New pricing will cost me over 500 $ monthly for my three sites... I am floored.

[Santiago Lema]

I have just seen the estimated for the new pricing model and to be
honest: I am horrified. I have spent a lot of time learning python and
the AppEngine platform because it seemed like a cheap and trustable
solution. I agree it was almost too cheap but a 17x increase is really
a bit much for me.

My websites extensively use memcache to store stats. Granted I get a
lot of requests per day (about 1 million) but I am still baffled by
the price increase. All my pages do is get 7 parameters from a get
request and store them in memcache, eventually writing to DataStore
every 15 minutes. I spent a lot of time optimizing for memcache
because DataStore writes where expensive...

My stats website will now cost me 7.05 $ per day as opposed to 0.41 $
before. I am afraid I am not going to pay 211 $ monthly (instead of 12
$). just to have my daily stats. I am just glad my business doesn't
depend on my stats and I'll just trash them. All I will have lost are
countless hours...

My company's website (http://smallte.ch hosted on
websmalltech.appspot.com) will now cost me about 144$ monthly. It's a
website with rather low traffic with just one mini webpage that gets
requested a lot (from memcache) by my iPhone apps. It was basically
free before if it hadn't been for the 9$ "always on" option I had.

Prior to switching to AppEngine this website was handled by a 19.99 €
server from OVH.com and it worked well. I just changed for the
security of being hosted on Google and to get rid of Linux
maintenance.

My smallte.ch website has 6 instances (3 resident doing nothing). How
the can six instances at 29 mb each (totaling 174 mb of memory usage)
cost so much money ?

How can these instances cost 140 $/month? https://skitch.com/smalltech/fipfa/instances-smallte.ch

I mean seriously for the 100 Euros per month I can get this machine:
http://www.ovh.com/fr/serveurs_dedies/eg_ssd.xml.
Yes that's 24 Gb, not 0.174 gb.

Sure it's not in the cloud but man aren't we being extorted here ? Am
I the only one seeing such an increase ?

[Anders]

Something must be wrong with the new pricing model. When I look at the Usage Report for 2011-08-26, then it says total: $0.30. With the new pricing model it says: $3.66.

With the same amount of usage for one month it will be around $110. For one application with tiny traffic! And that is only the price for frontend instance hours.

Is Google's motto: "Don't be evil, be very evil." ?

Or is this Google's way of saying: "We are killing Google App Engine, but will support it for another two years just to not piss you off too much." ;-)

[blackpawn]

I'm seeing an 8X increase with the new pricing model.  Combined with poor performance and reliability the last few months I'm thinking of spending my time moving to AWS instead of investing more time trying to optimize.

[RollingCircle]

Ditto - although the figure I'm looking at is moving from well within the existing free quotas to $120/month - for a non-revenue-generating application.

I suppose you could argue that as the service has been running for a couple of years, Google have already donated what they now value to be $2880 dollars of resource, but looking at the actual resources consumed, it is difficult to see how it has cost them that much.

Will still put the effort in to see how efficient things can be made in light of the new scheme, but would have appreciated maybe a couple of months rather than a couple of weeks to do so.

I suspect I'll end up spending $9 a month for a paid app that simply serves redirects to the app re-hosted on AWS until enough clients update bookmarks, and then remove billing and let it whither on the vine.

[Anders]

I took a look at the Amazon prices just now. It looks quite expensive too. And the same dubious and opaque cost per instance concept. Both Google and Amazon can cram in 1000 instances in each physical server! Talk about hideous profit margin.

The cost for GAE will be low if the instances is limited to only one. But what will the performance be then? And what about traffic spikes and/or traffic growth?

[blitz...@hotmail.co.uk]

I'm going from 6 cent per day to 3 dollars a day. A 50x increase.  $90 a month.  Wow google, just wow.  And that's with the 50% discount. Google really wants to kill this thing.

[Santiago Lema]

Update: I disabled "always on", set minimal latency to 60ms and now I
seem to have much less instances.

My three websites now seem to have each two instances. I am not sure
if my changes have anything to do with this.

My iPhone stats website (by *far* the one with more traffic) now seems
to be stable with two instances each getting a good share of requests.

It seems smaller websites are penalized. My "big" stats websites has
had about 23k requests in the last 37 minutes and they required two
instances with each an average 50ms latency. Compare that to my public
website which only got 280 requests in the last 20 minutes. It also
has two instances.

So each of these would cost 36$/month in instances although clearly
the stats has about 78x more traffic.

I don't really understand why my public website needs a second
instance though. The second one has got very few requests and has a
0.0 latency. First instance isn't hammered either with its 28ms
latency.

[Santiago Lema]

Actually I spoke/wrote too fast. My smallte.ch website now has closed
its useless second instance.

Also my 36$/month numbers were wrong since obviously the first
instance is free.

If it remains like that it could be tolerable but I am still baffled
by the numbers I saw in my history.

[psm]

Have your reliability issues been with HRD or M/S apps?

[Raymond C.]

Each EC2 instance can serve a lot more traffic than AppEngine instance (2 for python at a time at this moment).  I would say AppEngine instances are a lot more expensive (not to mention much more limitation).

[Ikai Lan (Google)]

Are you running Python or Java instances? A lot of the costs come from the fact that a single Python instance serves one request. We do expect there to be some price increase, but sometimes when someone says "20x" increase, it's not clear to us whether that's because you went from $1 -> $20 ($9 minimum base price) or because of the instance based pricing.

--

Ikai Lan 
Developer Programs Engineer, Google App Engine

plus.ikailan.com | twitter.com/ikai

On Wed, Aug 31, 2011 at 5:57 PM, Raymond C. <wind...@gmail.com> wrote:

Each EC2 instance can serve a lot more traffic than AppEngine instance (2 for python at a time at this moment).  I would say AppEngine instances are a lot more expensive (not to mention much more limitation).

--
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/-/98TgfWdWb-IJ.

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.

[John Wheeler]

What is with the front-end instance hours? Are the comparison reports broken or something? In my case, I am going from 30 cents to $5.00 a day. That can't be right. You guys would lose nearly all your customers.

[Santiago Lema]

When I started learning about AppEngine Python was obviously the
preferred language (features were being added to Python first)... so I
learned Python. Now it seems I made the wrong choice (especially since
I was already comfortable with Java).

Instance-based pricing is really killing me here. I don't know if
Python 2.7 will improve this but if I have like 2 weeks to implement
it I think I'll just disable all the non-vital services of my site (my
stats are not vital and they are the ones using the most instances).

@Ikan Lan:
In my case it's a 17x increase over all three sites based on the
Billing history for August 26 on smalltechads.appspot.com. The price I
paid was 0.41$ for this day but under the new billing it'll be 7.05 $,
effectively a 17x increase in price. This is similar for my other
websites.

I can understand that paying 12$ monthly wasn't much but I surely
didn't expect that this stats app would cost me 211$ to keep...

[Anders]

I saw a ten times increase for my paid application. In reality that will be much lower if I limit the number of instances, so we will see what the actual cost will be.

[Federico Elles]

I run 3x3Links.com and it will go up from $0.01 daily to up to $4.25.

Hope to reduce this using some configuration hints from this thread.

[moxa]

As I'm reading more and more about people's new pricing problems...
it seems to me there's either some calculation error on Google's
managers behalf
or they are slowly killing the product.

It seems reasonable that the new prices are mistaken by order of
magnitude.
For example 0.41$ => 7.05$ (better 0.71$)
0.30$ => 3.66$ (better 0.37$)
0.01$ => 4.25$ (better 0.43$)
0.30$ => 5.00$ (better 0.50$)

If small devs are penalized so the prices of big remain smaller then
logically
small devs sooner or later will flee and new pricing model would have
to be
revised once again ruining the credibility of the product and the
company
once again.

Another point is that the instances GAE offers are super small...
100MB RAM
for frontend instance should not cost 57.6$. It's not realistic at
all. 5.76$ sounds
more reasonable.

Just my 2 cents.

On Sep 1, 12:00 pm, Federico Elles <flas...@googlemail.com> wrote:
> I run 3x3Links.com and it will go up from $0.01 daily to up to $4.25.
> Hope to reduce this using some configuration hints from this thread.
>

>  gae1.jpg
> 68KViewDownload
>
>  gae2.jpg
> 117KViewDownload

[will]

Google's instance-hour is process-instance-hour whereas AWS's instance-
hour is machine-instance-hour. Every process GAE spins up count as
one instance-hour while you got the whole machine for a hour in AWS
and you can spin up as many processes as the machine instance can
handle.

When multiple web requests come in, GAE spins up multiple processes
(Python) to serve the requests. All those count as multiple
instances. The killer thing is most request handlers are not CPU
bound and just idling, waiting for network transfer from browser/
datastore/memcache/etc. All those idling times still count toward the
GAE instance-hour. That's why there's a huge increase when switching
from pure CPU-hour accounting.

In this case GAE's process-instance-hour ($.08/hr) is much more
expensive than AWS's machine-instance-hour ($.085/hr or $0.02/hr
micro). AWS gives you a whole spec machine which you can cram as many
processes out of it. GAE just gives you a process, which idles most
of the time for IO bound app but still charges you. Also the spec of
the GAE process-instance is pretty vague. You don't know what
capability you are buying.