Significant quota changes. . .

[theillustratedlife]

I just read the blog post and linked quota changes. Although I don't
think I'll be affected by the new quota yet, I think the magnitude of
changes is worth noting:

CPU
current quota - 46 hours
new quota - 1 hour
% change - 98% reduction

Bandwidth
current quota - 10GB
new quota - 1 GB
% change - 90% reduction


On a separate note, cheers to the App Engine team for bringing us
billing! =)

[Brett Slatkin]

> CPU
> current quota - 46 hours
> new quota - 1 hour
> % change - 98% reduction

I believe the new number will be 6.5 CPU hours:

http://code.google.com/appengine/docs/quotas.html#Free_Changes

[Brandon Thomson]

Has anyone seen a cost comparison between Google and the other
providers based on these new changes?

[theillustratedlife]

Typo on my part. Thanks for the catch. =)

Revised changes:

CPU
current quota - 46 hours

new quota - 6.5 hours
% change - 86% reduction

[Wooble]

On Feb 24, 5:57 pm, Brandon Thomson <gra...@gmail.com> wrote:
> Has anyone seen a cost comparison between Google and the other
> providers based on these new changes?

EC2 free storage: none
EC2 free CPU: none

you do the math.

[Jarek Zgoda]

Don't compare apples to oranges, EC2 is complete operating system
instance.

[peterk]

Maybe I'm misunderstanding you, but since when was it free to run a
EC2 instance?

EC2 Small Instance per CPU Hour is $0.10 (same as GAE's cost 'per CPU
hour')
EC2 outgoing bandwidth is $0.17 per GB for the first Terabyte out. S3
outgoing bandwidth is the same cost for the first 10TB out. GAE is
$0.12 per GB out.
S3 storage is $0.15 per GB for the first 50TB. GAE is $0.15 per GB.

I'm still reading, we may be comparing apples to oranges anyway with
regard to CPU time. Does CPU time on GAE cost if your application
isn't actually in use, or do you pay literally just for the cumulative
number of seconds your app spends actually executing? On EC2 you have
to keep an instance running 24/7 to host a site.

Anyway, I need to go back reading the fine print. But on the face of
it it looks very competitive, particularly if CPU charges are more
granular than EC2.

[Mike Chirico]

Do I read this correctly? There is a reduction and it is as follows?

> CPU
> current quota - 46 hours
> new quota - 6.5 hours
> % change - 86% reduction
>
> Bandwidth
> current quota - 10GB
> new quota - 1 GB
> % change - 90% reduction

Here's what's confusing. There is the following statement. Note the
phrase "applications with billing enabled will not be affected":

"These changes may also affect the fixed quotas applied to
applications without billing enabled. Fixed quotas for applications
with billing enabled will not be affected."
Reference: http://code.google.com/appengine/docs/quotas.html#Free_Changes

Specifically, does this mean I can sign up for billing, but if my
application does NOT go over 46 hours, then, I pay nothing? It sounds
like I'll get to keep the old, higher quota values, after May 25th,
2009, if I sign up for billing. True, I would run the risk of being
billed for anything above 46 hours; but, I would get to keep the old
quota.

Regards,

Mike Chirico

[Wooble]

On Feb 25, 11:11 am, mchirico <mchir...@gmail.com> wrote:
> Here's what's confusing. There is the following statement. Note the
> phrase "applications with billing enabled will not be affected":
>
> "These changes may also affect the fixed quotas applied to
> applications without billing enabled. Fixed quotas for applications
> with billing enabled will not be affected."
> Reference:http://code.google.com/appengine/docs/quotas.html#Free_Changes

I believe this is talking about just the "fixed" quotas, which is
contrasted with "adjustable" quotas. All 3 items mentioned are
adjustable, so we can assume that these change for accounts with
billing enabled.

[B.J.]

This change could lead to significant cost changes, too:

46 hours (current) - 6.5 hours (new) = 39.5 hours/day difference *
$0.10/hour * 365 days/year = $1441.75/year

10 GB (current) - 1GB (new) = 9GB/day difference * .10 * 365 = $328/
year

These are "worst case" scenarios in many ways. Still, worth
considering.

[Brett Slatkin]

Hi Mike,

> Here's what's confusing. There is the following statement. Note the
> phrase "applications with billing enabled will not be affected":
>
> "These changes may also affect the fixed quotas applied to
> applications without billing enabled. Fixed quotas for applications
> with billing enabled will not be affected."
> Reference: http://code.google.com/appengine/docs/quotas.html#Free_Changes
>
> Specifically, does this mean I can sign up for billing, but if my
> application does NOT go over 46 hours, then, I pay nothing? It sounds
> like I'll get to keep the old, higher quota values, after May 25th,
> 2009, if I sign up for billing. True, I would run the risk of being
> billed for anything above 46 hours; but, I would get to keep the old
> quota.

90 days after billing launch the free daily quotas for all
applications will change, as outlined here:
http://code.google.com/appengine/docs/quotas.html#Free_Changes

Until then, if you sign up for billing you would only pay for CPU
usage above the current ~46 CPU-hour free quota. After that time, you
would pay for usage above 6.5 CPU-hours.

As part of billing launch we have raised the fixed quota limits for
free and paid applications. These changes primarily affect the
non-adjustable, non-billed quotas that you see in the right-most
column of the tables in this section (look for "Maximum rate" under
"Billing Enabled Quota"). These quotas are mainly rate limits on
maximum usage of a particular API or resource in a short time span.

http://code.google.com/appengine/docs/quotas.html#Resources

We plan that the fixed quotas will remain at the current level after
the new free quotas have been rolled out in 90 days. We realize the
documentation on this quotas page can be difficult to understand so
we're working to make it more clear. Let me know if you have any more
questions.

-Brett

[Brett Slatkin]

On Wed, Feb 25, 2009 at 8:09 AM, peterk <peter...@gmail.com> wrote:
>
> Maybe I'm misunderstanding you, but since when was it free to run a
> EC2 instance?
>
> EC2 Small Instance per CPU Hour is $0.10 (same as GAE's cost 'per CPU
> hour')
> EC2 outgoing bandwidth is $0.17 per GB for the first Terabyte out. S3
> outgoing bandwidth is the same cost for the first 10TB out. GAE is
> $0.12 per GB out.
> S3 storage is $0.15 per GB for the first 50TB. GAE is $0.15 per GB.
>
> I'm still reading, we may be comparing apples to oranges anyway with
> regard to CPU time. Does CPU time on GAE cost if your application
> isn't actually in use, or do you pay literally just for the cumulative
> number of seconds your app spends actually executing? On EC2 you have
> to keep an instance running 24/7 to host a site.
>
> Anyway, I need to go back reading the fine print. But on the face of
> it it looks very competitive, particularly if CPU charges are more
> granular than EC2.

Indeed, App Engine's CPU accounting is extremely granular. We measure
the amount of CPU used for every request individually and add that to
our overall accounting. Your app does not consume resources and you do
not pay for anything if your application is not in active use!

For example, if your application uses 360 seconds of CPU time beyond
the free quota limits on a single day, you would only be charged for
one cent of usage ($0.10/CPU-hour * 360 = $0.01).

-Brett