Adaptive Applications Support

Document Charter Adaptive Applications Support WG (adapts)
Title Adaptive Applications Support
Last updated 2011-12-09
State Approved
WG State Concluded
IESG Responsible AD (None)
Charter Edit AD (None)
Send notices to (None)


Adaptive applications, those that can adapt their resource requirements 
variations in network resources, can provide a significant reduction in
Internet congestion.  In addition, adaptive applications can offer much
better quality to end users than can non-adaptive applications under
varying network conditions.  Such applications will represent an 
piece in the solution to the current Internet congestion control/QoS
(quality of service) problems.  The technology for making audio and 
applications adaptive has been under development by researchers over the
last 10 years.  

Applications built on protocols that do not do congestion control can
hamper the performance of the Internet. TCP backs off in the face of
congestion but other transport protocols, such as UDP and RTP, do not, 
thus can contribute significantly to network congestion.

Many Internet researchers and developers have come to believe that the
network forwarding devices (e.g. routers) must do internal congestion
control in order to provide robust and viable network service.  A 
of schemes have been suggested for policing traffic and managing router
buffer queues.  In such a controlled environment, non-adaptive 
will fare badly while those applications which can adapt will provide
better performance to the user.

This BOF will consider what can constructively be done to further the
development of adaptive applications.  For example, what additional
protocol support is needed from RTP?  While the technology for making
applications adaptive, does not require standardization, it would be
helpful to publish an Informational RFC which states what behavior is
required from applications to perform well in the Internet today and in
particular in an Internet where the routers implement internal 
control.  It would also be useful to make clear the issues involved in
constructing adaptive applications, and to show some example mechanisms 
algorithms which are known to work.   

This BOF focuses on best-effort traffic, rather than traffic for which 
reservations of some kind have been made (by RSVP for example). It is
reasonable to expect that best effort traffic will be the bulk of the
Internet traffic in the near future.