1.2 Ghz Intel single core? What about time sharing? I suspect Google 
squeezes in a lot of instances in each physical server, so that with time 
sharing of the CPU the instances in reality get a lot less than that when 
the load is high, and judging by how often instances are swapped in and out 
of servers, it means that there are way too few physical servers to handle 
the practical load.

Compare with a virtual memory. If there is too little physical memory, the 
page swapping algorithm, no matter how good it is, will have to do a LOT of 
swapping of memory pages. And what do we see the GAE instance scheduler do? 
A LOT of instance swapping.

Not only that, we as customers have to pay a large part of that page 
swapping, plus poor performance when the traffic to the application is low. 
AND, much if not most of the actual specs for the instances is kept hidden 
from us customers. So it's very hard to tell what we actually will be paying 
for, plus Google can change the specs at any time without adjusting the 
price model. It's opaque as hell.

-- 
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/-/TkMn50YXrhoJ.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/google-appengine?hl=en.

Reply via email to