WO1998004069A1

WO1998004069A1 - Bridge/repeater device

Info

Publication number: WO1998004069A1
Application number: PCT/GB1996/001697
Authority: WO
Inventors: David William Banes; Petros Santamas
Original assignee: 3Com Ireland
Priority date: 1996-07-18
Filing date: 1996-07-18
Publication date: 1998-01-29
Also published as: AU6466696A

Abstract

The invention provides a communication device for a Local Area Network (LAN) which can be switched to operate in at least two modes of operation. In the preferred embodiment the device can be switched to operate either as a repeater (12) or as a bridge (14). Switching between the modes of operation is done based on the volume of data traffic on the LAN to provide, when possible, the more or most efficient data communication. Typically, when traffic on the LAN is light the device operates as a repeater and when traffic is heavy the device operates as a bridge.

Description

BRIDGE/REPEATER DEVICE

This invention relates to Local Area Networks (LANs) for providing intercommunication between computers and/or other digital equipment (hereinafter called data termination equipment and abbreviated to DTE). More particularly, it is concerned with networks of the kind in which DTEs are connected (normally via a media attachment Unit (MAU), also called a transceiver) to a common transmission medium such as a coaxial cable, a twisted pair cable or an optical fibre. The invention is primarily (but not exclusively) concerned with networks operating Carrier Sense Multiple Access techniques with collision detection (CSMA/CD). The best-known networks of this type are those specified by the International Standards Organisation as ISO 8802/3 networks and by Xerox Corporation as "Ethernet" networks. In such systems, data is transmitted in frames which have a limited range of lengths and are normally made up of a meaningless preamble (for establishing synchronization), a start-of-frame indicator, a destination address segment a source address segment, a control segment (indicating, for instance, the frame length), a segment of data (often beginning with a frame or protocol identification) to be transmitted to the DTE identified by the address identification, and a frame check segment for verifying accuracy of transmission. In such LANs a number of different types of devices are used to interconnect the various DTEs present in the network.

Firstly digital repeaters (usually multiport repeaters, MPRs) are used to restore digital signals that have been attenuated or otherwise degraded and also to provide for branching as required. MPRs repeat frames received on an input port indiscriminately to all their output ports thereby providing for communication between all DTEs connected to that repeater. MPRs necessarily (because of delay limits imposed by network specifications) begin to retransmit an incoming data frame before the complete frame has been received.

Secondly DTEs in a LAN can be connected using an intelligent device which is capable of deciding whether to forward a data frame depending on its destination. These intelligent devices are usually known as bridges or switches which are considerably more expensive than MPRs and, because of their decision making capabilities as mentioned above introduce a delay in the transmission in excess of the overall frame length. For convenience we will call such intelligent devices "bridges".

A third type of component which can be used for interconnecting DTEs in a LAN is a "cut-through bridge". Cut-through bridges are discussed for instance in "On Transparent Bridging of CSMA/CD Networks" by Conrad K. wok et al in Globecom '89, IEEE Global Telecommunications Conference & Exhibition, November 1989 pages 5.7.1-6.

As described in that document a cut-through bridge operates like a normal bridge in that it only forwards a frame if such forwarding is appropriate given the destination address of the frame, but it begins such forwarding as soon as it has the destination address. This means that the delay introduced by a cut-through bridge is usually not as great as that introduced by a normal bridge.

At light loads a cut-through bridge as described in the above-mentioned document functions similarly to a normal repeater except that some delay is introduced while the destination address is read. It also has the function of queuing incoming frames if the destination network segment is already busy and therefore, if the transmission traffic is high volume, the cut-through bridge will appear to operate like a normal bridge. Bridges, repeaters, cut through bridges and other connection devices are typically used together in LANs. A bridge or bridges can be used to divide the overall network into a number of segments, the DTEs within each segment being interconnected via MPRs. Such a system has advantages in terms of ease of communication because DTEs within each segment can communicate with each other freely without affecting or being affected by communications between DTEs in other segments. However as mentioned above a bridge introduces a considerable delay into a transmission and therefore communications between DTEs in two segments in the arrangement just discussed are considerably slowed by having to pass through the bridge.

In designing a LAN therefore bridges and repeaters are used in combination according to the particular characteristics of the network under consideration and the expected usage of the network by the various users. In some circumstances however it is not possible to determine before hand what type of device would be best at a particular location in a network. This may be because it is unpredictable how much different users may utilize the network or because there is potential for extra users to attach to the network which will affect the network characteristics.

The present invention provides a communications device for use in a LAN, the device having at least two ports and comprising data transfer means operable according to at least two modes of operation to transfer, in use, data between the ports; monitoring means arranged to monitor, in use. the volume of data arriving at the ports: and control means sensitive to an output of the monitoring means to control the data transfer means to operate according to an appropriate one of said at least two modes of operation.

In the present invention therefore the device switches between the two modes of operation (eg repeater and bridge) according to the levels of traffic on the network according to which mode is more advantageous for the network in the prevailing traffic conditions.

The monitoring means can be any known means for measuring the level of activity at each port of the device. For instance there may simply be provided a counter at each port which counts the number of packets arriving at that port, the counter being read and reset periodically so that the read values give the activity since the previous reset. One particular data rate detector is shown in GB-A-2280574 which is used to decide whether a particular user port should be disconnected. Preferably the device according to this invention comprises a repeater selectively switchable into a communication path between the ports to provide one mode of operation and an intelligent bridge-like device to provide another mode.

The intelligent device with a bridge-like capability may be a bridge proper or a cut-through bridge.

In essence, then, in the present invention a particular type of interconnection device e.g. bridge, repeater or cut-through bridge, is used at any given time according to what will provide the optimum performance on the basis of the level of traffic in a LAN at that time. For instance, if one considers that the device according to the invention can be taken to divide a LAN into two segments each having a number user DTEs, when traffic is light it would not be a problem that all data transmitted by all of the DTEs is seen by all other DTEs, and therefore it is advantageous to configure the device as a repeater thereby minimising data transmission times. However in circumstances where traffic is heavy it is advantageous to configure the device as a bridge because, while this introduces a delay time in data transmitted between the two segments, all traffic is not transmitted between the two segments and this reduces the segment load with consequent improvement in speed and efficiency. In another example, a server device may be connected to one of the two segments of the network mentioned above and a number of users may be connected to the other. In such a situation when the traffic levels in the segments are approximately balanced this would indicate that most of the traffic was communication between users and d e server, because each transmission by a user causes a response from the server. In these circumstances it may be considered that the network efficiency is best served by having the device configured as a repeater which does not introduce delays into the communications with the server. Alternatively when the traffic levels are out of balance this indicates that the users are communicating with each other and not so much with the server. In these circumstances it may be better to configure the device as a bridge as this increases the efficiency of the server which now receives only those packets it should receive.

In an embodiment of the invention artificial intelligence is provided to control the switching of the device between its various modes of operation and therefore the switching can be controlled to optimize performance, according to the characteristics of the particular installation.

The invention has particular application in certain types of LAN for instance those operating according to a Novell standard, in which each transmitted data packet is acknowledged by the destination DTE. In such a system the level of traffic passing through the communications device can be simply monitored by a counter which is incremented when a data packet passes in one direction and decremented when a data packet passes in the other. If traffic is light and therefore an acknowledgement of a first data packet will be seen before the transmission of a second or subsequent data packet the counter will remain at or around zero. In heavy traffic the magnitude of the value in the counter may increase considerably and therefore the control device may be simply a device which operates according to whether the counter is above or below a threshold, switching the operation to be a bridge-like device when the counter is above that threshold, and to operate as a repeater once the counter is below a second threshold which is different to the first threshold.

Alternatively, traffic levels may be monitored by providing a counter at each port which counts the amount of traffic arriving at the port irrespective of whether it passes through the device. In order that the present invention be better understood, preferred embodiments thereof will be described in the following, by way of example, with reference to the accompanying drawings in which:

Figure 1 is a schematic diagram of a typical LAN layout; and Figure 2 is a block diagram of the communication device according to the preferred embodiment of this invention.

Figure 1 illustrates in schematic form a simple LAN layout. The LAN comprises a number of DTEs 1 connected to a common bus 2 and then to a port 3 of a communications device 10 which divides the LAN into two zones A and B. Zone A contains the DTEs bearing the reference numeral 1 and the bus 2 while zone B contains further DTEs 4 connected by a further bus 5 to a further port 6 of the device 10. In a known arrangement device 10 would typically be a bridge of the type discussed in general terms above which would look at the destination addresses of data packets reaching it and only pass those data packets through if that is necessary to allow the data packet to reach its destination.

In setting up a LAN, the nature of the various DTEs and their relationship to each other would be considered such that the LAN is configured so that most communications on the LAN occur within either zone A or zone B. That is most of the data transmissions made by a DTE in one segment would be addressed to another DTE in the same segment. Such data transmissions would not be passed by device 10 and therefore, when data transmission is occurring within each segment the bridge effectively divides the LAN into two effectively doubling the data carrying capacity of the system as a whole.

On the occasion of, for instance a DTE 1 in zone A wishing to communicate with a DTE 4 in zone B, the transmitted data will be received by bridge 10 which will establish that it needs to be received by a DTE in zone B and therefore pass the transmission through to zone B. Because of the nature of bridge 10, a delay is introduced into such communications between DTEs in different segments. In the preferred embodiment of the present invention, device 10 in Figure 1 is not a bridge but is a device as defined in general terms above and illustrated more specifically in Figure 2.

Figure 2 is a block diagram of a communications device according to the preferred embodiment of this invention which may be inserted as device 10 in Figure 1 with zone A on the left-hand side and zone B on the right-hand side as illustrated. The device comprises a repeater 12 and a bridge 14 which can be selectively switched by way of a selector 16 into the communication path between zone A and zone B. This means that the device can operate and appear to the LAN in which it is positioned as either a repeater or a bridge-like device. The selection between these two modes of operation is made according to which is the more advantageous for the LAN, given the traffic pattern on the network. The selection is effected by way of control means 18. The selection is made such that when the traffic is light on the network, the device is configured as a repeater and when traffic is heavy the device is configured as a bridge.

When traffic is light the device of this invention therefore provides the advantage that there are no delays introduced into communications between DTEs in the two zones A and B and it is as if the LAN as a whole is not divided into two segments. When traffic becomes heavy the advantages of having the LAN divided into two zones, as in the prior art. in order to provide the increased capacity, outweigh the advantages of reducing the delay time for communications between the two zones. Therefore at such times the device according to this embodiment is switched to be a bridge-like device. Consequently the device according to this invention is operated to provide the best mode of operation for the LAN given the existing traffic patterns for the time being.

In Figure 2 there is illustrated means which can make a simple assessment of the traffic pattern in order to effect the selection of the mode of operation. In Figure 2 there is illustrated a counter 20, this is operated such that the counter is incremented on receipt of a data packet and decremented if the next received data packet is on the other port. As will be appreciated the count in the counter will remain at or around zero if the traffic reaching the device from the two segments is relatively balanced, but if one segment becomes very busy the traffic pattern will become unbalanced and the count in the counter will increase. The control means is configured such that when the count in counter

20 passes a certain threshold it operates selector 16 such that the device operates as a bridge.

Alternatively, a separate counter may be provided at each port which simply increments on receipt of a data packet. At predetermined intervals, the control means would read the value of the counters and reset them, with the read values giving an indication of die traffic levels in each LAN segment. In this case the control means would be triggered to switch between modes of operation according to whether or not the segment traffic levels are approximately balanced.

As mentioned above this is a simple implementation of the traffic assessment required in the present invention and it is within the scope of this invention to provide other methods for assessing the traffic levels and distribution, and in particular to provide artificial intelligence within the control means 18 to make the assessment. This is particularly the case when the device 10 is a multi-port device as it is necessary for the control means to assess the amount of traffic on each of the ports and make connections between ports in the most efficient manner. It will thus be appreciated that a simple counter is merely illustrative of the basic function of assessing amounts of traffic on each port and the invention will use. in practice more sophisticated arrangements.

What is basically required is an indication preferably substantially instantaneous, of the amount of traffic on each side of a bridge or on each port of a multi-port bridge so that the changeover from a bridge to repeater and back can be achieved at the appropriate time. This could be conveniently done by software on a moving window time averaging basis or by the above mentioned counters.

The actual threshold for switching may be set according to the expected traffic patterns in a particular LAN and may be made dependent on any other relevant perameters of a particular LAN.

As mentioned above this invention has particular application in certain types of LAN where the protocol is such that any data transmission is followed by an acknowledgement by the receiving party. Clearly in such a network the amount of traffic for a given amount of data transfer is much increased and therefore providing a network which will constantly check that it is operating according to what is most advantageous according to the existing traffic pattern is advantageous.

Claims

1. A communications device for use in a LAN, the device having at least two ports and comprising data transfer means operable according to at least two modes of operation to transfer, in use. data between the ports; monitoring means arranged to monitor, in use, the volume of data arriving at the ports: and control means sensitive to an output of the monitoring means to control the data transfer means to operate according to an appropriate one of said at least two modes of operation.

2. A communications device according to Claim 1 in which the monitoring means comprises counter means arranged to count the number of data frames arriving at the ports.

3. A communications device according to Claim 1 or 2 in which the data transfer means comprises a repeater selectively switchable into a communication path between the ports to provide one of said modes of operation.

4. A communications device according to Claim 3 in which the data transfer means comprises a bridge selectively switchable into the communication path between the ports to provide one of said modes of operation.

5. A communications device according to Claim 3 or 4 in which the data transfer means comprises a cut-through bridge selectively switchable into the communication path between the ports to provide one of said modes of operation.

6. A communication device according to Claim 3, 4 or 5 in which said repeater is switched into said communication path when the volume of data is low.

7. A method of operating a communications device in a LAN, the device having at least two ports, the method comprising providing data transfer means operable according to at least two modes of operation to transfer in use, data between the ports; monitoring the volume of data arriving at the ports; and controlling the data transfer means to operate according an appropriate one of said at least two modes of operation according to said volume of data.

8. A method according to Claim 7 in which said step of monitoring the volume of data comprises counting the number of data frames arriving at the ports.