Home page logo
/

nanog logo nanog mailing list archives

RE: QOS or more bandwidth
From: Pete Kruckenberg <pete () kruckenberg com>
Date: Tue, 29 May 2001 14:07:25 -0600 (MDT)


TE isn't just for congestion avoidance/mitigation. It's also
for choosing the best path, based on the costs and
characteristics of the traffic, as well as the load on the
network.

So delay-intolerant (voice, video) traffic gets preference
on the lowest latency path, and pushes delay-tolerant flows
to alternate higher-latency (possibly longer) unused paths.
MPLS constraint-based routing allows this, as does ATM. PBR
w/ IP routing would be extremely complicated to do this.

Same rules can be applied to homogeneous traffic, where I
can manage at a finer granularity than
per-desination-CIDR-block. I can move traffic around my
network internally as well as to peers, to get full
utilization out of my network without introducing
congestion.

This presumes a network with multiple paths to a given
destination, with approximately equal performance and cost.

Pete.

On Tue, 29 May 2001, Ukyo Kuonji wrote:
How is TE part of QoS?  Maybe we are talking a different
type of TE.  To me, TE is the part that the traffic
takes, not really what queue it sits in to transmit.

I could see a arguement for QoS being part of TE, but...
(Assuming that you TE your netword so there is no
conjestion.)

--- Original Message ---

Everything that you said is correct.  However, you missed one
important part of QoS, which is TE.  TE is about avoiding congestion
in the first place by more efficiently using all of the bandwidth
you already have in the network.  To that extent, it is a substitute
for adding more pipe.


  By Date           By Thread  

Current thread:
[ Nmap | Sec Tools | Mailing Lists | Site News | About/Contact | Advertising | Privacy ]