JPS62249544A - Routing system for packet switching network - Google Patents

Abstract

PURPOSE:To reduce the number of packets for exchanging routing information by informing unusableness of all trunk lines or usableness of one trunk line to a network control station and determining the trunk line on the basis of a routing table. CONSTITUTION:The system is provided with the network control station 1 for calculating bias information for routing, exchanges 2-5 for storing a host computer terminals, etc., and executing packet switching and the trunk circuit 6 15 for connecting respective parts. Respecrtive exchanges 2-5 form respective routing tables providing trunk lines to be used successively from a repeating direction having the small number of repeating exchanges up to termination on the basis of received bias information. When no trunk line exists in the direction because of a fault in the trunk line or when one trunk line can be used as the result of restoration of the fault, the state of the trunk line is sent and the trunk line is determined on the basis of the routing table.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a routing system for a packet switching network,
[Conventional technology] Concerning the routing method suitable for load distribution of relay lines in particular, 1 Regarding the routing method of conventional packet switching networks, for example, "Packet switching technology and its 7th application" edited by Institute of Electronics and Communication Engineers (August 1980) As shown in the paper (first published on the 20th), the network management section grasps the topology and traffic situation of the entire network, calculates the route that minimizes the transmission delay time, and notifies each packet switch. has been proposed.

In these systems, the network management section calculates a route when a change occurs in the network topology or when the queue length for a relay line of a packet switch changes. Furthermore, when routing is performed on a packet-by-packet basis using these routing methods, even packets for the same logical channel may be routed via different routes.
Reducing the probability of queue control performed at the destination station in order to arrive at the destination station in a different order than the order sent by the originating station has not been taken into consideration, and furthermore, there are multiple relay lines between the same neighboring stations. It was not clearly indicated how to handle the case where it exists (so-called multi-link case).

[Problem that the invention seeks to solve]

In the above conventional technology, in order to always obtain optimal routing from the viewpoint of traffic load distribution on trunk lines in a switched network, changes in queue length are notified to the network management unit at short intervals and reflected in calculations. There was a need. However, since the queue length inside the packet switch changes from moment to moment, observing this and notifying the network management department of the results in short cycles would be difficult to do in a network with a large number of relay lines. The number of routing control packets exchanged with the packet switching center has increased, which not only affects network traffic but also requires high computing power in the network management section.

An object of the present invention is to reduce the number of control packets exchanged between the network management unit and each packet switching center, reduce the number of calculations in the network management unit, that is, reduce the calculation cost, and It is an object of the present invention to provide a routing method that is effective in distributing loads without being conscious of the configuration and reducing the probability of occurrence of order waiting for packets of the same logical channel.

[Means for solving problems]

The purpose of the above is that in routing calculations, the network management station considers multilinks to be one line, and when each packet switching station detects a change in topology, it uses one or more relay lines between adjacent stations. The network management unit is notified only when all lines have failed and when at least one relay line has recovered from that state, and changes in the queue length are not notified to the network management unit, but each packet exchange is performed instead. The station creates a routing table that gives the relay circuits to be used starting from the relay route with the least number of relay exchanges to the destination station, and based on this routing table, it uses the relay lines in order from the relay route with the fewest relay exchange stations to the destination station. Even if a relay line becomes unusable due to a failure, which is achieved by the routing method that determines the line, the packet switching center will not notify the network management department as long as there is another relay line on the route. , it simply deletes this failed relay circuit from the routing table, and even if the trunk line recovers from the failure and becomes usable, as long as there are other relay circuits on the route, this recovered relay line is simply deleted from the routing table. Just add it again.

[Operation] Each packet switching station does not notify the network management unit of changes as long as there is only one relay line between it and its neighboring stations (at least one relay line is available).
Furthermore, since there is no exchange of control packets due to key length changes, the number of packets for exchanging routing information can be reduced, and the network management station does not need to be aware of multi-links during calculation.

Furthermore, since each packet switching station only needs to update its routing table in response to failures and recovery of trunk lines, there is less intervention by the network management station, and fewer control packets are exchanged between the network management station and the network management station. Become.

〔Example〕

An embodiment of the present invention will be described in detail below with reference to the drawings. Fig. 1 shows the configuration of a packet switching network related to an embodiment of the present invention, and shows a network management station that calculates Bayanu information for routing. 1. It consists of packet exchangers 2-5 that accommodate host computers, terminals, etc. and perform packet switching, and relay lines 6-15 that connect these. However, in the same figure, the host computers and terminals accommodated in the packet exchangers 2 to 5 are omitted because they are not directly related to the explanation of the present invention.
and 15 are 9600 bits/sec, all others are 4800 bits/sec.

Figure 2 shows the packet exchanges sent and received between each packet exchange, including an intra-network header 16 in which control information between exchanges is written, a packet header 17 specified by CC Engineering TT Recommendations X and 25, and this. It consists of the user data 18 that follows. The intra-network header is added only during transmission between switching centers, and is not added to packets sent and received between host computers, terminals, etc. and the packet switching center. The intra-network header includes a type field 1/i-1 that indicates the type of packet, an originating terminal number 16-2 that indicates the number of the host computer/terminal that sent the packet, and an originating station number that indicates the number of the packet switch that accommodates this packet. number 16-3, originating logical channel number 16-4 indicating the logical channel number on the originating station side, destination terminal number 16-5 indicating the number of the host computer/terminal, etc. that receives the relevant packet, packet exchange that accommodates this A threat is detected from the destination number 16-6 indicating the station number, the destination logical channel number 16-7 indicating the logical channel number on the destination side, and information 16-8 necessary for transmission control within the network in addition to the above. FIG. 3 shows the recognition of the network topology in the network pipe Sakai station 1 and each packet switch 2-5.

In the figure, the subscript R indicates the own station (network management station 1 or packet switching stations 2 to 5), and j indicates the route to other stations as seen from the LP. Here, the route has two or more relay lines to adjacent stations and is defined as a t1 route.

FIG. 4 is an example of a bias table calculated by the network management station based on FIG. 3, and shows the bias table for the packet switching center 21C. Here, all routes shown in FIG. 3 are usable. For information on how to calculate this bias, see the Institute of Electronics and Communication Engineers, Information Network Research Group materials.
This is similar to that shown in N35-61 "Proposal of Centralized Control/Distributed Control Routing (SQ+B+NRC) System in Packet Switched Networks". However, the present invention's
In this case, the conditions for starting calculation due to queue length abnormality shown in the above document do not exist. In the figure, for example, the bottom row represents the following. That is, when packet switching center 2 relays a packet to packet switch 5, the number of relay switching centers is 0 if it is sent to route &1, and 1 if it is sent to route 1 or Rzs. If you send to route &4, the number of relay exchanges is 2 and 7P.
It is shown that it is possible to reach the packet switching center 5 via these two relay switching centers.

Figure 5 is a routing table used by the packet switching station 2 based on the bias table in Figure 4 for arrivals of 5 stations, and Figure 6 is a routing table for one arrival station based on the bias table. The flow of processing to obtain a table is shown.

The routing tables are Table 19, which shows the relay line that provides the least number of relay packet switching stations, Table 19, which shows the relay line that gives the second least number of packet switching stations, and Table 20. The relay line that gives the fifth smallest number of packet switching stations. It consists of table 21 shown in FIG. The packet switch 2 creates the table shown in FIG. 5 from the bias table shown in FIG. 4 in the following manner. First, the packet switching station 5 is the destination station.
In order to do this, we will focus on the bottom row of FIG. The number of relay exchanges, which is also smaller by the same amount, is 0, and the route that provides this is R1.
Knowing it's lightning (100~Ioz), Rtt
There are two trunk lines 11 and 15, and knowing that the transmission speed of trunk line 15 is 1102 times that of trunk line (103), Tables 19-1, 19-2, 1
Enter i 11, 15, and 1 in 9-5 (10
4 to 107), the reason why 15'ft is entered twice here is because the transmission speed ratio is twice that of 11. This is because the number of single-send, ie, two-way channels used is increased in proportion to the transmission speed of the relay line 11. Next is the second least (10B~10
9), Find the route that gives the number of relay exchanges from the $4 diagram and obtain routes Rt+ and Rts (101-102), Route stone 1 has relay line 8, and route Rffit has relay line 9 I learned that there are two .10 lines (103) and that the transmission speed of trunk line 9 is twice as fast as the other trunk lines.

Tables 2G-1 and 20-2 in the same manner as table 19
, 20-5 and 20-4 have trunk lines 8 and 9°9., respectively.
Write 10 6 (105-107), f & after Kf
- Find the pull 21 in the same manner as above and enter the trunk line 6 as 21-1 (101 to 104). In the example shown in the figure, tables 19 to 21 each have a small number of relay lines (one relay line), but there may be cases where there is no relay line in any of the tables.

FIGS. 7 and 8 respectively show processing flows when the packet switch detects a failure or recovery of a trunk line.

Furthermore, FIG. 9 shows the processing flow of the network management section when a change in route status is notified from the packet switch.

FIG. 10 shows a processing flow of trunk line selection in a packet switch.

Next, the process of notifying a change in network topology, distributing a bias table, and creating a routing table according to the invention will be explained using the above diagram. If a failure occurs in a relay line and it becomes unusable, the packet switching center checks whether there is another relay line on the corresponding route (
110). If there is at least one usable line, this failure will not be notified to the network management department.When all relay lines for the relevant route become unusable due to a failure, the network topology Notify the network management unit that the change has occurred (111).On the other hand, if the trunk line recovers from a failure, if there is already another usable trunk line on the corresponding route, notify the network management unit (112). If the relay line that has recovered from the failure is the only one related to the route in question, the network management unit is notified that a change in network topology has occurred (115).

When the network management unit 1 receives a notification from any of the packet switches 2 to 5 that there has been a change in the relay line, the network management unit 1 determines that a change has occurred in the topology of the network, that is, the route has become available or unavailable. Knowing that the path has occurred, the bias table is calculated (114), and the result is notified to the packet switches 2 to 5 (N15). In each packet switch, based on the bias table, routing tables 19 to 21 are created as shown in FIGS. 5 and 6 described above.
is created for each arrival station.

Next, as an example of the routing operation in each packet switch, we will show a case where the destination is packet switch 5 in the operation performed by packet switch 2. A packet whose destination is packet switch 5 at packet switch 2 is now sent to packet switch 2. It is assumed that the packet arrives from a host computer/terminal, etc. housed in , or from another packet switching center. The packet switching center 2 learns from the table 19 that the available trunk line selection is 3 (117), and further uses the intra-network header 16 (119) to obtain, for example, K = (mtd (originating terminal number 10 originating logical channel number).

? L ) l + 1- (1) to determine which trunk line to use (11
9,120). For example, if K=1, the packet is relayed to the trunk line 11, and if K=2 or 3, the packet is relayed to the trunk line 15 (121, 122). In addition, if the line described later becomes abnormal, or if there is no relay line in table 19 due to control based on the line usage rate, table 20 is used, and table 21 is used if there is no relay line in table 20. (123, 124).

Note that if a relay line cannot be selected even after performing the above processing on all tables 19 to 21, that is, if no relay line exists in tables 19 to 21, the corresponding packet is discarded (125).

Next, control when a relay line becomes a failure will be explained. In this embodiment, the relay line abnormality refers to a case where the line queue length exceeds a specified value and a case where a relay line in one route becomes a failure.

In either case, this is achieved by removing the corresponding line in the routing tables 19-21 from the table. Further, in the former case, if the queue length returns to the restriction release value, in the latter case, if the failure is recovered, the corresponding line is returned to the corresponding table. In the latter case, if the topology of the route changes due to failure or recovery of the relevant relay line, a change notification is sent to the network management station as described above.

Next (control by the C line user premises will be explained).

How much data does each packet switching center send each time it sends a packet to a relay line? It is counted in C units (bits) and the utilization rate ρ of the relevant trunk line is calculated every fixed time period T (seconds) using the following formula, for example.

Transmission speed C bits/sec) x T (sec)) -- (2) If the result is that ρ exceeds the regulation value, remove the corresponding line from the routing table in the same way as for the line abnormality described above, and set the regulation release value. When it returns below, it will be returned to the table again. However, if two or more identical lines are registered in the same table, only one is removed in one calculation.

As explained above, in the present invention, multiple identical lines may be registered in the routing tables 19 to 21 (8) The greater the number of registrations, the higher the probability that the corresponding trunk line will be selected according to the formula (1). As is clear from the above explanation, the number of registrations increases as the transmission speed of the relevant relay line is higher than that of other relay lines and as the utilization rate is lower than that of other relay lines.

〔Effect of the invention〕

As explained in detail above, according to the present invention, unlike conventional methods, queue length information, which is affected by traffic volume, is not notified to the network management station, and multilinks are treated in the same way as one trunk line. Therefore, the number of packets exchanged between the network service center and the packet switching center can be reduced, and the influence of routing control packets on normal traffic can be reduced.

Furthermore, since the network management station calculates bias information necessary for routing only when the topology changes, calculation costs at the network management station can be reduced. Since multi-links are not taken into account when calculating, for example, at a certain point a new relay line is added between two packet switching centers in addition to the existing relay line.
- Even if a relay line is added, only the corresponding packet switching station needs to recognize it, and the network management station and other packet switching stations do not need to recognize it, making it possible to configure a switching network with good expandability. Become.

Further, in each packet switching center, the probability of selecting a relay line can be weighted according to the transmission speed of the line, and the use of the relay line can be controlled based on the line utilization rate, so load distribution is possible. At this time, packets on the same logical channel are sent to the same relay line as much as possible, so the probability of waiting in line at the destination station can be lowered.

[Brief explanation of drawings]

Figure 1 is a configuration diagram of a packet switching network related to an embodiment of the present invention, Figure 2 is a diagram showing the format of packets sent and received within the network, and Figure 3 is a diagram showing the network topology with respect to Figure 1. 4 is a diagram showing an example of a bias table calculated at a network management station, and FIG. 5 is a diagram showing a part of a routing table created at a packet switching center based on FIG. 4.
Figure 6 is a processing flowchart for determining a routing table corresponding to one arrival station from a bias table;
9 and 8 are processing flowcharts when the packet switch detects failure 1 recovery of the trunk line, respectively.
The figure is a processing flowchart of a network management station notified of a route state change by a packet switch, and FIG. 10 is a processing flowchart of trunk line selection in a packet switch. 1...Network management station, 2-5...Packet switching station. 6 to 15...Relay line, 16...Network header. 17... Packet header, 18... User data. 19-21... Routing f-pull. Fig. 1 6-15; Ya 4! 6 Ho East Fig. 2 Fig. 30 Name 4 Fig. 5 figure

Claims (1)

[Scope of Claims] 1. A network management station that calculates bias information based on a plurality of packet switching centers and the status of relay lines sent from the switching centers and sends the bias information to the switching centers; In a packet switching network consisting of a relay line connecting adjacent stations, the switching station uses the received bias information to sequentially select a relay line to be used in order from the relay route with the least number of relay exchanges to the destination station. create a table,
When a trunk line fails and becomes unusable, and when the trunk line recovers from a failure and becomes usable, the trunk line is deleted and added to the routing table, and the trunk line is deleted and added to the routing table. When, as a result of this occurrence, there is no longer a single trunk line in the said route, and when at least one trunk line in the said route in which no trunk line exists becomes available as a result of recovery from a fault in the trunk line. A routing method for a packet switching network, characterized in that the status of the relay line is sent to the network management station, and based on the routing table, the relay line to be used is determined in order from the relay route with the least number of relay switching stations. 2. The switching center has n associated with the specific number of relay switching centers.
A patent claim characterized in that when selecting a trunk line to be used for routing from among trunk lines, the target trunk line is obtained by taking the modulo n of the sum of the terminal number of the packet-format terminal and the logical channel number. The routing method of the packet switching network according to item 1. 3. When a plurality of relay lines exist in relation to the specific number of relay exchanges, when determining the relay line, the switching center selects a relay line with a higher transmission speed with a higher probability than a relay line with a lower transmission speed. A routing method for a packet switching network according to claim 1, characterized in that the routing method is selected. 4. The switching center calculates the utilization rate of the relay line based on the amount of data sent to the relay line, and when the utilization rate exceeds a predetermined value, deletes it from the routing table;
2. The packet switching network routing system according to claim 1, wherein when the utilization rate falls below a predetermined value, it is added to the routing table again.