to systems like AWS and Azure means that there are is less diverse requirements on the market (transit, IX ports, co-location etc) but the requirement to get to the cloud providers get bigger. It’s a case of you are in the club and benefiting (Cloud Partners like carriers and co-location) or you are on the outside looking in (not a great place to be).

It might appear until now that I think its only the IX’s who need to adapt but they are not alone and are in good company, this shift could affect the traditional carriers and co-location providers.

A major shift when using the cloud services of the monsters such as AWS and Azure is the rise of bandwidth on demand offerings, the ability to connect to the platforms for a short time (sometimes measured in minutes) without the requirement to sign long-term contracts with bandwidth providers (typically 12 – 36 months). This is great for the end-user but how can the carrier operate a long-term strategy without secured revenue? Will this led to lower service levels, old networks and general over booking? For the carriers who own lots of assets and have for many years this isn’t so much of a problem (the lack of long-term commits) as this probably makes up a smaller percentage of their revenue. I am however interested to see how some of the new players fair as they must need to have the access to high levels of bandwidth (not to mention co-location costs) but their customers have no commitment to them*.
*This is an assumption as I have not seen end-user contracts.

One of the consequences of the shifting market (or is it a driving force behind it) is that in some cities (Amsterdam, Frankfurt, London) some of the DC;s are turning into Megahubs where most of the key players are – there is simply little need to be elsewhere. This shift is generally limited to 4 provider in Europe, this is good for the vendors and good for the end users in terms of a) market share and revenue and b) ease of connecting to the major players but it might have an adverse effect on the local players. The opinion in the market is very split about how things will pan out, will the smaller vendors slowly fade away or will they take the opportunity to be creative and provide a real alternative to the Mega hubs. The large traditional vendors are starting already offer additional services (IAAS, Firewalls, IP etc) and are beginning to expand their service with connectivity into the cloud providers.

Some of the ‘upcoming’ DC providers are even offering co-location ‘by the hour’ with people only paying what you use, again that’s what they say and I have not seen the T’s & C’s. This business model is something I am trying to get my head around, yes it is not so common to have people leave once they are in but with no commitment from them how do the co-location providers plan their business and structure the ongoing investments.

In summary as ever it is an exciting market it is clear that ‘traditional’ companies like Carriers, IX’s and co-location vendors are in a changing market and more changes appear to be coming. If you look into the history or the IT industry (or any industry come to that) the biggest companies have to evolve or become extinct. All parties need to evolve and if they don’t then it might be a rough ride.

In a couple of weeks I am off to the European Peering Forum and it will be interesting to get people’s views there.

Till the next time!

The changing IP interconnection scene, observations.