[asterisk-dev] rtp scalability improvement...

[SteveK]

On Mar 19, 2006, at 5:25 PM, Anton wrote:

> On 20 March 2006 03:15, alex at pilosoft.com wrote:
>
>
[...]
> same. BUT in smaller scale. OR with more servers at once.
> Which is not good.

Why is that a problem?

It seems to me that while there may be scalability improvements that  
can be made to improve particular aspects of performance on a given  
system, how much can you expect this gain to be?  2x?  3x? 10x?

The only way to scale to very large installations is to address  
horizontal scaling, and see why more servers is "not good", and make  
it "good".  Ensure that it is easy to make clusters of asterisk  
servers, where it's easy to bring additional nodes into a cluster,  
bring nodes out, and take over from failed nodes.   Even in the  
present state of asterisk scalability, how much does the cost of the  
hardware running asterisk figure into the cost of most deployments?   
Is it significant?  It doesn't seem to be a major player when  
asterisk is used as a softswitch, where the station costs (at $70- 
$400 per station) dominate.

-SteveK