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.
Its been a busy day so I haven't been able to answer all the questions in the group, but I will do so tomorrow morning (well, morning my time) as well as cull for the questions that seemed concerning and people wanted answered.
I'll make sure to include all of these, but for now, some answers below:
On Thu, May 12, 2011 at 12:14 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?
The smallest granularity will be 15 minutes, but part of the scheduler change is to ensure we don't start instances to serve 1 request.
> 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)?
Yes, you will be able to tune the scheduler, specific controls are still being worked on, all of the ones you mentioned seem like reasonable ones.
> 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?
We are still evaluating various methods to potentially handle Python concurrency, so don't yet have an answer to whether it will require code changes. In terms of the number of requests per second that can be handled depends primarily on your code. We've done stats on what the max requests that can be handled and I'll post those numbers.
> 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?
Soon. :)
> 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.
The current plan is that all frontend instances have the same CPU and memory limits but there is the potential to later allow customers to configure the instance size for frontends.
I'll gather all of these into a document tomorrow.
> -- > 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-appengine@googlegroups.com. > To unsubscribe from this group, send email to > google-appengine+unsubscribe@googlegroups.com. > For more options, visit this group at > http://groups.google.com/group/google-appengine?hl=en.
>> The smallest granularity will be 15 minutes, but part of the scheduler
change is to ensure we don't start instances to serve 1 request.
This is useful to know so thank you for letting us know this. But it's disappointing to say the least. We're going from millisecond granularity with CPU-hours to chunks up to 15 minutes depending on how many requests you get out of a new instance.
Anyway, in the FAQ, I'd like a transparent, honest answer about why the switch from CPU-hours to instance-hours (not a vague 'based on the value of the service', 'based on feedback'), and a comprehensive outline of the ramifications.
On Thu, May 12, 2011 at 1:47 PM, peterk <peter.ke...@gmail.com> wrote: > >> The smallest granularity will be 15 minutes, but part of the scheduler > change is to ensure we don't start instances to serve 1 request.
> This is useful to know so thank you for letting us know this. But it's > disappointing to say the least. We're going from millisecond granularity > with CPU-hours to chunks up to 15 minutes depending on how many requests you > get out of a new instance.
> Anyway, in the FAQ, I'd like a transparent, honest answer about why the
> switch from CPU-hours to instance-hours (not a vague 'based on the value of
> the service', 'based on feedback'), and a comprehensive outline of the
> ramifications.
"In its three short year history, Google App Engine has evolved from
its grass roots developer origins to a technology used more and more
by global businesses. This session will review App Engine’s history
and explain how it will be continuing to evolve to serve an increasing
Enterprise audience."
Well, if that's the motivation, if it's coming from 'enterprise', then to rephrase, I'd like to see that explained and outlined fully in this new FAQ, and what from Google's point of view this orientation will mean for 'grass roots' developers.
I mean, they talked so much about how they wanted to support the people building the next Facebook - that kind of developer. 'That kind of developer' is people working out of dorms and garages and at home and in small teams, not suits in established companies. The next facebook won't locate on AppEngine if 'grass roots' devs are to be abandoned in favour of extracting as much values as possible out of Best Buy.
> 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.
> 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 to Kennet for the right subset of first questions.
Thanks to Greg for his answers, especially that about scheduler tuning by user: it will put a cap to skyrocketing bills with a fair performance cap instead of out-of-quotas service disruption.
1) What is the justification behind the extremely high price for additional instances?
Here is a comparison between a GAE Frontend Python Instance and a small Rackspace Cloud Server:
GAE Frontend Python Instance: ------------------------------------------------- Monthly price: $57.60 ($36.00 if reserved) RAM: 128 MB memory cap CPU: 600 MHz limit Capabilities: Limited by the Python sandbox. No native code execution. Single-threaded right now, possible future multi-threading obstructed by the GIL.
Rackspace Cloud Server 256 MB: ------------------------------------------------- Monthly price: $10.80 RAM: 256 MB fixed CPU: Guaranteed proportional minimum; Free CPU bursting (I'm using the full power of 4 cores 99% of the time) Capabilities: Full-featured Linux box, can do whatever you want on it.
On Thu, May 12, 2011 at 10:14 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
> -- > 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-appengine@googlegroups.com. > To unsubscribe from this group, send email to > google-appengine+unsubscribe@googlegroups.com. > For more options, visit this group at > http://groups.google.com/group/google-appengine?hl=en.
> 1) What is the justification behind the extremely high price for additional > instances?
> Here is a comparison between a GAE Frontend Python Instance and a small > Rackspace Cloud Server:
> GAE Frontend Python Instance: > ------------------------------------------------- > Monthly price: $57.60 ($36.00 if reserved) > RAM: 128 MB memory cap > CPU: 600 MHz limit > Capabilities: Limited by the Python sandbox. No native code execution. > Single-threaded right now, possible future multi-threading obstructed by the > GIL.
> Rackspace Cloud Server 256 MB: > ------------------------------------------------- > Monthly price: $10.80 > RAM: 256 MB fixed > CPU: Guaranteed proportional minimum; Free CPU bursting (I'm using the full > power of 4 cores 99% of the time) > Capabilities: Full-featured Linux box, can do whatever you want on it.
> On Thu, May 12, 2011 at 10:14 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
>> -- >> 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-appengine@googlegroups.com. >> To unsubscribe from this group, send email to >> google-appengine+unsubscribe@googlegroups.com. >> For more options, visit this group at >> http://groups.google.com/group/google-appengine?hl=en.
> -- > 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-appengine@googlegroups.com. > To unsubscribe from this group, send email to > google-appengine+unsubscribe@googlegroups.com. > For more options, visit this group at > http://groups.google.com/group/google-appengine?hl=en.
Yes, they are different kinds of beasts. But the resources needed to provide them are the same. And still, the GAE instance costs more than 10 times higher than a Rackspace VPS consuming the same resources. This is an order of magnitude difference.
Being "managed" can justify a higher price, but not 10 times (the "management" in this case is 100% automated).
On Thu, May 12, 2011 at 7:25 PM, Nischal Shetty <nischalshett...@gmail.com>wrote:
> You cannot just compare GAE with Rackspace simply because GAE is unmanaged, > we don't do a thing.
> Though I would still maintain that the prices are indeed high.
> On 12 May 2011 21:46, Peter Petrov <onest...@gmail.com> wrote:
>> My single question for Greg's FAQ:
>> 1) What is the justification behind the extremely high price for >> additional instances?
>> Here is a comparison between a GAE Frontend Python Instance and a small >> Rackspace Cloud Server:
>> GAE Frontend Python Instance: >> ------------------------------------------------- >> Monthly price: $57.60 ($36.00 if reserved) >> RAM: 128 MB memory cap >> CPU: 600 MHz limit >> Capabilities: Limited by the Python sandbox. No native code execution. >> Single-threaded right now, possible future multi-threading obstructed by the >> GIL.
>> Rackspace Cloud Server 256 MB: >> ------------------------------------------------- >> Monthly price: $10.80 >> RAM: 256 MB fixed >> CPU: Guaranteed proportional minimum; Free CPU bursting (I'm using the >> full power of 4 cores 99% of the time) >> Capabilities: Full-featured Linux box, can do whatever you want on it.
>> On Thu, May 12, 2011 at 10:14 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
>>> -- >>> 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-appengine@googlegroups.com. >>> To unsubscribe from this group, send email to >>> google-appengine+unsubscribe@googlegroups.com. >>> For more options, visit this group at >>> http://groups.google.com/group/google-appengine?hl=en.
>> -- >> 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-appengine@googlegroups.com. >> To unsubscribe from this group, send email to >> google-appengine+unsubscribe@googlegroups.com. >> For more options, visit this group at >> http://groups.google.com/group/google-appengine?hl=en.
> -- > 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-appengine@googlegroups.com. > To unsubscribe from this group, send email to > google-appengine+unsubscribe@googlegroups.com. > For more options, visit this group at > http://groups.google.com/group/google-appengine?hl=en.
I am curious how "Always On" will be handled? Right now for $8.40 I
get 3 instances always on. Can I still have this? If so will I be
billed 3 * .05 * 24 * 30, or will I only be billed for the actual time
the instances are "used" ?
On May 12, 11:31 am, Peter Petrov <onest...@gmail.com> wrote:
> Yes, they are different kinds of beasts. But the resources needed to provide
> them are the same. And still, the GAE instance costs more than 10 times
> higher than a Rackspace VPS consuming the same resources. This is an order
> of magnitude difference.
> Being "managed" can justify a higher price, but not 10 times (the
> "management" in this case is 100% automated).
> On Thu, May 12, 2011 at 7:25 PM, Nischal Shetty
> <nischalshett...@gmail.com>wrote:
> > You cannot just compare GAE with Rackspace simply because GAE is unmanaged,
> > we don't do a thing.
> > Though I would still maintain that the prices are indeed high.
> > On 12 May 2011 21:46, Peter Petrov <onest...@gmail.com> wrote:
> >> My single question for Greg's FAQ:
> >> 1) What is the justification behind the extremely high price for
> >> additional instances?
> >> Here is a comparison between a GAE Frontend Python Instance and a small
> >> Rackspace Cloud Server:
> >> GAE Frontend Python Instance:
> >> -------------------------------------------------
> >> Monthly price: $57.60 ($36.00 if reserved)
> >> RAM: 128 MB memory cap
> >> CPU: 600 MHz limit
> >> Capabilities: Limited by the Python sandbox. No native code execution.
> >> Single-threaded right now, possible future multi-threading obstructed by the
> >> GIL.
> >> Rackspace Cloud Server 256 MB:
> >> -------------------------------------------------
> >> Monthly price: $10.80
> >> RAM: 256 MB fixed
> >> CPU: Guaranteed proportional minimum; Free CPU bursting (I'm using the
> >> full power of 4 cores 99% of the time)
> >> Capabilities: Full-featured Linux box, can do whatever you want on it.
> >> On Thu, May 12, 2011 at 10:14 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
> >>> --
> >>> 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-appengine@googlegroups.com.
> >>> To unsubscribe from this group, send email to
> >>> google-appengine+unsubscribe@googlegroups.com.
> >>> For more options, visit this group at
> >>>http://groups.google.com/group/google-appengine?hl=en.
> >> --
> >> 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-appengine@googlegroups.com.
> >> To unsubscribe from this group, send email to
> >> google-appengine+unsubscribe@googlegroups.com.
> >> For more options, visit this group at
> >>http://groups.google.com/group/google-appengine?hl=en.
> > --
> > 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-appengine@googlegroups.com.
> > To unsubscribe from this group, send email to
> > google-appengine+unsubscribe@googlegroups.com.
> > For more options, visit this group at
> >http://groups.google.com/group/google-appengine?hl=en.
Also currently the python instance can only serve 1 request at a time,
so 128mb ram doesn't matter much
My question for FAQ is similiar too:
Is there a possibility of Python not getting a solution for
concurrency and us paying high amounts for this reason?
On May 12, 7:16 pm, Peter Petrov <onest...@gmail.com> wrote:
> 1) What is the justification behind the extremely high price for additional
> instances?
> Here is a comparison between a GAE Frontend Python Instance and a small
> Rackspace Cloud Server:
> GAE Frontend Python Instance:
> -------------------------------------------------
> Monthly price: $57.60 ($36.00 if reserved)
> RAM: 128 MB memory cap
> CPU: 600 MHz limit
> Capabilities: Limited by the Python sandbox. No native code execution.
> Single-threaded right now, possible future multi-threading obstructed by the
> GIL.
> Rackspace Cloud Server 256 MB:
> -------------------------------------------------
> Monthly price: $10.80
> RAM: 256 MB fixed
> CPU: Guaranteed proportional minimum; Free CPU bursting (I'm using the full
> power of 4 cores 99% of the time)
> Capabilities: Full-featured Linux box, can do whatever you want on it.
> On Thu, May 12, 2011 at 10:14 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
> > --
> > 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-appengine@googlegroups.com.
> > To unsubscribe from this group, send email to
> > google-appengine+unsubscribe@googlegroups.com.
> > For more options, visit this group at
> >http://groups.google.com/group/google-appengine?hl=en.
My FAQ question (which went unanswered by Google in an earlier thread, I assume because of the sheer volume of questions right now):
> High Replication Datastore as default: ... encouraging everybody to begin plans to migrate…. > Mail API: ...we’ve reduced the number of free recipients per day from 2000 to 100 for newly created applications...
So if we migrate to HR Datastore, does that mean we have a "newly created" application, and will get dinged by this new, rather low, free quota for email? Could you grandfather in migrated apps at the old 2000 limit?
no FAQ from you yet... so in the meanwhile here it is a list of links you (and Ikai, Nick, Justin, etc.) should visit to know what gae users said in the last couple of days about pricing news.
I agree with Kaan. I believe Google needs to take an official stance for all of us python developers who have invested heavily in learning how to optimize GAE for python development.
1) Would you please state weather GAE team will be releasing an equivalent multi-safe-threading python environment (without the GIL) similar to java, or have a different pricing model for python all together.
2) If GAE team is not going to change pricing for python devs, or has no plans to release a multi-safe-threading python environment, they could at least be kind enough to not keep this a secret and let us know now, so we can each make an informed decision on how we should proceed with our projects. Keeping us python developers in the dark is not a very pleasing thing to experience, especially with so much effort invested on our parts. I lost sleep last night not knowing weather the past year of my life has been wasted learning python GAE development for my startup.
3) And weather or not we would be better off using GAE Java. Since I understand App Engine, it would be easy to make a switch, but we must not be kept in the dark. Please make a statement answering weather Java will be the better option for those without extremely deep pockets launching startups.
Still working on pulling all the information together (rather than answering the questions one off). Thanks for the consolidated list of comments that you found.
Greg
On Thu, May 12, 2011 at 2:46 PM, Vanni Totaro <vanni.tot...@gmail.com>wrote:
> no FAQ from you yet... > so in the meanwhile here it is a list of links you (and Ikai, Nick, Justin, > etc.) should visit to know what gae users said in the last couple of days > about pricing news.
> -- > 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-appengine@googlegroups.com. > To unsubscribe from this group, send email to > google-appengine+unsubscribe@googlegroups.com. > For more options, visit this group at > http://groups.google.com/group/google-appengine?hl=en.
The new pricing announcement has put instance performance into the
spotlight, and GAE comes out looking very bad compared to other cloud
solutions*. My question is how can Google justify roughly equivalent
pricing for a product that is 15-30 times less powerful? I'm happy to
pay a margin for GAE's scalability and platform management, but I'd
suggest that margin should be more like 50%, than 1500% to 3000%.
*An EC2 instance running Drupal can handle 45-60 requests a second.
Because it's single-threaded, a GAE instance running Django can only
handle 2-3 requests a second.
I recommend that Google figures out how to do concurrent requests per
instance before adopting an instance-hour pricing model, or else
discounts the floated instance-hour price by at least 90%.
Hacker News isn't the fount of wisdom it once was (if its mythology is to be believed), but here's the money quote from a Hacker News thread:
I guess I felt that my implicit feelings on App Engine were something like, "Hey hackers! You should totally rewrite your apps for our Google systems that are a lot more efficient than other systems. Yeah, there are some annoying restrictions that you'll have to get used to and are totally a pain for some things. Still, out service is cheap for loads of usage and really cheap even after that so you're spending a little programmer time for no-hassle-scaling and cheaper hosting than anything you can get!" However, they've consistently lowered the free usage tier to being a fraction of what it once was, they're now charging a ton more with their instance-hour model compared to the old CPU based model, a bit of the reliability/scaling sheen has worn off as it's had problems, other competitors have been aggressively entering this space, and you still have to alter your apps specifically for their architecture. I'm not saying that App Engine doesn't have value, just that it feels very different.
> no FAQ from you yet... > so in the meanwhile here it is a list of links you (and Ikai, Nick, Justin, > etc.) should visit to know what gae users said in the last couple of days > about pricing news.
> -- > 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-appengine@googlegroups.com. > To unsubscribe from this group, send email to > google-appengine+unsubscribe@googlegroups.com. > For more options, visit this group at > http://groups.google.com/group/google-appengine?hl=en.
Here is an idea for the FAQ: How will versions work with the new pricing?
Currently, when switching default versions, old and new instances are spun up simultaneously for a while. Will switching default versions thus incur additional fees? How does the 15-minute granularity on instance charges work here? Will it make a difference if you deploy on top of the current version instead of incrementing the version?
If the "version trick" is used to run, say, one version as Python and one version as Go within the same app, does that count as two billable instances? Does it count even if they both use less than 20MB of memory?
how does the instance work? If I have multiple modules in one app, how many instance is that? Or does it just mean that for as long as my app is a singleton, no matter the amount of modules I have, it is still a single instance.
At what point is Google distributing my app unto more than one server, which will definitely cause more than one instance? Is it possible to control the instances created? I mean like force App Engine to manage resources within a fixed number of instances instead of trying to be so perfect (unmanaged) that it costs me lots of money.
Greg, I'll be looking forward to your answers on these FAQs because I'll be giving a talk on Google App Engine at upcoming CloudCamp event.
Regards
On 5/12/11, 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
> -- > 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-appengine@googlegroups.com. > To unsubscribe from this group, send email to > google-appengine+unsubscribe@googlegroups.com. > For more options, visit this group at > http://groups.google.com/group/google-appengine?hl=en.
