Yea i think so too :)
On Aug 22, 2008, at 12:38 PM, Kevin Brown wrote:
I think you meant QPS.
On Fri, Aug 22, 2008 at 3:00 AM, Chris Chabot <[EMAIL PROTECTED]>
wrote:
It depends on a whole lot of different factors so it's hard to give
an
exact number. One of the important factors is how your social api
service
interface is implemented for instance, what kind of caching schemas
your
using, etc.
However in practice I found that in my situation, just to give you
a rough
ballpark kind of number, one off the rack Dell server (single quad
core
xeon) can support some 200-300 QPM before latency starts to become
something
of an issue.
To achieve that you do need to take very good care of the caching
layers in
your social api code, make sure you run either Zend Platform
(commercial) or
APC (free) for byte code caching, and make sure everything is
generally
properly configured and tuned.
Hope that helps !
-- Chris
On Aug 22, 2008, at 8:06 AM, Daniel Ruspini wrote:
I've looked around and haven't found any load test numbers for
supporting
a
Shindig server. Meaning if I wanted to run an installation of
Shindig on
Apache (default installation), how many servers would I need to
support X
number of manifest requests?
