If you correctly architected your application around the constraints of AppEngine, then it wouldn't be costing you a lot of money. That is the whole point... you know the constraints, so you architect wisely and you spend money where you need to. You decide which indexes you need and which ones you don't need. You decide how many reads and writes you are going to make. You figure this stuff out in advance and if you make a mistake, you refactor your code to improve things.

I'm not saying it is easy, it isn't. It takes a skilled engineer to learn this stuff and make it work. But, when it does work, it really does work.

It sounds like you failed to create an application that took all of that into account and of course you are going to look 'real stupid' when your clients figure out that you took shortcuts. There is no way that could be the fault of AppEngine.

I'm curious where you went after AppEngine. Are you hosting on AWS now? Heroku? EngineYard? Do you have the same reliability and scalability as what AppEngine provides? Maybe that isn't something that is beneficial to you and in that case, I can see how AppEngine is not your cup of tea.

For me, AppEngine is amazing. I love the fact that I'll never have to hire a sysadmin. I love not carrying a pager. I love knowing that when my site gets an asston of traffic, I won't ever have to think about making it scale. I love not having to worry if my database is on big enough hardware, replicated across data centers, backed up. I'll never have to think about whether or not my OS needs an upgrade to plug a security hole or ssh'ing into a server at 2am to figure some esoteric problem out. To me, all of these things are worth the 'cost' of AppEngine. I'd rather spend my time adding features than doing sysadmin.

If something is marketed as a preview release, don't rely on it for critical infrastructure.