Quantum Flow Control (QFC)

Material dated 1995-96 that was saved in Feb.-May 1997 from

http://www.qfc.org/

but no longer exists there (as of 2003).


[I have placed this material here in order to preserve it, because I believe it is of interest to the networking scientific community (M. Katevenis, August 2003)]:



                               [QFC Alliance]

                  Contacts | Members | Front page | Events
----------------------------------------------------------------------------

                          QFC Questions and Answers

What is Quantum Flow Control?

Quantum Flow Control (QFC) is an ATM protocol supporting the Available Bit
Rate (ABR) service necessary to support bursty applications for which
bandwidth requirements are difficult to predict in advance. ABR is
distinguished from other ATM services by its ability to support
instantaneous access to unused (available) network bandwidth with very low
cell loss rates comparable to link error rates. Current LAN-based
applications that have no mechanism for bandwidth reservation can be fully
supported by ABR services based on Quantum Flow Control.

What is the Quantum Flow Control Alliance?

The QFC Alliance was formed in 1995 to develop a suitable flow control
protocol that can comprehensively meet the requirements of the ATM Available
Bit Rate Service. Alliance members fully support the ATM Forum's objective
of making ATM technology specifications available to the industry as early
as possible. And, they believe that the significant work already completed
on the QFC specification can accelerate the ATM Forum's Traffic Management
ABR service objectives.

What is the Flow Control Consortium?

We originally called ourselves the "Flow Control Consortium", but changed
the name to "Quantum Flow Control Alliance" to align with the name of the
QFC Specification.

Do you intend to submit QFC to the ATM Forum?

Yes, when it is ready, when it has sufficient support, and when it will not
slow down the Forum's work. The Forum needs every chance to complete the
definition of the Explicit Rate (ER) mechanism without distraction, and QFC
needs to be considered on its own technical merits, which are considerable.
In the meantime, the specification is available to all, and companies with
something positive to contribute are welcome to join the QFC Alliance.

Is Quantum Flow Control expensive to implement?

For terminal equipment such as network interface cards, the ER mechanism is
far more complex than QFC. For example, it requires a variable rate
controller rather than a simple on/off, it uses floating point, and it has a
large number of parameters which must be signaled and stored. QFC does
require per-connection accounting in its feedback information, as does the
ER mechanism. This is not surprising, as per-connection accounting is a good
idea for any control mechanism.

Do you expect that QFC might replace ER?

In the long term, possibly, depending on experience. In the short term QFC
is likely to be implemented first in LANs and other computer data intensive
applications, and use tunneling to get through long-distance connections.
QFC outperforms ER in supporting high-performance LAN-based applications
over integrated services ATM.

It is quite possible that ER and QFC will coexist indefinitely,
complementing each other and possibly fulfilling the needs of different
segments of the ATM market. We have been careful to design QFC so that it
can coexist with and even enhance ER, and so that it aligns with many other
sections of the ATM Forum's traffic management version 4.0 specification.

Our products are going to have to support ER anyway. Why should we support a
second ABR protocol?

A growing number of companies already support QFC; products lacking it may
not be competitive. The complexity of implementing QFC in an end system is
relatively small and the potential performance benefits are substantial
throughout a network. QFC supports point to multipoint service naturally,
requires minimum new signaling capabilities, and is self-configuring.

Why should my company join the QFC Alliance?

To have a voice in the further development of QFC, including the right to
participate in discussions and votes. To have immediate access to the
working documents of the consortium. To participate in the mutual agreement
to cross-license any patents necessary to implement the specification. And
most important: to support the standardization and widespread use of QFC.

What are the requirements for joining the QFC Alliance?

An interest in developing and using QFC, in promoting its standardization
through relevant bodies such as the ATM Forum and ITU-T, and willingness to
grant other members a license to any patent necessary for the implementation
or use of any portion of the Quantum Flow Control Specification. See the
membership agreement for details. Copies are available for your review.

What are the signaling requirements of QFC?

We have minimized the signaling requirements. QFC is compatible with UNI
3.0/3.1 signaling.

What about compatibility with existing EFCI switches?

Since QFC is hop-by-hop it can be added to networks one link at a time,
improving overall performance at each step. Thus it creates
better-performing links that will make ER, EFCI, or anything else work
better.

The real legacy problem is whether ATM will support existing LAN-based
applications better than today's LANs. QFC supports the requirements of
high-performance LAN-based applications today.
----------------------------------------------------------------------------

Copyright 1996 QFC Alliance. All Rights Reserved.