Warren Wilder wrote:
1. Security
2. Cost.

I can understand these reasons. However, the programs you refer to can
do a lot more than I would need. They can browse/check all the lines
from everyone. That is an obvious no-go for 'us plebs'.

I am wondering if the QoS API allows the requesting application to
receive constant notifications about the actual shape that its sole
channel is in. If, say, there are five routers between the start and end
point, and the one in the center decides to downgrade the quality of
your connection, because another one came along with a higher rating
(which should not be able to downgrade yours, I know) does your
application get any intel about this?
I suspect that the entire QoS negotiation is being dealt with before any
data has been sent, and the talk is over when the channel has been
established.

Warren


I think the problem you may have with your interest is that between the main nodes in the network they (the operators) set-up VCs (Virtual Circuits) using MPLS[mainly] to give a level of service guarantee to their customers (the ISPs and 3rd party telcos). One MPLS VC could [and probably does] carry ALL the traffic for that customer. It is running on a very high speed line (155Mb or higher) which is itself a VC on a SDH or ATM backbone.

Also, when I was selling this stuff, QoS was a "best effort" system. As the whole network is now packet based (rather than circuit switched) it is very hard to give absolute guarantees about bandwidth or latency as your network is dynamic and susceptible to load variations.

As far as I can remember QoS on IP traffic was never really successfully implemented by any of the main router vendors. That is why MPLS took over. There is a better implementation of QoS in IPv6 but, no-one uses it much yet and, again, it is down the infrastructure to actually do something with the QoS/ToS header information and at the moment it is largely ignored...

Hope this helps.

Al

--
http://linuxfromscratch.org/mailman/listinfo/lfs-chat
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Reply via email to