Shared publicly  - 
 
Siegfried Hirsch originally shared:
 
Before dropping Google AppEngine think about Tricking the new price list of App Engine

Dropping the price by a factor of 50 is possible. A lot depends on the code and how you use App Engine for the better.

You remember +Russell Beattie rant about the new price structure of App Engine. He has taken down plusfeed.appspot.com because of the tremendous price increase and a lot of people have shared this feelings.

After some time I decided to optimize the code, cause it was obvicious how the code make better use of App Engine's features. I don't think this is always as easy as this time - see the postings of +Emlyn O'Regan who is optimizing his own code.

Find the old and new listings of plusfeed2.appspot.com (optimized with red markings) and the old plusfeed.appspot.com and compare for yourself.

I have also included a charts for the last two days to show how many instances are used to do the workload of plusfeed2.appspot.com

As you can see, plusfeed2 now uses between 2 and 4 instances - but most of the time it were just 2 instances. Compare this to the about 35 instances before and you see, where the higher costs came from.

I hope this gives other programmers some insights, that it is possible to optimize for the new price structure of App Engine. I don't say, that it is always possible, but a better algorythm or other solutions could help a lot.

App Engine is often compared to running lamp servers by yourself, but scaling and maintenance are a lot of work and if your app needs more resources App Engine just grows, where on other platforms you need to think about new servers and system admins. And sometime there are only slashdot or techcrunch/mashable peeks, that your app has to handle. So automatic scaling, no admins and much less baby sitting of your app are the main plus points for me to stay on App Engine.
28
20
Alper Çuğun's profile photoMarcin Lulek's profile photoSteve Sherrie's profile photoalang yin's profile photo
6 comments
 
the return of time-based (and use-based) billing marks a new era in coding/consulting
 
+Ahmet Alp Balkan if writing good code doesnt make sense... then i feel sorry for you. Good luck with heroku/whatever ;-)
 
Whatever the end result, this probably has served as a wake up call to a lot of people with regards to vendor lock-in not being a good idea.
Add a comment...