Re: [httpstreaming] [conex] [dispatch] Q-HTTP

[Mikael Abrahamsson <[e-mail]>]

On Thu, 11 Nov 2010, David Singer wrote:

> It's interesting you say this in the context of real-time media, because 
> I believe a network can give time guarantees ("any packet I deliver on 
> this will be delivered within x ms") or delivery guarantees ("all 
> packets will eventually be delivered") but it's hard (impossible, maybe) 
> to do both.

Are we talking 100.000% or 99.5% ? With cheap residential broadband, I 
want the norm to be that the ISP is not congesting their 
core/distribution. I don't want formal guarantees, I just want the normal 
mode to be that I can use my full access bandwidth, regardless of what 
application I'm running. With flash events, both telephony and packet 
networks can't handle peak load, but this doesn't happen every day, it's 
"few times per year" events.

> Personally, I want my fair share of the shared 'roads' no matter how 
> many others are sharing them, and I expect the 'last mile' personal link 
> to have (at least) the capacity that was sold to me.

And what is "fair"? I want my access link to be the limiting factor, not 
my ISP distribution/core/peering links. I was sold X megabit/s "Internet 
Access". I want that.

I don't want "X megabit/s Internet Access unless it's that time of day 
when most people want to use the Internet, then you get much less but we 
won't state how much less, and oh by the way, you'll get different amount 
depending on what L3/L4 information you provide to us in your packets".

Let's work on empowering the users so they can understand when their 
internet supplier is not giving them what they paid for. The users need 
tools, the ISP doesn't. The ISP can look at their mrtg graphs showing 
octets in and out, and if the 5 minute average utilization of their link 
is over ~75-85% or so, they're most likely not doing their job right.

-- 
Mikael Abrahamsson    email: [e-mail]