44 messages in com.googlegroups.google-appengine[google-appengine] Re: A really good ...
FromSent OnAttachments
Brett Morgan21 May 2008 02:51 
Ben the Indefatigable21 May 2008 04:45 
Brett Morgan21 May 2008 04:47 
Dado21 May 2008 08:41 
Dado21 May 2008 08:45 
Filip21 May 2008 08:59 
Filip21 May 2008 09:00 
Ben the Indefatigable21 May 2008 11:01 
toddhoff21 May 2008 12:01 
bowm...@gmail.com21 May 2008 12:19 
Ben the Indefatigable21 May 2008 12:25 
Ben the Indefatigable21 May 2008 12:38 
bowm...@gmail.com21 May 2008 12:48 
Filip21 May 2008 12:51 
Filip21 May 2008 12:55 
Ben the Indefatigable21 May 2008 13:50 
Brett Morgan21 May 2008 14:07 
Dado21 May 2008 15:11 
Brett Morgan21 May 2008 17:57 
Brett Morgan21 May 2008 17:58 
Brett Morgan21 May 2008 18:02 
Dado21 May 2008 19:28 
toddhoff21 May 2008 23:48 
Barry Hunter22 May 2008 00:34 
Brett Morgan22 May 2008 01:03 
Dado22 May 2008 01:50 
Brett Morgan22 May 2008 03:08 
Filip22 May 2008 04:55 
Filip22 May 2008 04:58 
toddhoff22 May 2008 08:04 
Dado22 May 2008 11:48 
Filip22 May 2008 13:40 
Brett Morgan22 May 2008 14:14 
Dado22 May 2008 15:58 
Filip22 May 2008 16:23 
bowm...@gmail.com22 May 2008 20:00 
Filip23 May 2008 03:13 
toddhoff23 May 2008 08:36 
Dado23 May 2008 08:41 
bowm...@gmail.com23 May 2008 09:24 
Filip23 May 2008 09:39 
Dado23 May 2008 12:20 
Scott23 May 2008 12:39 
Dado23 May 2008 13:18 
Subject:[google-appengine] Re: A really good summary of AppEngine data modelling ideas
From:Filip (fili@gmail.com)
Date:05/22/2008 04:23:18 PM
List:com.googlegroups.google-appengine

No, I don't have experience in data warehousing... but has anybody tested the GAE datastore performance when it reaches terabytes levels??? That's the kind of info we need from Google engineers, because I don't like to base my choice on assumptions or, even worst, faith.

Better yet, I'd sign for the ability to test it myself on that scale :-)

They seem to imply that getting a hundred records is OK. If that is the scale of joining, then you might be right, especially if joining really is really about getting a predefined set of records. If you intend to join millions of records and select on that, as is common using SQL, you might be out of luck. Again, the truth is in the middle, according to me.

You're right, but for the average web page rendering I don't foresee joining millions of records (at least in my app). However, goigh through millions of records (offline I mean, for example to create a recommendation system) is still a requirement for many web apps. We definitively need answers from Google in terms of their willingness to allow background/scheduled processes and access to map/reduce.

Couldn't agree more. It seems more than likely that background processing is coming, but a simple "yes we're working on that" would be nice. By the way, at least in my case, precalculating and creating ready to go copies of the data is exactly what that mapreduce/ background process will do. You could even call it denormalisation.

I am definitively looking forward to having access to that information!!! Let's hope it will clarify best-practices for the datastore.

Filip.

--~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups
"Google App Engine" group. To post to this group, send email to goog@googlegroups.com To unsubscribe from this group, send email to
goog@googlegroups.com For more options, visit this group at
http://groups.google.com/group/google-appengine?hl=en -~----------~----~----~----~------~----~------~--~---