Special Delivery

This guest editorial comes from Pat Chapman-Pincher, who has worked within the telecommunications industry for more than 30 years - with experience ranging from multinational Internet and telecoms companies to technology start-ups.

Chapman-Pincher has held senior roles at UUNET International, MFS International, IBM UK, Logica and BT. Today she is chief executive officer of CacheLogic, a global provider of content delivery network services.

Here, Chapman-Pincher discusses the games industry's shift towards digital distribution - and questions whether traditional CDNs will be able to cope.

We don't encourage the use of stereotypes at CacheLogic, but you might be thinking I don't look much like a gamer. And not much like a plumber either, let alone a FedEx or UPS truck driver. You'd be right. I am none of those. But, hopefully, you will find them - my analogous alter egos for the sake of this argument - nearly as useful as I do in understanding the role and practicalities of online retail games distribution.

Before we talk plumbing, some context and a prediction. In the UK, according to ELSPA, the number of interactive entertainment software units sold in 2006 was up 7 per cent on 2005. There were 65.1 million unit sales, worth a whopping GBP 1.36 billion. Cue sound effect of old-fashioned cash register. The US games market meanwhile was worth USD 18 billion according the Entertainment Software Association. Double kerr-ching.

Now for the prediction - not mine, thankfully: at the end of 2005 Screen Digest said that digital distribution of games "is expected to be strong and constant from now until 2010". Anybody noticed most of their retail games business migrating online yet?

Okay, back to plumbing. In the late nineties and early noughties, as president of the international part of the world's largest ISP I oversaw the planning and installation of much of the Internet. And very nice, reliable and useful pipes they are still.

Clever people have figured out how to use this for all manner of business and leisure purposes requiring ever-more bandwidth and many thousands of servers. For purposes of this discussion the most pertinent current - aspirational, actually - use being online distribution of very large games files and patches. Not to mention movies and all of that zeitgeisty IPTV stuff.

Right now there isn't a whole lot of online games distribution. And if suddenly it were to happen, to be brutally honest with you, that plumbing that I am so proud of wouldn't cope too well. There'd be plenty of people swearing at their PCs.

Moreover, as incumbent content delivery networks (CDNs) continue to throw their customers' or investors' money at the problem, the economics of large object distribution remains so broken that it makes for an unsustainable business model.

Before I finish thrashing my plumbing analogy with a lead pipe, one more thing: actually the plumbing is in good shape. Yes, I would say that wouldn't I? But I am about to explain why the economics of online large object distribution are so badly broken that it is they - the broken economics - not dodgy plumbing that's holding things up.

Now I need you to imagine something for me. Shouldn't be too difficult; gaming is all about imagining. Imagine I am not the CEO of CacheLogic but, instead, the boss of a famous games developer and I've decided to take the plunge and make Doom 4 available for download after a successful PR and marketing ramp-up. It's going to be huge!

And let's imagine I haven't heard of CacheLogic. I've signed up with a "traditional" CDN. Everything's uploaded onto their many thousands of servers connected to giga-oodles of bandwidth.

At the appointed hour the formidable monsters inside my latest gaming masterpiece will be available to a flash crowd of unprecedented size - hundreds of thousands. Millions? Will the system cope without the longest ever worldwide wait, frustrated gamers and a mess of bad press for me and my service provider? Maybe; maybe not.

Maybe not because the existing infrastructure wasn't designed for hugely simultaneous distribution of large files. It was designed for rapid delivery of web pages of a few tens of kilobytes at most. Incumbent CDN providers have tried to get round this by throwing expensive servers and bandwidth at the problem.

On its website one of the largest boasts something like 20,000 servers currently coping well with populating tens of millions iTunes libraries. That's 20,000 servers before they've up-scaled to deal with my keenly anticipated release. That's 20,000 servers doing their best to serve single-figure megabyte MP3s, not multi-gigabyte games files. But while we're imagining, let's imagine my CDN gets its sums right and it all goes well. What about the cost? Imagine that!

Putting it simply, traditional CDN providers read the meter. Their content-owning customer's bill is based on megabytes of data leaving the CDN's servers regardless of whether or not the end user had a successful download first, second or, even, third time lucky. That's right, delivery isn't guaranteed or, for that matter, recorded. And the more gigs in your game, the more it'll cost you. It's hardly FedEx or UPS.

Which is a convenient link to my final analogy. You can't blame incumbents for imposing this unsustainable business model on content owners and you can't blame content owners for not queuing up at the gates to the server farm with truckloads of DVDs. After all, you have to use the tools at your disposal, don't you? And there's no real-world service with sustainable courier-like accountability and pricing/speed flexibility available. Is there?

Well, no you don't, and yes, actually there is. You could - if you had a lot of experience of Internet engineering, of intelligent caching and a realistic understanding of the strengths and weakness of peer-to-peer - you could build a CDN that is fit for the purpose from the ground up. You could build an intelligent, peer-assisted CDN that charges only once for each game delivered, that guarantees and records delivery, and can cope with the biggest flash crowds, and you could offer as sustainable, flexible, per asset pricing model.

Okay, fair point, you'd be more inclined to build such a splendid CDN if you didn't already have a huge amount of capital sunk into thousands of servers and phenomenal quantities of bandwidth.

Here's my prediction: it's a tautological tour de force. More big games will be distributed online when more CDNs are designed and optimised for distributing big games. Right now I can think only of one such outfit. But I promised I wouldn't plug it too much!