On 3 Jul 2007, at 13:56, Brian Butterworth wrote:
I think you will find that there is a longer window due to "series
stacking".
I think you'll find that applies to 15% of the total content, which
means for the majority of content, it's not there. And in fact this
may make things worse, not better.
TCP/IP is specifically designed to share bandwidth, and that's what
it will do.
Sharing bandwidth is not the same as sharing bandwidth in the most
efficient way for particular applications.
TCP/IP on its own is actually incredibly inefficient at sharing
bandwidth. Because all TCP/IP packets are created equally, it
doesn't know the difference between a packet containing voice data
(which needs to get to its destination promptly) and one containing
email data (which can be delayed by a few seconds without any
meaningful damage). If you were designing a protocol from scratch
for bandwidth efficiency and considering quality of service, it
would look very different from TCP/IP.
What utter rubbish.
You might say that, but I think it really just demonstrates that you
don't know what you're talking about. There's a very good reason why
protocols like MPLS have been developed to correct TCP/IP's failings.
TCP/IP was designed for two things: simplicity and resilience. It's
great at both of them. But that doesn't make it either bandwidth-
efficient or capable of providing QoS.
The whole reason the internet exists is TCP/IP. Otherwise we would
all be stuck with our 64k synchronous stuff that the Telcos wished
us to use.
So? This is completely irrelevant to what I posted. Please actually
apply an argument, instead of doing your standard thing of making an
ideological statement and trying to bend the facts around it.
It's still better to understand the nature of the network, rather
than fight against it!
And this is exactly what I mean by an ideological statement. You've
decided that TCP/IP is the be-all and end-all of protocols, and it is
therefore superior and must be used for everything. Thankfully,
people working on core backbones don't agree with you, otherwise in a
few years time QoS would have fallen apart for voice and video.
MPLS exists between the traditional levels 2 and 3 of the TCP/IP
network model precisely to correct its failings as protocol,
particularly for time-sensitive data like streamed voice and video.
There is no real limit to the speed data can be flashed down a
fibreoptic cable - the limitation is the equipment, which can (and
has for decades) been improved by Moores Law.
This is just a silly, meaningless statement. It's like saying there's
no limit to the amount of data you can broadcast via radio -
theoretically (almost) true, but in practical terms bull.
The best networks currently being planned should hit 800Gbit/second
throughput "eventually". And that bandwidth will be limited to
traffic between academic and research establishments. As soon as you
hit one of the hundreds of thousands of T1/DS3 circuits which form a
vast chunk of the backbone of the internet - and will for many, many
years to come - you're screwed.
Sky will be dead in five years time, Moores Law will make IP
delivery more capable than satellite - don't forget that Sky
doesn't own the satellites, the uplinks, the encyption/subscription
system.
You think Sky will only do satellite broadcast? How quaint.
-
Sent via the backstage.bbc.co.uk discussion group. To unsubscribe, please
visit http://backstage.bbc.co.uk/archives/2005/01/mailing_list.html.
Unofficial list archive: http://www.mail-archive.com/backstage@lists.bbc.co.uk/