Browse Prior Art Database

Fast Topology Update in Asynchronous Transfer Mode Networks

IP.com Disclosure Number: IPCOM000116884D
Original Publication Date: 1995-Nov-01
Included in the Prior Art Database: 2005-Mar-31
Document File: 2 page(s) / 88K

Publishing Venue

IBM

Related People

Basso, CL: AUTHOR [+2]

Abstract

This describes how information about dynamic link properties, such as available link capacity, can be distributed quickly in an Asynchronous Transfer Mode (ATM) network. The current state of the art is to distribute such updates hop-by-hop through the network. This method results in a relatively long period of time during which the information is out of date. In highly dynamic ATM networks, this can result in non-optimal routing of connections, the failure of connection setups when it is determined that a particular path does not have adequate bandwidth (or other resources), or failure of a connection because the database indicates that there are insufficient resources along the connection path, when in fact, there are sufficient resources. This method uses a broadcast that is sent through all switches in the network.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 52% of the total text.

Fast Topology Update in Asynchronous Transfer Mode Networks

      This describes how information about dynamic link properties,
such as available link capacity, can be distributed quickly in an
Asynchronous Transfer Mode (ATM) network.  The current state of the
art is to distribute such updates hop-by-hop through the network.
This method results in a relatively long period of time during which
the information is out of date.  In highly dynamic ATM networks, this
can result in non-optimal routing of connections, the failure of
connection setups when it is determined that a particular path does
not have adequate bandwidth (or other resources), or failure of a
connection because the database indicates that there are insufficient
resources along the connection path, when in fact, there are
sufficient resources.  This method uses a broadcast that is sent
through all switches in the network.  The update message need not be
processed before it is forwarded.  It is simply forwarded by the
switching hardware.

      Methods for computing a spanning tree on a network are known.
The ATM switches from which a network is constructed generally
support cell multicast function.  The switch can be set so that a
cell received on a switch port carrying a particular label Virtual
Path Indicator/Virtual Circuit Indicator (VPI/VCI) can be propagated
on a set of output switch ports.  In this scheme, we choose a 'well
known' label value for the broadcast update function.  In each
switch, the spanning tree is consulted in order to set the switch so
that a cell carrying the 'well known' label will be switched from the
input port (or edge in the spanning tree) to the output port (or edge
in the spanning tree).  Since a 'well known' label value is used, the
value is left undisturbed in the cell.

      When a switch learns that a property of one of its links has
changed (perhaps by more than a threshold amount), it issues a cell
containing the update and carrying the 'well known' label value.  The
cell is broadcast on the spanning tree of switches and the switches
can change their link information accordingly.  The latency is just
the time required for cell propagation across the network and
processing at the source and target nodes.

      The scheme described so far would work if the size of the link
update was limited to the payload available in an update message that
could be carried in a singl...