Browse Prior Art Database

Connection Establishment in Hierarchical Networks with Bandwidth Management

IP.com Disclosure Number: IPCOM000110322D
Original Publication Date: 1992-Nov-01
Included in the Prior Art Database: 2005-Mar-25
Document File: 4 page(s) / 214K

Publishing Venue

IBM

Related People

Doeringer, WA: AUTHOR [+6]

Abstract

Current procedures for the establishment of bandwidth-reserved connections in high-speed networks all build on the implicit sumption that all information required for the path computation and bandwidth reservation is available at the network level, that is, is represented in the topology database and accessible through algorithms operating on it. This obviously implies that the potentially considerable complexity at the node level is ignored, i.e., the resources, algorithms, and protocols that relate to a network node's internal structure. However, the growing complexity of (distributed) switching nodes in large-scale networks presents severe problems to a straightforward application of the current signalling procedures, some of which may even become insurmountable: .

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

Connection Establishment in Hierarchical Networks with Bandwidth Management

       Current procedures for the establishment of
bandwidth-reserved connections in high-speed networks all build on
the implicit sumption that all information required for the path
computation and bandwidth reservation is available at the network
level, that is, is represented in the topology database and
accessible through algorithms operating on it.  This obviously
implies that the potentially considerable complexity at the node
level is ignored, i.e., the resources, algorithms, and protocols that
relate to a network node's internal structure.  However, the growing
complexity of (distributed) switching nodes in large-scale networks
presents severe problems to a straightforward application of the
current signalling procedures, some of which may even become
insurmountable:
.   Any non-hierarchical architecture for large-scale networks with
complex switching nodes implies that the intra-nodal links of
switching fabrics have to be included into the topology database
since they are subject to bandwidth management.  As a consequence,
the size of the database will increase drastically, and since
utilization changes of such links then have to be globally
advertised, the network-wide traffic for database updates will also
induce enormous control overhead.
.   Performing bandwidth management for inter-node links only is not
sufficient to guarantee an efficient operation of the entire network.
.   Since the network nodes may employ different internal structures
and functions, they will make use of potentially very different
internal control algorithms.  It is, however, highly undesirable to
see changes of such internal algorithms reflected at the network
level.
.   The sequential execution of path computation and bandwidth
reservation is no longer appropriate in networks with a high
complexity at the node level, i.e., with complex network nodes whose
internal structure is not reflected at the network level.

      The proposed solution to the above list of problems is based on
the notion of a hierarchical network as a network with a non-trivial
node level.  In contrast, non-hierarchical networks are those, where
the node level does not contain any information and algorithms apart
from those visible in the topology database and executed at the
network level.  The novel approach builds on the following two
concepts:

                            (Image Omitted)

First a suitable structure for the network nodes is defined that
keeps node-level details out of the topology database and the
network-level algorithms, yet allows the use of node-level features
where appropriate.  Secondly, a complementary connection
establishment procedure is proposed which makes efficient use of this
node structure to exploit node-level functions for global
optimization and which increases the concurrency of path computation
and bandwidth reservation....