April 08, 2008

Google App Engine to challenge Amazon, Ning etc.

Write your applications in Python ... w00t!

Update : OK, this is big, at least this is going "mainstream". Poor Ning, they had all the great ideas, Amazon too ... but it sounds like Google are taking the whole package : integrated development environment, database, python, logging, version control, multiple developers working on same code-base etc. etc. and bringing it to the masses.

But only a limited number of free accounts ...

Update 2: early chatter is complaining about Python-only, actually I'm not sure it's gonna be Python that people struggle to get their heads around ... I think it's gonna be BigTable.

3 comments:

chad said...

why is this so exciting to everybody? is there some problem with the $15 linux / mysql hosting providers out there? why would somebody want to learn all of these proprietary apis to do what we do on our accounts with ftp / email / http hosting already? it works perfectly well.

Python only? this looks like a rush job. i mean compare the feature set to Mosso and its no comparison. This is just the google internal dev toolkit. Why would anyone building a web business want to be totally dependent on google hosting tools?

Composing said...

chad ... what, to me is exciting about this :

a) it's good confirmation of the trend towards commodity application hosting and management ... see Simon Wardley on why this is potentially so interesting. ( http://blog.gardeviance.org/2007/11/web-20-expo-berlin.html )

But, of course, as he's been warning, unless apps. are portable there's a vicious risk of lock-in. I agree about the danger.

b) The difference with the $15 dollar hosting (of which I have two accounts) is *partly* the automatic scaling if my application suddently takes off, but *mainly* the rest of the infrastructure ... particularly the support for developers. That's what excited me about Ning too.

I forsee that these managed hosting stacks are going to also be integrated development environments ... look how GAE seems to have built-in version control and rollback to earlier versions if you make a mistake; how it allows multiple developer accounts etc.)

At some point we'll see integration with the sort of functionality of a fogbugz or edgewall trac (bug tracking, wiki, project management, maybe hyperlinking within source-code, maybe realtime or twitter-like chat between developers)

Developing is going to change considerably in these environments ... a lot of the drudgery will become a thing of the past.

Imagine if there can be automatic publishing of GAE apps. to GoogleCode (if you want) and installing from GC to GAE. So anyone can pick up an existing application direct from GoogleCode and start customizing it. That was the original Ning vision (of course) before they reinvented themselves as a YASN.

Not only a lot of the hosting becomes commodity, so does a lot of the "application" on top.

c) for me *personally* the python-only thing is a bonus ... one in the eye for Rails (bwaahaahaa) - er, no ... I mean ... er - it will boost python's popularity, and as I already have Python CGI apps that I can experiment porting to it I have a head-start on everyone else ;-)

chad said...

Well - yes I guess the promise of headache-free scaling is indeed useful. Lets see what happens, clearly this is a VERY early iteration of this kind of product. I'm sure in time we'll all be using this kind of auto-scaling system.