Google App Engine team answering about New Pricing (ongoing FAQ)
groups.google.com
groups.google.com
That makes a lot of sense, I have a few apps that I get bills for a few cents here and there, that can't be cost effective after credit card fees.
$9 will not be a fee, but a minimum spend.
If your effective resource usage is <$9 you'll pay a $9 bill (not $9 + the bill amount, as previously Google stated); if your bill is >= $9 you'll pay that bill amount (not $9 + the bill amount, as previously Google stated).
It is however a good change.
Hope it helps.
On months you go over the old free limit by $12.34 you will be charged $12.34 instead of $21.34.
It is a good change (minimum spend v. privilege to pay) but still quite a big change from the old pricing structure.
No, you will be charged at least $9 every month because you can't tell in advance when/if you will exceed the free quota, see you must leave billing enabled all the time, just in case.
The new proposal is better than the previous proposal but it still sucks. Large customers get $9 of value and startups get shafted.
Now, I want to try out GAE. I do so with local / dev appserver and free junk until I get things kind of working and start getting traffic / traction.
I guess the ~good~ point to all this is that if you are going over the free limit (ie: start getting serious traffic) what el goog wants you to do is put on adwords, earn $9/mo and then turn right around and give it back to them for "hosting costs".
Really it just leaves a huge sour taste in people's mouth that this is such a dramatic (and un-warned!) change in pricing for really a monopoly service. Assume that you were heavily invested in GAE / DataStore and have... ohh... 1tb of data or something with them. Where / how else are you going to solve these "impossible" problems that the datastore solves for you?
Much better to not get tied down with google's quirky (yet scalable) model in the first place so at least you have the ability to move between competitive providers instead of getting stuck with a monopoly provider that doesn't know how (socially) to charge people for services.
I expect that is exactly the kind of application that they want to dissuade with this change.
I'm working on something where I was thinking about using S3 for static and building a really simple GAE datastore-backed JSON API pulled in from the static files.
Nothing complex but I want it to perform as well as possible. If GAE would work just as well, I'm thinking I should just keep everything there.
Do the front-end cache servers prevent unnecessary hits on the app server instances?
So people can only make testing app that doesn't cost anything or $9 for each paid app. Judging from the "tick" list (marketing props) there is not much you would get for $9 one.
I've been a strong GAE proponent since the very early days [1]; GAE's sudden price announcement has made me question my belief in PaaS. Perhaps IaaS really is the level to build on: unless you're using bespoke stuff like AWS auto-scaling you can plausibly keep your infrastructure code "on the outside" of the service itself.
--
[1] Here's my talk at I/O 2009, when App Engine was young! http://www.google.com/events/io/2009/sessions/AppEngineNitty...
I have convinced all my clients this year to do their new projects on GAE.
The first thing I worked on was a simple abstraction library to sit on top of GAE so you could port away from it easily if you need to later on. I plan on releasing the code at some point in the next few months (the db part is the hardest, going to support running on Riak for now)
Also, I recently integrated the Google Prediction API into a product. Damn that thing is hot - coolest new thing I have played with in a long time. Those additional service API's make GAE a dream.