Multiplex communications – Pathfinding or routing – Switching a message which includes an address header
Reexamination Certificate
1998-03-06
2001-02-13
Ton, Dang (Department: 2731)
Multiplex communications
Pathfinding or routing
Switching a message which includes an address header
Reexamination Certificate
active
06188696
ABSTRACT:
FIELD OF THE INVENTION
This invention relates to communication networks and in particular to a network which includes at least one multi-segment ‘Ethernet’ repeater hub, that is to say, a multi-segment hub comprising at least two repeater segments and a multiplicity of ports to which end stations or other hubs may be attached. Such a multi-segment hub has an internal switching matrix that allows any port to be coupled to any repeater segment.
BACKGROUND OF THE INVENTION
Such a hub is organised so that for each segment or repeater conflict between transmissions is resolved by a carrier-sense multiple access technique, which is well established in the art and is the essence of an ‘Ethernet’ communication system. A station wishing to transmit will first detect whether there is a carrier signal present in the segment to which it is connected and will either not transmit or will wait until an end of transmission is detected and then transmit its signal packet.
It is known that a system which resolves signal contention in this manner is subject to difficulties in conditions of heavy loading. This is particularly true for an Ethernet hub wherein, for example, there is signal traffic between stations connected to different segments which are connected via a bridge, because traffic is then subject to a plurality of possible contentions.
Multi-segment hubs of this character commonly include an internal switching matrix by means of which any port can be connected to any segment. It is therefore possible to contemplate measuring the traffic across the ports and controlling the connections so as to equalize the traffic across the segments. One such technique computes the volume of traffic through each segment in a given period of time and reallocates the ports connected to those segments to provide, so far as possible, an equalization of the traffic through the segments.
One disadvantage of such a technique known as octet-based load balancing, is that there tends to be a dispersal of end stations within a work group, so that forms an initial condition, wherein a multiplicity of stations forming a work group are connected to the same segment, at least some of the stations in such a work group have their ports associated with different repeaters. It will thus be apparent that there is a high likelihood that a given packet will appear on two segments, that is to say the segments connected via respective ports to the source and destination of the packet, and therefore the traffic across the bridge between segments remains high.
SUMMARY OF THE INVENTION
The present invention is primarily concerned with an improved method of allocating ports to segments based on the collection of data representing the traffic through the multi-segment hub. The technique involves computing, for a multiplicity of ports in turn, preferably starting with the busiest, the effect of moving the port to each of the participating segments and evaluating the effect in terms of the utilization of segments and the volume of inter-segment bridging for each possible or selected position of the port. The solutions may be compared against the ‘best so far’. Further candidate solutions may be generated by computing the effect of moving the next busiest port to a variety of segments. In this way, one may generate a ‘tree’ in which each node is a possible solution to the load balancing problem. Theoretically, if all possible movements of ports were considered, the tree would have depth ‘n’ where ‘n’ is the number of ports and the total number of possible solutions would be m exp (n) where m is the number of segments. It will be apparent that the number of solutions becomes unmanageably large for practical networks, where m may be substantially more than 2 and n is large. Accordingly, one may limit the depth of descent of the tree, since less busy ports will affect the result less significantly, and one may limit the number of solutions pursued at each level in the tree. A chosen solution is the one yielding the best feasible result.
The invention requires for its performance a means of collecting data representing the traffic between ports and the segments that the traffic utilizes and a programmed data processor which can evaluate that data and ultimately control the internal switching matrix to govern the allocation of the ports to the segments in accordance with the chosen result. Such a memory and data processor are themselves commonplace parts of a hub of this character, the invention lies in the process which is implemented by the processor and associated memory.
REFERENCES:
patent: 5936962 (1999-08-01), Haddock et al.
patent: 5999538 (1999-12-01), Haddock et al.
patent: 0615362 A1 (1994-09-01), None
patent: 0701347 A2 (1996-03-01), None
patent: WO 96/29802 (1996-09-01), None
Moran Paul J.
Quine Richard A.
3Com Technologies
Nixon & Vanderhye P.C.
Ton Dang
LandOfFree
Method of operating a communication network to provide load... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of operating a communication network to provide load..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of operating a communication network to provide load... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2558483