[38119] in North American Network Operators' Group
RE: QOS or more bandwidth
daemon@ATHENA.MIT.EDU (Pete Kruckenberg)
Tue May 29 11:28:37 2001
Date: Tue, 29 May 2001 09:19:05 -0600 (MDT)
From: Pete Kruckenberg <pete@kruckenberg.com>
To: <nanog@merit.edu>
In-Reply-To: <F58XdoYfYoiB5ANH2cv000032ea@hotmail.com>
Message-ID: <Pine.LNX.4.30.0105290849510.14812-100000@localhost.localdomain>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Errors-To: owner-nanog-outgoing@merit.edu
On Tue, 29 May 2001, Ukyo Kuonji wrote:
> The problem is, while most vendors support tagging and
> priority queuing, non of the current vendors can support
> true end to end QoS. Instead, we have taken to calling
> their options CoS.
The elusive "end-to-end QoS" seems to be a network
management issue, motivated primarily by the number of knobs
(and some featuritis) and lack of QoS best-practice (because
there's just not a lot of QoS practice, period).
Simple end-to-end technologies (TOS, DiffServ) are enough to
handle congestion-management QoS strategies (and some
differentiated services), are interoperable across most
vendors, and are supported in most edge- and core-class
devices.
Beyond simple QoS schemes, the complexity mandates an
end-to-end management tool. I suspect the cost/benefit curve
gets pretty flat above a simple QoS strategy, and anything
more complex has diminishing marginal value.
Any studies been done on cost/benefit of QoS, it'd be
interesting to see where the technical/business case is
compelling and where it makes no sense.
Pete.
