How Riot Games Became a Telco

Telecom networks are changing in some important ways. A few weeks ago, I wrote a somewhat fanciful piece about the ways in which telecom networks are being unbundled by the growing convergence between ‘enterprise’ and ‘carrier’ networking technologies.A few days later, a friend pointed me to an interesting blog post (Part 1 and Part 2) written by Riot Games’ engineering team that provides a very interesting, real world example of what I was trying to describe in my post. Riot Games is the developer behind League of Legends, one of the most popular online games out there. For those not familiar with LoL, it is a real time combat game pitting teams of five players against each other. Games like this rely on fast Internet connections, if one player has a slow data rate their character lags, and in split second fights often lose. This is no fun, as it means that bandwidth can become more important than game skill. And since LoL is one of the leading eSports events, with literally millions of dollars in contest prizes available, a slow Internet connection can have some real, physical world implications.

The Riot blog post comes in two parts, the first outlining the limitations of today’s Internet architecture for high-speed, low latency gaming. The second post details how Riot sought to improve speeds for its players. To put it simply, their solution was to build their own Telecom network, or as they put it, they built their own Internet. Riot leased their own fiber backbones, installed routers in key Internet exchange points, and signed peering agreements with many (dozens?) of Internet service providers (ISPs). This is serious stuff, very low-level plumbing work. It is basically everything a telco operator does with the exception of building last-mile connections. And it worked, getting 80% of their North American players below 80ms lag, basically the threshold below which delays happen too fast for humans to notice.

All of this for a game.

I found this relevant because Riot has basically unbundled the networking stack of the telecom operators. A concrete implementation of exactly what I was hypothesizing (fantasizing?) about in my post. The Riot post even details the reasons why they tried but failed to work with the big carriers – basically the carriers could not customize their service to deliver the speeds Riot required.

This kind of thing calls into question the very nature of a Telecom Operator’s existence. Riot has some very talented engineers and architects, but I do not think they are the pinnacle of talent. Many others could replicate the work of Riot if they chose to.

In the comments of their post, several people asked the next obvious question – would Riot offer other game companies access to their networks. While they did not address that, and have no reason to do so, it is a powerful idea. In fact, I think there is a big opportunity for someone to build just that sort of network. There are many games out there that require this sort of low-latency specialization. And I would argue that the economics favor a platform shared by many customers. The basis of such a network are:

• Routers

• Real estate at key Internet junctions

• Peering relationships with ISPs

For the most part, those are all fixed cost, upfront capital investments. Build out the service and the more the owner loads it up the better the profits. The real question is how much would gaming companies pay for this? I do not have an answer to the question, as it is hard for them to monetize directly – they cannot have a two-tier pricing system for their players. Nonetheless, I think many would pay for the improved performance if it made their game more enjoyable. I imagine the serious online gaming companies can precisely quantify the impact of lag on their revenue. And, eventually competition in gaming (which is immense) will drive those who do not participate to fall behind those who do.

And if it works for gaming, why not other fields? The game companies want fast, low-latency networks. Other companies may want cheap, high-latency connections. Or networks with extremely low number of intermediate connections for security reasons. There are many possibilities. This is not a new idea. People have been kicking these ideas around since at least the 1990’s Internet bubble. What has changed is that the enabling technologies are now much cheaper and simpler to build. You would not need a team of some of the world’s best telecom architects to implement this, the required expertise is now much more widespread.

The point being that the way we think of Telecom Operators today is steadily changing. This opens the door for new business models as technology enables implementation of segmented marketing. If Riot has done this, so will others.


4 responses to “How Riot Games Became a Telco

  1. I would not say that they “built their own Internet”; they simply connected to the Internet as a first class citizen like carriers and ISPs do. It may seem a small difference but it isn’t, because they are still connected to the Internet; in fact they’re now a member of the Internet backbone. They did the same Google and Facebook have done before; they became a content provider network (CDN).

  2. Another important point is that regardless of how big is the investment, or how complicated it is to setup this network, they’re still just a content provider and not a telco. Telcos run the access business too, which Riot Games don’t have to run; it’s a completely different business, with different engineering and operational challenges. They’re very far from becoming a Telco from this perspective.

    • That’s my point. The definition of telco is changing. While I agree the metro access business is hard to get into, it has become vouch easier to get into the other parts of the business. This is going to make the telcos rethink their competitive advantages.

  3. Great article, thanks. Thinking about it more I believe you can think of Xbox Live as a specialised low latency network that Microsoft charges customers for (£40 a year) and that games companies pay access fees to use in their platform sales royalties. They’ve even started to integrate it with their cloud compute offering (although I don’t believe many people are using that yet). Valve has matchmaking and content hosting services as part of Steamworks but I don’t think they have a similar low latency backbone.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s