"... With Google, all the costs are hidden or down the road, which is especially scary given the lock-in ..."
I'm not going to start advocating this google model is good for everyone. If fact I'm pretty sceptical of google ~ http://flickr.com/photos/bootload/tags/google/page2/ but I found this particular tool good enough, now for me. For example. If I want to get a site up and running with django, python 2.5, a scalable db, with a domain name with an isp right now, I'm pretty sure I'd either a) have to host my own, $$$... (cost of pc, cable, time with admin) or b)
get some rack space at an ISP that has a later version of python (if I use CGI) and django again $$$, $$$. Now looking at the tools available I'm pretty sure you could get users to register their details with your application & keep the data & export it later. I haven't read the license details about this. (there's one problem, checking for and asking for permission) As for the tools I'm pretty sure you could re-create the db access on another system.
What I do say to myself is, "could I prototype an idea here" then investigate moving if I move to the next level that may incur hidden costs.
"... Need to extend your application with features Google's platform can't provide? Well, you'll need to rewrite the back end of you application. And ask all your users to re-register ..."
I think you can solve the later problem as suggested above. As for the first. Well there seems at the moment things you cannot do with this platform. Computationally extensive applications that use 'C' based code. It appears what google is offering is the front-end tools that scale. Not the back-end computational processing.
"... The only people who are really squealing are the hardware/software vendors ..."
Tim O'Reilly argues this maybe a lock-in play ~ http://radar.oreilly.com/archives/2008/04/is-google-app-engi... Until I read the license to verify you have access to your data I'm less worried about this than google fighting the Internet. I think we are going to see more "level 3 platforms" which have ' runtime Environments' ~ http://blog.pmarca.com/2007/09/the-three-kinds.html so they are not going to go away. Trying them out might allow you to see a problem to work on exploiting insight you have gained.
Some reasonable points, especially if you're already committed to Django as a platform.
BTW, I meant back end as in data store, not computation. From what I understand, using BigTable differs markedly from using MySQL or Postgres, and you would have to port to one or the other to leave Google unless you want to do some serious development work on your data back end.
I hope you're right about being able to salvage your user data, but I would study both the license agreement and the technical capabilities very closely on this one. It would be a shame to build an early site thinking it was just a starting point and then discover that you could never change key aspects of it.
"... Some reasonable points, especially if you're already committed to Django as a platform. .."
Not really. There are some aspects of Django that really suck. Complex template logic for one . I really appreciated the simplicity of webpy after wading through the django api. On a side note, friendfeed's (Bret Taylor) ~ http://bret.appspot.com/entry/experimenting-google-app-engin... worked on google appengine. When he built friendfeed he chose webpy as a starting point. There is a lesson somewhere in there. They key thing is I can use CGI, webpy or django so I'm not bound to any one choice.
"... I hope you're right about being able to salvage your user data, but I would study both the license agreement and the technical capabilities very closely on this one. It would be a shame to build an early site thinking it was just a starting point and then discover that you could never change key aspects of it. ..."
Good points I'll have to think about them. The one that will matter most will be the user information. The app I have in mind is a design tool of sorts that generates object code for users. I don't think appengine as it is will work for the back-end ideas I have but who knows if users really want this?
I'm not going to start advocating this google model is good for everyone. If fact I'm pretty sceptical of google ~ http://flickr.com/photos/bootload/tags/google/page2/ but I found this particular tool good enough, now for me. For example. If I want to get a site up and running with django, python 2.5, a scalable db, with a domain name with an isp right now, I'm pretty sure I'd either a) have to host my own, $$$... (cost of pc, cable, time with admin) or b) get some rack space at an ISP that has a later version of python (if I use CGI) and django again $$$, $$$. Now looking at the tools available I'm pretty sure you could get users to register their details with your application & keep the data & export it later. I haven't read the license details about this. (there's one problem, checking for and asking for permission) As for the tools I'm pretty sure you could re-create the db access on another system.
What I do say to myself is, "could I prototype an idea here" then investigate moving if I move to the next level that may incur hidden costs.
"... Need to extend your application with features Google's platform can't provide? Well, you'll need to rewrite the back end of you application. And ask all your users to re-register ..."
I think you can solve the later problem as suggested above. As for the first. Well there seems at the moment things you cannot do with this platform. Computationally extensive applications that use 'C' based code. It appears what google is offering is the front-end tools that scale. Not the back-end computational processing.
"... The only people who are really squealing are the hardware/software vendors ..."
Tim O'Reilly argues this maybe a lock-in play ~ http://radar.oreilly.com/archives/2008/04/is-google-app-engi... Until I read the license to verify you have access to your data I'm less worried about this than google fighting the Internet. I think we are going to see more "level 3 platforms" which have ' runtime Environments' ~ http://blog.pmarca.com/2007/09/the-three-kinds.html so they are not going to go away. Trying them out might allow you to see a problem to work on exploiting insight you have gained.