> Being that an SLA typically just means I will receive a refund for
> down time, in other words a fraction of $9/month and $.08/hour, I'd
> like to propose an alternative model for people to have paid apps
> without an SLA.  I am definitely interested in uptime, however a
> refund of my $.08/hour for 30 minutes downtime doesn't get me too
> excited.

> On May 12, 2:14 am, Kenneth <kennet...@aladdinschools.com> wrote:

> > Greg mentioned he was putting together an FAQ so let's help him out!

> > If you're going to answer this just put in your question into a single line,
> > let's not try and answer them here or give opinions, there's plenty of other
> > threads for that. I do understand that Google doesn't have answers to some
> > these.

> > Here's my list:

> > 1) What is the time granularity of the instance pricing?  ie if I have an
> > instance up for 5 minutes, what am I charged, $0.08 / 60*5?
> > 2) Will I be able to tune the scheduler myself, ie set it to performance or
> > low cost,  Will I be able to limit the min or max number of instances
> > created (with the obvious impact on user experience)?
> > 3) Python concurrency, will this require any code changes, do you have any
> > estimates based on your testing of the number of well behaved requests per
> > second a single instance will be able to handle for a given framework?
> > 4) Database charges, when can you give us more details over what Max gave in
> > the other thread, are you charging for deletes, what do you expect the ratio
> > to be between the new pricing metric and the Datastore API calls metric we
> > have today?
> > 5) Will you be charging differently for instances that use different amounts
> > of memory, since this seems to be the cost that you're going after that
> > isn't charged for in the current model.

> > Thanks,
> > Kenneth