Harry's choices

When Harry looks at what he can use, his choices do not look good...

• Distributed computing platforms. There are several of these available. But deployment with these is difficult. Harry's code needs to move around. It needs to find new places to host itself as it grows to meet players' demands. And it needs to do this by itself. Current platforms can't do that. They need a bit of help.
• His own hosting. Harry could become his own ISP and run his own trusted global network. But this is not his core skillset. He also finds himself several million dollars short.
• Traditional ISPs. Leave it to the experts. The downsides are: large up-front costs; global integration is difficult and expensive (particularly when he will be using different ISPs); and if he wants to set up in Japan then he needs to start spending several months ahead, with a prediction that that market will take off when he is complete--and he cannot afford to be wrong.
• Write everything himself. Harry will write the basic application, but does not want to re-invent services which others already provide.
• Traditional ASPs. He can use ASPs, but this is usually a design-time decision. Changing to another ASP will be a pain. Harry does not like vendor lock-in. And he will spend too much time integrating with them.
• Remote-services platform. Some services are provided remotely. They use protocols like SOAP. But not every service is suited to this. And even if a particular service usually is, then that will still not be right for every client machine. Not everyone is always connected. An alternative is...
• Application-oriented systems. But again, not every service and every players' machine is suited to large client-side applications. The third party services he uses--and his own--must be adaptive.
• Money. Harry's game code should pay its way, but he cannot afford large up-front costs. He can afford to pay an ISP if his game uses a little bandwidth, but he does not want to pay the ISP's minimum charge of a 2Mb link on day one. He does not want to buy 100 servers months before he needs them. VC funding would help him in the short term, but would reduce his share of the profits in the end. Harry wants to pay for what he uses, and no more.
• Building in bank integration. As his application moves and grows round the world, so the players will want to pay from various bank accounts in various currencies. His code needs to enable it, but the banks should be able to work out the detailed mechanics themselves. That's a quagmire for someone like Harry.