My app uses MS with python 2.5 and initially I was trying to maximize cold 
start performance as cheaply as possible by setting min instances = 1 and 
max = 2, but I was getting huge latencies due to requests getting routed to 
cold starts which then led to over quota errors. I then set max instances = 
1 and performance more than doubled and no more over quota problems... 
weird. It seems having just one always hot instance bypasses the 
scheduler's odd behavior.

I think GAE is great if you have a fairly large budget but it just isn't a 
viable platform for low traffic (and low budget) applications given the 
unpredictable performance and almost zero support. The idea of 
automatically spinning up instances to handle traffic spikes is great, but 
in practice it seems to be really expensive and leads to spurious latency 
issues. I've been using GAE since 2009 and it's been really great to use, 
especially when it was free, but now I'm thinking an EC2 instance with 
Appscale might not be such a bad thing and possibly less of a hassle in the 
long run... I suspect Google would rather see apps like mine go away anyway 
since they don't generate much income.



On Thursday, March 15, 2012 9:41:53 AM UTC-7, prashant wrote:
>
> Almost identical setup and observations. 
>
> In addition,
>
> We are observing that one instance with very few warmups (spin offs of new 
> instances) can be achieved with *min instances = auto and max = 1. *Our 
> traffic is not that much ~ 1 QPS with mostly < 1 s latency
>
>
> Thanks
> Prashant
>

-- 
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/-/2OpVOJQxXPUJ.
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