I am super excited about Google's recent foray into cloud computing, the app engine. They are not first, I realize Amazon has been doing some pretty cool things with the S3 services and their computing services but the ability to simply and easily have an application that can leverage BigTable and GQL, google account authentication, image manipulation, memcache and having that all live within the seemingly infinite scalability of the google platform is extremely exciting for me. In fact it's the whole reason for my recent resurgent interest in python. App engine out of the gate is a python runtime environment which is robust enough even to run frameworks like Django for your web apps.
ArsTechnica had some good comments on the service and one of their cheif concerns was being locked into the platform. I think this is not an invalid concern, but that the tools will be written to make migration from one platform to another a relatively trivial matter. From what I understand of GQL it is a subset of normal SQL with limitations like no joins which should make data/code migration easy, but scalability and hosting maybe not so much. There are open source implementations such as HyperTable which is modeled after Google's data platform, but ultimately the article is right. Personally I don't see myself writing large scale applications on this platform though, I see this much more as a utility model for small useful services.
I registered for the preview of app engine on the same day it was announced but not early enough to be one of the lucky first 10,000. When I did get my invite a couple weeks ago I immediately started to create applications. It's clear this is exactly what everyone else did and I'm glad Google has capped usage at three applications per developer. Already it's almost as bad as trying to register a domain. For each of my three applications I had to try a series of names before getting a clear one. In other words none of my ideas are at all unique. :-) On a positive note when I checked the names I was looking for there were also no uploaded applications for any of them. I predict Google may have to have an expiry date on these - I can see myself taking weeks or months to finally get anything uploaded.
http://silitrader.appspot.com (silverlight port of the old game OilBaron)
http://wherediditallgo.appspot.com (personal spending tracker)
I actually created that last application as I was trying to find a simple means of having a rotating list of hand picked quotes show up in the header of my blog. Back when somatose.com was hosted with LunarPages this would have been trivial, throw the quotes into a textfile, write a script to randomly choose and insert one as I was building the page. I moved from a hosted environment to using blogger so that I would stop tinkering and focus on actual content. I spent most of tonight tinkering though, building up to actually building blogquotes. (By the way I am sure there are other blog quote like tools out there, but this is a rare case where I don't mind reinventing the wheel somewhat)
Now the nature of the tinkering has just changed. Rather than write a small shell script that spits out random lines I will actually build something of a small application/service. The quotes will be stored using the datastore api (GQL), and will be publicly visible. Other users will be able to authenticate and store their quotes in a similar manner. A widget will be built and hosted from that application that enables anyone to paste in a snippet of html to add random quotes from their personal db. I as a user can subscribe to my friend's quotes. It's remarkable really, the differences and yet familiarity of this new ecosystem. And while I do miss the control of simply writing a little shell script I have to say I love the idea of the internet being populated by these little widgets of functionality. The emergence of ajax and increasing power on the client means that millions of non-technical users who blog on platforms like blogger and wordpress can very easily use my quote engine. How cool is that?
Google's API's (and Amazon's) are proving to be the next Win32. Massively widespread and base functionality that anyone can build on. The plumbing is there for us to be able to chain together these small tools/widgets in much the same way a clever unix user could within their shell. And once the honeymoon wears off and I actually give some though to hosting data in the US and dealing with popularity should one of my apps get hammered and I end up having to pay Google usage fees... well until all those little issues I can just be excited again about the changing landscape.