No subject

users on a server, and probably more than that. The cost
per user is less than 10 cents assuming a colo price less
than $100/month. Even higher than that you're still talking
a few cents more per user.

The only problem I found scaling is the need for some type
of load balancing - and I'm taking your advice on that one
(thanks!) and looking to put Kamailio in front of an array 
of media servers. Once that's done and working properly, I 
really see no reason why we couldn't scale to tens of 
thousands of users by just adding more media servers as
we need them. :)

One of the things I am sturggling with is how to get
Kamailio to account for geographic location and divert
a west coast user to a west media server rather than east,
but I guess that shouldn't be too hard to implement 
eventually. (maybe look at the IP address - or if possible
with kam, just store an indicator in the database that will
tell it what array of servers to divert to)

Anyway.. point being - servers are cheap, colo is cheap,
and bandwidth is cheap and always getting cheaper. The
cost of proxying media are negligble in the type of
business model we run.

Again- a wholesale operation may very well find it makes
more sense not to proxy the media since they don't have
to deal with NAT and other such headaches anyway.

-- Nitzan
http://www.comparevoipproviderrates.com/