Browse Prior Art Database

Reversible Automatic Network Routing

IP.com Disclosure Number: IPCOM000100946D
Original Publication Date: 1990-Jun-01
Included in the Prior Art Database: 2005-Mar-16
Document File: 2 page(s) / 91K

Publishing Venue

IBM

Related People

Cidon, I: AUTHOR [+2]

Abstract

We describe an invention which permits the simple definition of reversible routes using a slight modification of the Automatic Network Routing technique.

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

Reversible Automatic Network Routing

       We describe an invention which permits the simple
definition of reversible routes using a slight modification of the
Automatic Network Routing technique.

      The Basic Technique Automatic Network Routing (ANR) is a
technique used for routing packets in high speed networks.  We
briefly review the basic technique used there.  The ANR field is
composed of n words.  The ith word in the ANR field defines the
outgoing link label of the ith hop along the packet path. The
outgoing link label is essentially the internal switch ID or address
(SID) of the outgoing link adapter. Thus, the packet header contains
all the routing information necessary for the routing of the packet
within each intermediate node along the path.  As the packet
progresses through the network, the "used" SIDs are stripped off, so
that the first bits in the ANR field always contain the routing
information for the current node.  This process is depicted in the
figure.  Thus, every node will examine a fixed location in the header
without having to know its position in the path.  No external table
look-ups or processing is necessary, thereby ensuring minimal nodal
delay.

      In the structure described above, the ANRs in either direction
of a route may differ even if both directions traverse the same sets
of links and nodes.  For example, a path that traverses nodes A-B-C
may have ANR 3-6-2, while the reverse path C-B-A may have a
completely different ANR, say, 9-10-2.  The problem occurs because
each node has the freedom to label its links in the way it chooses
and there is no guarantee that the link from A to B will have the
same label as the link from B to A.

      Reversible Automatic Network Routing In reversible ANRs, rather
than having two separate ANRs that refer to the two halves of a
route, we provide a simple transformation that obtains the reverse
ANR from the forward ANR and vice versa.  This property is
particularly useful for network management and control purposes.  In
particular, this transformation shall be the bit reversal operation...