AppEngine Pricing

[Rüdiger Jungbeck]

We currently have a Python App which basically does nothing more than receive 4000 request a day, and in halve of this requests decides to lookup a single key_name in datastore and update the data record (350 bytes) or insert it.
(NB: The app is able to do a lot more thing, but we are currently only using the other functionality only for very few requests (< 200)  in testing).

Running this app with "always on" (because else we had lots of timeout errors during app loading) resulted in 6 concurrent instances so would cost 5* 57,60=288 USD per month with new pricing. Current cost is 9 USD per month.

288 USD would buy 57 reserved Amazon EC2 Micro instance (each of which has 3 times the amount of RAM AppEngine has) or  13 reserved Amazon EC2 Small instances (each of which had 8 the amount of RAM AppEngine has). I would think that 2 EC2 instances of any type should be enough to handle a lot more  load and have redundancy)

Do you think that the  pricing is correct?
Do you think that your scheduler is running correct?

[Gregory D'alesandre]

Hi Rüdiger,

This is a common scenario and in this case the comparison bills are a bit deceptive in terms of what you'll find once the pricing comes into effect (it is an issue we are currently working on resolving).  The scheduler has been tuned for the new pricing model and so does not take advantage of the resident instances that always-on creates.  Unfortunately these resident instances are being shown on the comparison bills.  Once the new pricing is out always-on will no longer exist and instead you'll be able to set a minimum number of idle instances to keep running.  So, in your case, in order to both understand what your bill will be as well as manage it, there are 2 things to do:

- Subtract 72 instance hours from your bill

- Set max idle instances as it will limit the number of idle instance that you will be charged for (and your traffic seems low enough that it seems unlikely it will impact performance)

- Consider using reserved instance hours that are $0.025 rather than $0.04 and it will reduce the amount you spend on the instances that are left

As mentioned we are looking into fixing the comparison bills to avoid this issue, thanks for highlighting it,

Greg

--
You received this message because you are subscribed to the Google Groups "Google App Engine" group.
To view this discussion on the web visit https://groups.google.com/d/msg/google-appengine/-/oRQHZiqa1d8J.
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.

[Stephen]

On Tue, Sep 6, 2011 at 9:14 AM, Gregory D'alesandre <gr...@google.com> wrote:
>
> - Consider using reserved instance hours that are $0.025 rather than $0.04

To predict your spend under the new price scheme you need to know that
on-demand instance hours are $0.08 and reserved instance hours are
$0.05.

http://www.google.com/enterprise/cloud/appengine/pricing.html