I've been going to some sites located on GAE and that i found these phones be very slow. Virtually them all take more time than normal to load.

Time: (in seconds) [ YSlow ]

9.9 giftag.com

3.1 hotskills.internet

1.9 jeeyo.internet

1.5 appspot.com

One thing Application Engine Cloud is simply too slow, Bigtable is simply too slow ... or what?

You are while using YSlow wordpress plugin to measure this, and YSlow informs you why the website is slow (the cunning title may be the clue). For instance, within the situation of gifttag.com, YSlow reviews that:

This site has 9 exterior Javascript scripts. Try mixing them into one. This site has 3 exterior stylesheets. Try mixing them into one. This site has 13 exterior background images. Try mixing all of them with CSS sprites.

Therefore it is have an 'E' grade for your. That will get rid of the perceived load performance from the site.

None of the has anything related to appengine.

YSlow is not related to the rate from the web application about the server side becasue it is a totally client side speed measurement (css, javascript, browser rendering, image loading, etc). But on the other hand, I've heard that the application might be slow on Application Engine if does not cash hits and traffic. This will make the Application Engine to not cache the python runtime atmosphere (have cold start), which means this could make factor in performance of programs with low traffic.

GAE's data access is incorporated in the order of seconds in comparison to some database that is measured in milliseconds. The main difference is the fact that BigTable scales towards the an incredible number of concurrent access because of the natural isolation degree of Read Uncommitted and also the relaxed consistency.

No RDBMS can compute with this but still give consistency guarantees. To tell the truth, you do not actually want to because for many programs you would like strong guarantees over scalability.