I made a teensy-weensy mistake when I started to look at Google App Engine: I downloaded and installed the SDK and read through the Getting Started Guide fairly thoroughly before signing up for an account. As a result, I've been wait-listed. I think that means that more than 10,000 others have already signed up for the free App Engine beta. Oh well, I can still develop locally until my invitation comes through.

As about a million other bloggers have already discussed, Google App Engine feels like a direct competitor to Amazon's three Web services (EC2, SimpleDB and S3) all rolled into one Python framework.

I think it would be really nice to be able to target the Google infrastructure "cloud" for a Web application at need, just as it's really nice to be able to target the Amazon infrastructure and the SalesForce.com infrastructure at need. I can see different uses for the various platforms as currently constituted; I can also see why the choice might confuse people.

I like the choice of Python as the first implementation language, unlike many other bloggers who seem to be whining about the lack of Ruby and PHP support. I also like the way Google has given us a local server for development, and given us access to most of Django (a Web-development framework), WebOb (which provides objects for HTTP requests and responses), and PyYAML (a parser) as well as most of the standard Python runtime libraries. I think I can learn GQL without a problem: it's basically a subset of SQL.

I'll pass over the way the HuddleChat demo ripped off the 37Signals Campfire real-time chat application, for two reasons. First, about half a million other bloggers have already complained about it; second, Google has already bowed to the pressure and pulled the app.

I wonder what the 10,000 others who have already signed up for the free App Engine beta are going to do with it. In fact, I wonder what I'll do with it when I eventually get access.

What's a potentially profitable Web server application that needs great scalability, doesn't need table joins, and hasn't already been done to death?

Posted by Martin Heller on April 9, 2008 12:02 PM