Whatever happened to intelligence in the applicattion [Was: Re: Th e Qo s PipeDream]

Fri Dec 16 16:04:17 UTC 2005

On Dec 16, 2005, at 3:49 PM, Fergie wrote:

>
> I certainly don't endorse placing _all_ of the intelligence
> in the application, but look at it this way -- if you expect
> to have a 'stupid' CPE handset rely on 'intelligence' in the
> network for voice quality, you're probably going to be disappointed.
>
> And no amount of leveraging smoke-and-mirror QoS frobs to generate
> additional revenue will help you out.

You'd have to convince them first that they need to care about QoS  
and all its gory details rather than whether it works or not at a  
price they're willing to pay.  There's a fundamental disconnect  
here. ;-)  If you have to deploy QoS to make a service happen, no  
volume customer is going to care.  And the cost vs margins game  has  
always looked pretty frightening to me every time I've been near  
those trying to erect a business case for it.

Too many people know too little about the fundamental mechanics of  
networking (even though they all believe they do) to make an accurate  
decision about whether their choices of dealing with capacity (pick  
whatever real time metric you want, to me delay/jitter is a capacity  
function and nothing less) vs dealing with QoS is the right choice.   
But the glossy slideware looks so impressive. *sigh*

I'm much reminded of how, by some accounts, we were all supposed to  
live in a CoS'ified SVC to the brainstem world a few years back.

Best regards,
Christian