The whole instances pricing thing is too complicated for average brains like mine

195 views
Skip to first unread message

Daniel Florey

unread,
Sep 3, 2011, 5:13:40 AM9/3/11
to google-a...@googlegroups.com
Is there no way to create a simple pricing model that will not force me to learn everything about the internals of the app engine scheduler?
I'm fine with paying for consumed resources at a reasonable price but I'm not feeling comfortable with the instances/hours approach.
It makes me feel like the scheduler impl is responsible for my bill.
I'd prefer a simple approach (from a users perspective) where I would get charged for the cpu resources/memory consumed + be able to spend a budget on "scalability/processing power" and let app engine take care of whatever it needs to speed up / slow down my app according to my settings.

Nick Rudnik

unread,
Sep 3, 2011, 12:16:42 PM9/3/11
to google-a...@googlegroups.com
I completely agree. This is why the cpu time was a superior metric to instance hours. That way we had a reason to write efficient code and Google had every reason to write an efficient scheduler to not waste instance hours.

Francois Masurel

unread,
Sep 3, 2011, 12:35:24 PM9/3/11
to google-a...@googlegroups.com
I totally agree too.

Arun Shanker Prasad

unread,
Sep 3, 2011, 2:19:18 PM9/3/11
to Google App Engine
+11111

absolutely agree, may be Google can put up an an article detailing
what this means to normal users?

Deepak Singh

unread,
Sep 3, 2011, 2:46:48 PM9/3/11
to google-a...@googlegroups.com
+11111111111

Totally agree. There should be simplified billing process.


--
You received this message because you are subscribed to the Google Groups "Google App Engine" group.
To post to this group, send email to google-a...@googlegroups.com.
To unsubscribe from this group, send email to google-appengi...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.


Frank Müller

unread,
Sep 3, 2011, 2:56:12 PM9/3/11
to google-a...@googlegroups.com
I would like a package based pricing. Like

  • Free with limits of package S or M for non-commercial projects
  • Package S with 1 non-always-on instance, 1 GB database, ...
  • Package M with 1 always-on instance, 2 GB database, ...
  • Package L with 2 always-on instances, 10 GB database, ...
  • Package XL with 4 always-on instances, 25 GB database, ...
  • Package XXL like the pricing today
So choosing a package to start would be simple. An additional warning mail when a limit is almost reached and a simple switch between those packages would be great.

My 2c

mue

an0nym

unread,
Sep 3, 2011, 3:35:45 PM9/3/11
to google-a...@googlegroups.com
Totally agree. New billing is too complicated to control and too expensive when you don't do it. Old billing wasn't and it's low cost in absolute wash't the reason.

Tim Hoffman

unread,
Sep 3, 2011, 7:10:17 PM9/3/11
to google-a...@googlegroups.com
Oh this is all nonsense.

You are all smart enough to understand how it works.  

Google has dropped the ball, not providing appropriate real time tools to monitor instance hours and what it will cost you.  

You can also argue about the level of pricing, but to suggest it is any more complicated is just silly.  

Every one use to complain about the inscrutible way cpu time was measured before.


Francois Masurel

unread,
Sep 4, 2011, 4:11:44 AM9/4/11
to google-a...@googlegroups.com
Hi Tim,

You right, CPU measure is quite mysterious, but there is nothing we can do about it and we just have to optimize our code to reduce CPU use.

Francois

Raymond C.

unread,
Sep 4, 2011, 4:23:14 AM9/4/11
to google-a...@googlegroups.com
I agree the instance based pricing is too complicated because of the nature of how GAE works.  On GAE we have no control on resources, Google control it all. What we can control is what our code does during the life time within a request.  

On instance based pricing (the upcoming one) all we are ASKED to do is to make our request return faster.  However there are many area we cannot control regarding the time: datastore wait time, URL fetch (the only way to communicate between backends and frontend instance), etc

GAE have been charging for CPU time for all operations because thats what we can almost fully control (except extra datastore performance due to google's server performance), thats why most GAE users (at least the heavy users) think the old charge model make more sense, even GAE charge us more on CPU based pricing.

Many of us have raised the concern from the beginning that instance based is not how GAE works.  Still they are going for it without a valid reason, at least without much reasonable reasons (to me).  There are so many methods to charge us for using their machines or occupying the memory of their machines, yet they go for a way thats not how GAE works.

Tim Hoffman

unread,
Sep 4, 2011, 4:28:51 AM9/4/11
to google-a...@googlegroups.com
I totally agree with you francois I just found the whining without  constructive dialog starting to be a more than a little painful.

I don't have a problem helping people get more out of appengine, but baseless unsupported sniping isn't helping anyone, and the signal to 
noise ratio is drowning out contructive suggestions for people trying to work out what they do about the coming changes.

As I said before I am not a google aplogist. They do good things and silly things ;-)

See ya

T

IPv6

unread,
Sep 3, 2011, 7:34:38 AM9/3/11
to Google App Engine
not agree... hours stuff much easy to understand that CPU+API calls
+bla-bla-bla
But prices... they are just too high for low-profile application
Imho
Reply all
Reply to author
Forward
0 new messages