Browse Prior Art Database

Flow Control for Broadband Networks Control Operations

IP.com Disclosure Number: IPCOM000122831D
Original Publication Date: 1998-Jan-01
Included in the Prior Art Database: 2005-Apr-04
Document File: 6 page(s) / 172K

Publishing Venue

IBM

Related People

Cretegny, A: AUTHOR [+6]

Abstract

Broadband networks control flows - In current communication networks, many applications, with various requirements of service, are mixed together on broadband supports. The correct functioning and the reliability of such networks are based on quite heavy control flows, which can spread in the whole network information such as the availability of the links or nodes, the activity of the connections, etc.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 35% of the total text.

Flow Control for Broadband Networks Control Operations

      Broadband networks control flows - In current communication
networks, many applications, with various requirements of service,
are mixed together on broadband supports.  The correct functioning
and the reliability of such networks are based on quite heavy control
flows, which can spread in the whole network information such as the
availability of the links or nodes, the activity of the connections,
etc.

      In addition to these classical flows, new architectures propose
to their customers additional features, as for example the automatic
and non disruptive rerouting of the connections in case of internal
link failure, or the automatic increase or decrease of the
connections bandwidth allocation as a function of the traffic really
sent by the connection or the bandwidth availability in the network.
This means that additional control messages must be exchanged between
nodes in the  network in order to convey the information necessary
for these new functions.

      It is easy to reduce the steady state control messages through
appropriate choices of connection activity survey timer and link
availability update periodicity.

      However, automatic rerouting triggered by an internal link
(also called 'trunk') failure is unpredictable and may consume a lot
of resources.

      Depending of the time in the day, peaks of automatic
connections bandwidth adjustments are triggered, when all the network
users starts/stops the half day activity almost at the same time.

These additional processes impact two main resources:
  o  the links, which must sustain during these periods an
      important control traffic.  Usually, a certain limited
      amount of bandwidth is "dedicated" to the control traffic,
      in term of percentage of the total bandwidth.
       While this is sufficient which high throughput links,
      it is clearly insufficient when the links used in the
      network are low rate, such as for example for 64 Kbps,
      128 Kbps or 256 Kbps
  o  the processors in the switching nodes, which must generate
      and process this additional flow.
       When a lot of connections are supported, both type of
      resources may be saturated during non steady-state periods.

      It is easy to see that the amount of control traffic which may
be sent in the network, even if it is reasonable in steady state, can
be very important depending on the events which disturb the stability
of the network: sudden peak of activity, as well as links failure
trigger particularly intensive exchanges of messages, directly
function of the number of connections affected by the operation.

      To avoid performance impacts on data quality of service, a top
level flow control mechanism must be put in place in end-point nodes
processors, to take into account the availability of the internal
resources and the av...