JP2001053806A - Automatic route selecting address setting method and route selecting method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically set a route selecting address over a plurality of physical networks by giving preference to one of a plurality of route selecting addresses based on priority order information, changing the other simultaneously with or before/after the preferred address and setting the route selecting address having singularity. SOLUTION: The optical route selecting address is given to a plurality of nodes first. In a succeeding step, priority order information is given to the optional route selecting address. As the way of giving priority order in the step, intrinsic priority order information given to the node or an output interface can be used or another kind of information can be used. In the next step, preference is given to one of the plurality of route selecting addresses based on the priority order information, the other is changed simultaneously with or before/ after the preferred address and, then, the route selecting address with the singularity is set.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a method for automatically setting a route selection address and a route selection method used in a network system.

[0002]

2. Description of the Related Art The topology (connection state of nodes) in a network may change depending on the situation. In such a case, it is necessary to update the routing table or to assign an address to a new node. It is generally desirable that such operations be performed automatically.

[0003] Protocols (route selection protocols) for automatically constructing a route selection table include OSPF and RI.
P, RTMP of AppleTalk, etc. exist. However, these protocols do not automatically set the route selection address itself.

There are DHCP and IPv6 stateless automatic setting as a protocol for automatically setting a route selection address. However, they assign routing addresses on a single physical network.

Therefore, the conventional method cannot automatically set a route selection address over a plurality of physical networks, that is, over a plurality of physical networks via a router.

[0006]

SUMMARY OF THE INVENTION The present invention has been made in view of such a problem. An object of the present invention is to provide a method of automatically setting a route selection address that enables a route selection address to be automatically set across a plurality of physical networks.

Another object of the present invention is to provide a route selection method suitable for the above-described automatic setting method.

[0008]

According to a first aspect of the present invention, there is provided a method for automatically setting a route selection address including the following steps. (A) a step of assigning an arbitrary route selection address to a plurality of nodes; (b) a step of assigning priority information to the arbitrary route selection address; and (c) based on the priority information, Setting a unique routing address by prioritizing one of the plurality of routing addresses and changing the other simultaneously with or before or after this.

According to a second aspect of the present invention, in the method of automatically setting a route selection address according to the first aspect of the present invention, the route selection addresses in the nodes are hierarchized into a plurality of levels, so that the route selection table is stored. It can be small.

The method of automatically setting a route selection address according to claim 3 is the method of automatically setting a route selection address according to claim 1 or 2, wherein the route selection address includes a group address and a physical address. The group address is assigned to a group obtained by dividing the node so as to satisfy the following conditions. (Condition) For an arbitrary physical network, whether the group is completely included or not included at all in the set of all nodes connected thereto.

According to a fourth aspect of the present invention, there is provided a method for automatically setting a route selection address, wherein a route selection table having a plurality of entries is prepared for each node, and the plurality of entries are used as candidates for the route selection address. And the configuration further includes the following steps. (A) a step of giving priority order information to each entry; (b) a step of storing the number of hops from each node to the node in the entry; and (c) a step of storing information in the entry from the node. Broadcasting the hop count and replacing the hop count of the lower priority entry with the hop count of the higher priority entry according to the priority order of the entries in the table; (d) then, based on the hop count in said entry,
Assigning a unique routing address to the node.

According to a fifth aspect of the present invention, in the method of automatically setting a route selection address according to the fourth aspect, when information of the route selection table in the control element is broadcast, the amount of information to be broadcasted is set. Is reduced compared to the amount of information in the case where the information of the routing table in the router is broadcast. As a result, the required communication capacity can be kept low. Also, unlike the router, the information in the route selection table of the control element has only one interface, so that the information in the route selection table of the control element almost matches the information of the interface to which the broadcast is made.

According to a sixth aspect of the present invention, there is provided a method for automatically setting a route selection address according to any one of the first to fifth aspects, wherein the unique address of the plurality of nodes is used. In the setting step, the address is configured to start from a small value.

According to a seventh aspect of the present invention, in the automatic route selection address setting method according to any one of the first to sixth aspects, the node is replaced with an interface, and the physical network is replaced. After reading the internal bus of a physical network or a router, reading the group address as a network address, reading the interface as an interface connector, and setting the unique routing address, the routing address or the routing If the table has not changed for some time, the most used network address on the physical network to which each interface is connected is
The network address of the interface is rewritten.

The route selection method according to the eighth aspect is the fourth aspect.
And a group address cache set for each output interface, wherein the group address cache includes, as information, a group belonging to one physical network to which the output interface is connected. And includes the following steps. (A) After the packet arrives at the node, if the group information included in the packet matches the group information included in the group address cache, the packet is sent to the physical address included in the packet. If not, sending the packet to the next node according to the routing table.

[0016]

DESCRIPTION OF THE PREFERRED EMBODIMENTS A method for automatically setting a route selection address according to an embodiment of the present invention will be described.

First, terms used in this specification are defined as follows. Network system: The entire network subject to automatic assignment Physical network: Network defined by physical layer in OSI (Open System Interconnection) Node: Computer system or hardware that performs some operation based on information from a physical network. Nodes are classified into control elements and routers according to the number of output interfaces (sometimes abbreviated as interfaces) they have. The control element has only one output interface, and the router has a plurality of output interfaces. The state indicated by these terms is shown in FIG. Physical address: A unique address assigned to each output interface in a physical network. The physical layer of a network generally has a function of automatically allocating a physical address to an output interface. Even in a network not equipped with this function, it is possible to implement the same function by software. For networks where this function is implemented by software, for example, Apple Talk, ARCNET
(Reference: Toyo Microsystems "What is ARCNET?"
8 edition, 1993). That is, having the function of allocating physical addresses does not limit the type of physical network.

Further, it is assumed that all elements connected to the network system in the present embodiment have a three-layer protocol stack shown in FIG. The automatic setting method in the present embodiment is executed in the network layer, but is not limited to this.

The automatic setting method of the route selection address in the present embodiment basically has the following steps (a) to (c). (A) assigning an arbitrary route selection address to the plurality of nodes; (B) assigning priority order information to the arbitrary route selection address. As a method of assigning a priority in this step, information of a unique priority assigned to a node or an output interface can be used, or other information can be used. (C) setting a unique route selection address by prioritizing one of the plurality of route selection addresses based on the priority order information and changing the other at the same time or before or after the other. .

First, prerequisite matters will be described.
The route selection address in the present embodiment includes a group address and a physical address. Here, the group address is a code or number assigned to a group including a plurality of nodes. As the group address, any display that can identify the node may be used. Groups are divided so as to satisfy the following conditions. (Condition) For an arbitrary physical network, whether the group is completely included or not included at all in the set of all nodes connected thereto.

This condition simply means that no group spans a plurality of physical networks. When this condition is satisfied, different group addresses are always assigned to nodes connected to different physical networks. Also, under this condition, one router forms one group. The reason is as follows. The router is connected to a plurality of physical networks (see FIG. 1). If routers and nodes in other physical networks are grouped into one group, the group exceeds the range of one physical network. This does not satisfy the condition.

Also, nodes belonging to different groups may be connected in one physical network. That is, a plurality of groups may exist in one physical network.

In a conventional network layer protocol, for example, IP (Internet Protocol), a network address (an address for identifying a physical network) and a host address (an address for identifying a plurality of output interfaces in the physical address) are used. Are combined to form a route selection address. However, since a router is connected to a plurality of physical networks, a plurality of network addresses must be automatically set for the router.

On the other hand, in the group address according to the present embodiment, since one group address is assigned to one router, there is an advantage that management becomes easy.

In addition, the use of group addresses has the advantage that the procedure for automatically setting addresses is simplified. This is because, in order to automatically set a network address in IP, (1) the network address must be the same in all output interfaces connected to the physical network (identity assurance)
(2) It is necessary that two different physical networks have different network addresses (uniqueness is guaranteed). When a group address is used, only uniqueness is guaranteed, and no guarantee of identity is required. Therefore, the automatic setting of the address can be performed in a simpler procedure.

The group addresses of this embodiment are multi-staged (see FIGS. 3 and 4). In the following, a description will be given of route selection based on multi-stage group addresses. However, it is naturally possible to avoid multi-stage.

First, the entire set of nodes is divided into several sets of nodes. A new D-bit address is assigned to each new divided group. Since the number that can be represented by a D-bit address is ^2D , the division is performed so that the number of new groups is ^2D or less. A group formed by one division is called a level 0 group. The address assigned to this group is called a level 0 group address (see FIG. 4).

Next, each of the level 0 groups is divided into a plurality of groups. The new group formed by division is called a level 1 group (see FIG. 4). Level 1
The address in the group (group address) is
The address (D bit) of the level 0 group to which the level 1 group belongs is further added with D bits (see FIG. 3). In order to express the data with D bits, the number of groups at level 1 is also set to ^2D or less. This is the same as in the case of level 0.

In this way, the group addresses from level 0 to L-1 and the group specified thereby are defined (see FIGS. 3 and 4). That is, there are L levels. These groups are designed to satisfy the conditions of the groups described above. Also,
The group address satisfies the above guarantee of uniqueness in a stable state, but the process of determining the address will be described later. Further, each group at all levels is configured to maintain group connectivity. Here, the connectivity of the group means that “given any two nodes in the group, there is at least one route in the group that passes through the two nodes”.

In the following description, the group and the group address at the level L-1 (that is, the lowest level) are simply referred to as a group and a group address. A level with a small value is called an upper level or a higher level, and a level with a larger value is called a lower level or a lower level.

Further, level 1 (that is, an arbitrary level)
, The total length D × (l + 1) bits are divided by D bits, and are expressed as (a ₀ , a ₁ , a ₂ ,..., A _l ). A group having this group address can be written as G {a ₀ , a ₁ , a ₂ ,..., A _l }.

By performing the group multiplication in this manner, a path selection table (a table for subtracting the output interface of the transmission destination and the physical address of the next node based on the destination group address in the packet received by the node) Is ^2D ×, as shown in FIG.
It has a form with L entries. In the example of FIG.
2, L = 8. The top row in the figure corresponds to the level 0 group address. In the figure, “hop” indicates “the number of physical networks that pass from the node having this table to reach the target group”. Hop = 0 at a group address at a certain level 1 means that "this group address is a group address of a group to which the user belongs". Such hop = 0
Is attached at each level. That is, hop = 0
By tracing the group where the entry exists in each level, the group address assigned to the own node can be found. For example, in the example of FIG. 4, this route selection table is for a node having a group address of (0, 1, 2, 3, 0, 1, 2, 3).

As an auxiliary function for setting a group address, a variable used_level indicating up to which level the address is set is set (see FIG. 4). This is a kind of pointer indicating the set level. The reason that the variable used_level is set is
It is as follows. When the group address is automatically set, a level lower than a certain level is in an undetermined state in the route selection table transiently. So the variable used
_level tells you what level has been set. The variable used_level takes a value from -1 to L-1. When used_level = -1, the route selection table indicates a state in which no level is set. used_l
When evel ≧ 0, it indicates that the level up to the level indicated by the variable used_level has been set.

Next, the contents of information included in each entry will be described. Each entry is a hop, for example, as in the entry labeled 10 in FIG.
It has information of number, output interface, physical address, priority, time, addnos, and ifnos. The number of hops has already been described. The output interface captures the physical address of the output interface that has received the broadcast packet. The physical address of the node that transmitted the broadcast packet is taken in as the physical address. The priority order will be described later. The time indicates an elapsed time from when the information of the entry arrives from another node. If this times out, it determines that the entry is invalid and marks the entry unused.
Invalid entries will be described later. addrnos is a variable used when setting a route selection address described later. ifno
s is a variable used in the lowest level entry to identify whether the node of the group corresponding to this entry is a router or a control element. When setting the route selection address,
This is used to implement a mechanism that does not set an address that matches the router.

Next, an algorithm for automatically setting a route selection address based on the group address having the above structure will be described. This algorithm includes an algorithm such as a transmission / reception algorithm and a method for determining a route selection table. This relationship is shown in FIG. The part surrounded by a square frame is involved in the automatic setting algorithm.

First, the basic items will be described. In the transmission algorithm, the contents of the route selection table are broadcast (broadcast) to surrounding nodes. A packet that is or is broadcast is called a broadcast packet. In the receiving algorithm, the contents of the broadcast packet sent to the node from the surroundings are reflected in the route selection table of the node.
With these two algorithms, information of each route selection table can be propagated to each node. On the other hand, the group address determination algorithm is an algorithm for determining (expressing) its own group address from the contents of the route selection table.

In this embodiment, the concept of priority is introduced. The priority determines which table information has priority in order to resolve inconsistencies between the route selection tables broadcast to each node (that is, information that is candidates for a route selection address). The priority is given to each node and each entry of the route selection table of each node (see the entry 10 in FIG. 4). Each entry may be considered to indicate each group that exists at that level. The priority of each node is composed of a node-specific numerical value and an extension field. The numerical value specific to the node is a numerical value fixed to the node and does not change. This is called an identification address. The extension field is a bit that changes according to the algorithm of the present embodiment (described later). The priorities at all nodes are different.

Next, the priority of each entry in the route selection table is determined as follows. (1) The priority of the entry at the lowest level, hop = 0, is matched with the priority of its own node. (2) At other levels, the priority of the entry of hop = 0 is the same as “the highest value among the priorities of all entries at the next lower level”. (3) The entry of hop @ 0 is the same as the "priority of the corresponding entry in the route selection table in the node adjacent to the node". If adjacent nodes have different priorities, the higher priority is selected, as described later.

In the state where the route selection address is set at the output interface of the entire network system and there is no change, the priority of each entry (that is, a group or group address formed by dividing at a certain level) is determined by the network system. Matches throughout. If the two entries have different priorities in the route selection table of two nodes, it means that there are two or more groups represented by the same entry, that is, the same group address.

According to the method of the present embodiment, it is possible to keep entries with a higher priority while propagating the route selection table. Then, one group will eventually correspond to one group address. That is, a group address whose uniqueness is guaranteed can be automatically generated. Details will be described later. Next, the structure of the broadcast packet will be described. As described above, in order to automatically generate an address, the contents of the route selection table are broadcast to the physical network as a packet. The structure of this packet (broadcast packet) is as shown in FIG. This packet has almost the same form as the routing table. But a variable indicating the level used
There is no used_level. Unused entries are set to hop = 0 to indicate that they are unused. This is because they are unused and may be used to indicate their own group. The content of each entry is 20
, The number of hops, the priority, and the addrnos. These contents are the same as those used in the routing table.

Next, a transmission algorithm for transmitting a broadcast packet will be described. In the transmission algorithm, a broadcast packet is sent to each output interface at regular time intervals. This time interval is called a broadcast interval. When the sending algorithm creates a broadcast packet from the routing table, it performs the following calculations for hop: As viewed from the node that receives the broadcast packet, the number of hops to the group indicated by each entry when the route is traced toward the transmitting node is set to the hop of the broadcast packet. When each entry of the route selection table of the transmitting node is broadcast to the output interface indicated by the entry, the entry corresponding to this entry in the packet is unused. This is because the information of this entry comes from this interface, and the meaningless operation of returning the same information is avoided. This transmission algorithm is
Basically, it is the same as a conventional RIP, and therefore, further description is omitted.

Next, the receiving algorithm will be described. This operates when a broadcast packet is sent from a surrounding node to a certain no complaint.

In the receiving algorithm, although not essential,
Perform filtering to increase efficiency. here,
If all the entries of the transmitted broadcast packet are not set, nothing is performed. Also, when the number of hops in the entry is larger than the threshold value, nothing is performed because there is a possibility that information transmission may be looped. This threshold can be pre-given to each output interface a value deemed sufficiently large, eg, 100.

After the filtering, each entry in the route selection table and the broadcast packet is processed in order from the highest level. The contents of this processing are shown in Table 1 below.

[0045]

[Table 1]

The algorithm shown in Table 1 is also shown in the flowchart of FIG. 7 for reference. Hereinafter, description will be made based on Table 1.

The processing in Table 1 is for correcting the information of the entry in the route selection table. In Table 1, the three columns from the left represent the states of the entries in the table, and the two columns on the right represent the operation when the respective states of the three columns on the left are detected. "-" In the operation column means that nothing is performed, and "replace"
This means that the information of the route selection table is replaced with the information of the broadcast packet. Also, "table"
Indicates a route selection table, and “packet” indicates a broadcast packet. In Table 1, the comparison between the packet and the table compares the level of priority. Here, the equal sign indicates the same priority.

First, if an entry in the table is not used, replacement is performed, and nothing is performed on the variable used_level.
Next, if an entry in the table is used and hop =
When it is 0, the following is performed. If the priority in the packet entry is lower than that in the table, do nothing. This is because the entry of that node has a higher priority than the entry of another node using the same address. In other words, this has the advantage that if the same address is assigned to a node, address duplication can be prevented depending on the level of priority.

When the priority in the packet entry is equal to that in the table, the number of hops in the packet is 1
It depends on whether or not. If it is 1, the process proceeds to a lower entry. At this time, the packet sent from the own node is returning via the adjacent node. If the hop number of the packet is 1, nothing is done.

If the priority of the packet entry is higher than that of the table, replacement is performed and the variable
Set used_level to the level of the entry being compared.

When an entry in the table is used and hop ≠ 0 (that is, hop> 0), as shown in Table 1, the priority level and the packet h
Processing is performed according to the number of ops. This processing is apparent from Table 1 and FIG. 7, and therefore, further description is omitted.

In this way, the contents of the entries in each table can be modified by using the priority. That is, even if an overlapping address is assigned, it is possible to prevent the address from being duplicated and to guarantee a unique address.

Further, in order to determine the address of the own node in addition to the above algorithm, the following is performed.

For example, it is assumed that no address is assigned in the network. At this time, immediately after the node is powered on, used_level = −1, that is, the route selection table is completely empty. The entry of the route selection address is determined by the above-described reception algorithm. However, the used_level cannot be reduced unless the node belongs to which group, that is, the entry of hop = 0. That is, the address at the lower level cannot be determined. Hop while lowering the used_level little by little
= 0 is determined, and finally, the lowest order, that is, the group address of the node is determined.

For the purpose of generalization, it is assumed that the route selection table has been determined up to used_level = 1−1. That is, the entry of hop = 0 exists up to the level l. Then, since the entry of hop = 0 can be traced to the level l, the route selection address of this node is determined to the group address of the level l-1. Next, the variable used_level = 1
, And all the entries of level 1 are made unused.
In this state, it waits for a while to receive a broadcast packet. At this point, the route selection table is not yet in a normal state (a state finally required). When the level 1 entry is gradually filled with information from the broadcast packet and the route selection table is in the normal state, or the normal state is not reached forever, and when the time-out occurs (a certain time or more has elapsed). ), The level l group address is determined. Details of the determination method will be described later. If the priority of the entry is lower than the priority of the node (priority given to each node) at the time of determination, the priority of each entry is replaced with the priority of the node. After that, the route selection table is normalized after a time considering that there is a broadcast packet that has not been received yet.

The algorithm for determining the level 1 group address will be described with reference to FIG. First, the level 1 is set to 0 (step S8-
From 1), it waits until entries at level 1 (that is, level 0 in this case) are set to some extent by broadcasting (step S8-2). Then, l =
It is determined whether it is L-1 (step S8-3). l ≠
In the case of L-1, an address can be assigned by any of the three methods shown in step S8-4. Method 1 is, in short, a method of assigning a group address so as to have a small value. The method 2 is a method in which unused groups are not used as much as possible, but among the used groups, a group having the smallest number of nodes belongs is used. Method 3 is a method of starting with the group to which the least number of nodes belong, regardless of whether it is currently unused or not, so that the addresses used are scattered throughout the group addresses.

In these three methods, it can be expected that there is no bias in the use route selection address among all the addresses in the order of the methods 1, 2 and 3. If there is a bias in the use of routing addresses, nodes that later try to join the network are likely to be stuck in one location with surrounding routing addresses and are likely to be depleted. Is done.

Here, the depletion state will be described. The depletion state refers to a state in which even when all group addresses have been searched, there is no route selection address to be assigned.

There is a method in which a node in a depleted state is given a higher priority and is re-participated. Although this causes a problem that the route selection address of the node to which the address has already been assigned changes, the addresses are rearranged, and the depleted node can obtain the route selection address. However, it can be considered that this method also eliminates the bias of the use route selection address among all the addresses in the order of the methods 1, 2, and 3. Although the occurrence of the depletion state can be solved, the route selection addresses of the surrounding nodes are reset. Then, not only is it not suitable for attaching / detaching a new node, but also the time from turning on the power supply to ending the assignment of the route selection address to the entire network system may become long, which is not preferable.

In step S8-3, used_level =
In the special case of L-1 (that is, the lowest level),
Proceed to step S8-6. In step S8-6, processing differs depending on whether the relevant node is a control element or a router. In the case of a router, each unused entry at level 1 is found, and the address represented by the unused entry is set as a route selection address. In the case of a control element, the method 1 is different from the methods 2 and 3. This is illustrated in FIG.
Since it is as described in S8-6, further description is omitted. If there is no unused entry, no address is assigned, and this is called a depleted state.
The solution of the depletion state will be described later.

As described above, the group address can be automatically set by the method of the present embodiment.

Here, a method of counting the number of group addresses existing in the used group in the above methods 2 and 3 will be described. For this purpose, an integer variable called addrnos in the route selection table and the broadcast table is used. Level 1-1
Hop = 0 (route selection table) or hop = 1 (broadcast table) is the sum of the addrnos of all entries at level l, and the addrnos of the entry used at level L-1
= 1. For the entry of hop @ 0 in the routing table, addrnos of the broadcast packet
Is set, the addrnos of each entry is the total number of route selection addresses existing in the corresponding group. A specific example is shown in FIG. In this example, at the position corresponding to each entry shown in FIG. 4, (hop number, addrno
Information is described in the format of s). For example, looking at the entry of hop = 0 at level 4, addrnos = 20 = 1
+ 17 + 1 + 1. This is the sum of addrnos at level 5 below. Thus, the number of group addresses is counted.

A method of eliminating the depletion state will be described. When a certain node is depleted, an address in a group to which this node cannot belong is intercepted. As a method of this, by scanning the route selection table from the higher level value, an entry with a lower priority than the priority of this node and hop ≠ 0 is found. Then, the entry is set as the group address of this node. In this case, a group having the same group address temporarily exists at a remote location on the network system. However, since the priority of the group address of this node is higher, it is not possible to assign the group address to this node. it can. In this case, the node having the lower priority will eventually lose the priority and the address will be reset. If the priorities of the entries in the route selection table are all higher than the priority of this node, the value of the extension field constituting the priority of this node is increased by 1 to raise the priority, and
_level = −1 and all the route selection tables are emptied.
Next, the setting of the route selection address is performed again.

Next, the discovery of an invalid entry will be described. This is a process of searching for an invalid entry in the route selection table. An invalid entry is an entry for which an entry is reserved, but for which the group indicated by this entry does not actually exist. Here, it refers to a group in the entire network system
Focus on one entry. Whether this entry is not invalid can be determined from whether there is an entry having the same priority and hop = 0. because,
This is because a node having an entry with hop = 0 means that it belongs to the group indicated by this entry. Further, the fact that groups having different priorities are different groups is a definition of the priority of the groups. Therefore, it can be said that this entry is invalid when the following two conditions are satisfied. (1)
hop ≠ 0. (2) In the information of the corresponding entry in the broadcast packet from the surroundings, the hop number is always larger than the hop number of this entry. Alternatively, the information of the corresponding entry has a different priority.

Therefore, in order to make the algorithm concrete, a timeout is provided for a certain period of time, and it is checked that a broadcast packet from the surroundings within the period always satisfies (2). In addition,
The condition (1) can be determined immediately.

An unused entry sends a cancel packet in order to quickly inform the surroundings that it has become unused. The node receiving the cancel packet looks for an entry in the corresponding route selection table. If there is no entry or the priority is different, the node ignores the packet, and if the priority matches, the packet is unused. Further, a cancel packet is sent to the surroundings.

Next, the timeout process of the route selection table will be described. For each entry other than hop = 0, information is sent from another node at regular intervals by a broadcast packet. The absence of this information means that the contents of the entry have become invalid. A certain magnification C is determined, and the broadcast interval C is determined.
If information about this entry does not come in from the next node or the next node indicated by the entry at twice the time, even if the number of hops in the packet is large, the information is replaced with information from another node. If the information does not come from anywhere even after waiting for 2C times, the entry is deleted. After a while, the route selection table is normalized after a while.

Next, a description will be given of a route selection method using the group address set as described above.

Basically, this route selection method is as follows.
It can be executed by a method such as a conventionally used RIP. Here, a method that can further increase the efficiency when a group address is used will be described.

In this method, a group address cache (hereinafter, referred to as "cache") is provided for each output interface 2 (see FIG. 10) of the router 1. This cache has, as information, a list of groups 4 inside the physical network 3 to which the output interface is connected. It should be noted here that, from the definition of the group, no group exists across the physical network, and therefore, the group exists inside the physical network. The cache is created by holding the number of hops as 0 or 1 in one path selection table and holding the group located on the output interface side of the output interface as information. The contents of such a cache can be obtained from the contents of the broadcast packet. In other words, the broadcast packet traverses the entry of hop = 1, and if the broadcast packet can be traced to level L-1, the group address corresponding to this is located in the physical network from which the broadcast packet has entered. Become. Therefore, the cache can be created by using the group address obtained in this way as an index of the cache so that the output interface into which the broadcast packet has entered can be retrieved.

The route selection method will be described with reference to FIG. First, when a packet arrives at a node, it is determined whether or not the group indicated by the packet is in the group address cache of the node (step S11-1). Then, if it is a control device,
According to the information of the physical address indicated in the packet,
It is sent to the control device (indicated by reference numerals H1 to H3 in FIG. 10) (step S11-2). Since the physical address is uniquely determined, the packet can be delivered to the target address by this operation. If the determination in step S11-1 is false, the packet is sent to the next node based on the information in the route selection table (step S11-3).

When the group address is multi-staged at a plurality of levels, the group address of the packet and the group address of the node are compared from the higher level. If a match occurs at a higher level, the lower levels are compared. If they match up to the end, it is determined that the packet has reached the target group, and the route is selected toward the physical address of the packet. If there is a mismatch at any level, it is sent to the next node based on the information in the route selection table.

In the above description, a method using a group address has been described. However, a unique address can be assigned to a conventionally existing IP address using the concept of priority. I
In the automatic setting of P, there are two points to be considered. First, the IP is not the address assigned to the node, but the address assigned to the output interface, and the network address that is automatically set instead of the group address is guaranteed to be the same in the physical network. That is what we have to do.

To solve the former point, a route selection table may be prepared for each output interface. To address this, the terminology described above has been changed to
As shown in FIG. 12, the replacement is performed as follows. Thereby, a network address can be automatically set instead of a group address. Physical network → physical network or internal bus interface of router → interface connector node → interface group address → network address

However, when assigning the level L-1 group address, the interface across the internal bus has a different network address.
Also, the interfaces across the physical network are assigned so as to have the same network address as much as possible.

However, the network addresses allocated in this way are not always the same on one physical network. Therefore, a process for changing the network address to ensure the identity is prepared so that all the output interfaces connected to one physical network have the same network address.

This identity assurance process involves first collecting the network addresses of the nodes on the network, and then setting the most frequently used network address in its own network. This guarantees identity. If this process is operated at the same time as the address automatic setting algorithm, it may cause a deadlock with each other. Therefore, when it is confirmed that all the route selection tables of the node remain unchanged for a certain period of time, it is considered that the network address has been allocated to the entire network system for the time being, and the process of ensuring the identity is started.
Here, it is determined that the route selection table is unchanged. However, since the route selection table is information to be the route selection address, the method of checking the route selection address is substantially the same. It is. For example, in a system that does not use the concept of a route selection table, such a method can be adopted. With the above process, the automatic setting of the IP is completed.

(Experimental Example) The results of evaluation by simulation are shown below. Evaluation method The time from the simultaneous turning on of the power supply of the network system to the end of the assignment of the route selection address is called the automatic setting time. The state in which the assignment has been completed is called a stable state.
In this experiment, the scale of the network system was 100,
When the number of bits D and the number L of levels are changed to about 000 nodes and the number of bits assigned to the group is changed, the following points are considered by simulation. (1) How long the automatic setting time will fit. (2) In a stable state in which the network is actually operating, can a new node or network be attached / detached without disturbing the stable state?

When considering the insertion of a node into a network, when a control element is inserted, only the same group address as that of the other control elements connected to the physical network is set, and there is no particular shortage of addresses. , Only the router is evaluated. The problem is when new routers and new physical networks are inserted.

The degree of exhaustion is defined as an index for measuring the possibility of attaching / detaching a router in a stable state. When a new router is inserted into each physical network, consider all patterns of the route selection table obtained from the surroundings, and in these tables, the number of unused entries is g, and the average m
(g) is defined as the degree of exhaustion. As m (g) is smaller, the inserted router or a node on the network connected by the router may be depleted.

Since it is desired to reduce the memory capacity of the route selection table as much as possible, L and D should be set so as to reduce the number of entries in the route selection table. The number r of entries of the basic route selection model is r = 2 ^D L.

The number of bits B of the group address of level L-1 of a is B = DL. In order to minimize r while keeping B constant, the value becomes minimum around D = 1 / log _e 2 regardless of B.

The number of nodes is about several hundred thousand to several million nodes,
Assuming that about 200 nodes are attached to one physical network, the number of necessary group addresses is contained in B = 16 bits. However, considering the possibility of the exhaustion state, B = 16 bits or more, that is, 24 bits and 32 bits must be considered with a margin. Then, when considering combinations of D and L which are these divisors, Table 2 is obtained.

[0084]

[Table 2]

When the above method 1 is selected, the size of the broadcast packet and the size of the route selection table are determined in proportion to the number of entries. When the methods 2 and 3 are selected, the variable addrnos is equal to B and changes according to B. Therefore, if the number of entries is the same, the smaller B is, the smaller the broadcast packet and the route selection table can be.

With 128 entries, the size of the data portion of the packet is about 2048 bytes (adjusted by methods 1, 2, and 3).
It is. This size is based on the Ethernet MTU (Max Transfer Un
it) is slightly larger than 1
This is the boundary between whether the packet can be terminated or not. Therefore, the size up to this point was targeted as the range of evaluation.
Also, it can be predicted that the larger the value of D, the better results such as the automatic setting time and the degree of exhaustion will be obtained. Therefore, D = 2
Up to 4 were evaluated.

Next, a simulation method will be described. Performing an evaluation experiment on a network system with hundreds of thousands of output interfaces actually takes a lot of time. Although it depends on the combination of parameters, as an example where time is required, D = 4, B = 32 and the network scale is 30
It takes about one month to simulate when changing from the output interface to 12,000 output interfaces. It takes more time if there are more output interfaces. However, the CPU assumed Sparc170MHz.
In this experimental example, a simulation in which some combinations of various parameters are thinned out and the number of nodes is further reduced is performed. On the other hand, a result at about 100,000 nodes is predicted from the result. For this reason, a verification simulation that the prediction was correct was performed.

First, a prediction method will be described. Since Method 3 seems to be superior to the other methods, Method 1,
No. 2 performs a simulation using four combination patterns of B = 16, 32 and D = 2, 4. On the other hand, Method 3 performs all seven patterns. Next, in the present embodiment, it is highly likely that terminals in the same physical network will be assigned the same group address, and that the number of nodes connected to the physical network will not affect the automatic setting time and the degree of exhaustion. it can. Assuming that the physical network is 2,000, the number of nodes is 200 in one physical network, which is 400,000 nodes, which is the scale of the target network system. Therefore, even if the condition of 2,000 physical networks is not changed and 200 units are changed to 5 units,
It seems that almost the same result is obtained.

First, the control element 5 is assigned to each physical network.
The simulation is performed under the condition that the number of connected routers is one and the average of one router is one per physical network. This is referred to as simulation pattern 1. Next, the condition that the average number of routers is one per physical network does not change. Of the total, up to 200 physical networks, 200 control elements are connected, and the remaining physical networks are five control elements. Then, a network system is generated and simulated. This is referred to as simulation pattern 2.

The prediction is made based on the result of the simulation pattern 1. Verification that the above prediction is correct is made by comparing the simulation patterns 1 and 2.
In the simulation pattern 2, when there are 2,000 physical networks, 10% of the physical networks are connected to 200 control elements. On the other hand, it is concluded that the evaluation is not affected by the number of nodes connected to the physical network if the difference between the simulation patterns 1 and 2 is much smaller than 10% for each measurement value such as the automatic setting time and the exhaustion degree. Table 3 shows the difference between the two patterns.

[0091]

[Table 3]

Finally, in order to reduce the evaluation time of the simulation pattern 2, the simulation pattern 2 is performed only on the simulation pattern 1 for which a good result is obtained.

Next, the configuration of a network system for simulation will be described. For the simulation, a system for generating a network system is required. This system includes two types of programs: a program that randomly configures a network with a router and a physical network; and a program that randomly adds a control element to a generated network.

In a program for constructing networks from the given number of networks and the number of routers, first, the number of interfaces for each router is determined and the network is randomly connected. The number of interfaces is also determined randomly from a given range of numbers. In the network generated in this way, the entire network system is not always connected, so a portion that is not connected is found and a correction is made to connect. The network system output last has serial numbers assigned to the physical network and the router. In a program for adding a control element to a network, a fixed number of terminals are added in the order of the serial numbers of the physical network.

In the simulation patterns 1 and 2,
Each parameter was selected as shown in Table 4.

[Table 4]

Next, the handling of the broadcast interval will be described. The broadcast interval must be determined so as not to burden the physical network.
The detailed description is omitted because it departs from the gist of the present invention, and only a rough guide is described for considering the evaluation result.

At the time of automatic setting, there is a demand to make the broadcast interval as short as possible. However, the communication capacity of the broadcast cannot exceed the communication capacity of the physical network. Therefore, in order to shorten the broadcast interval, the communication capacity of the broadcast must be reduced. One effective approach is to focus on the fact that most nodes connected to the physical network are control elements, and optimize the broadcast packets of the control elements.

Since the control element has only one output interface, the information of the route selection table in the control element except for the information about the group address allocated to the control element is the physical interface to which the interface is connected. This is information that already exists on the network.

Therefore, the control element has its own priority.
(12 octets), used_level (1 octet), and only the group address (2 to 4 octets) need be broadcast. It is data that can be sufficiently expressed if a total of 17 octets is used.

Assuming that there are 200 control elements and 3.5 routers in the physical network, the broadcast packet is assumed to be about 30 octets including the header, the broadcast packet of the router is assumed to be 2 koctet, and the broadcast interval is i seconds. The communication capacity required for broadcasting is {200 · 240 (bit) + 3.5 · 16k (bit)} / i = (104 / i)
k (bit). This means that if the entire communication capacity is 1Mbps, i =
1 indicates that broadcast packets can be exchanged with a communication capacity of about 10%. Therefore, in the evaluation of this paper, the broadcast interval was set to 1 second as a guide.

Next, the evaluation will be described. The newly introduced group address does not specify the number of group addresses that exist in one physical network. Therefore, the number of group addresses per physical network may increase.

In FIG. 13, the horizontal axis represents the number of physical networks in the entire network system, and the vertical axis represents the number of group addresses used. This is an example of a case where a simulation is performed using the simulation pattern 1. From this figure, it can be seen that the number of group addresses is proportional to the number of physical networks. The proportionality constant does not greatly differ depending on the address assignment method and D, and the group address used is 2.2 times the number of routers or physical networks on average.

Accordingly, the group address assigned to the control element is 1.2 times as large as that of a router or a physical network (assuming that the number of routers and the number of physical networks are the same (see Table 4)). An average of 1.2 group addresses exists in one physical network.
+ [Number of routers], indicating that the group addresses do not multiply significantly.

Next, when the scale of the network system was changed, the total number of output interfaces that became depleted during the automatic setting time was examined (see FIGS. 14, 15, and 16). The simulation pattern is 1. In the above-described method 1, a depleted node appears with about 200 networks. In method 3, when B = 16, the number of networks is 100 to 200, and when B = 32, about 300 nodes appear in a depleted state. In method 3, (B, D) =
It can be seen that the depletion state hardly appears except for (16, 2).

From this, using method 3, (B, D) =
(16, 4), (24, 2), (24, 3), (24, 4), (32, 2), (32,
Use the combination of 4).

Further, the automatic set time when there was no depleted node was measured (FIGS. 17 to 21). In method 1, it may take an extremely long time, and attention is required. Methods 2 and 3 do not take much time except for D = 2 and B = 16. In particular, in the case of D = 3, 4 in Method 3, the time is around 200 broadcast intervals, which is practical.

In the case of the method 3, it is understood that it depends not on B but on D. It can be seen that the larger the D, the shorter the automatic setting time.

[0108] For Method 3, which yielded good results, the degree of depletion after the network system was stabilized was measured by simulation pattern 1 (Figs. 22 and 23). It can be seen that the degree of depletion depends on B, not D. Naturally, the larger B is, the smaller the depletion degree is.

In the scope of this evaluation, any network
The scale of the system has not been exhausted, and it has been shown that nodes can be practically attached to and detached from a moving network.

Next, the number of used route selection addresses, the automatically set time, and the degree of address depletion in the simulation pattern 2 were measured, and compared with the simulation pattern 1 by the following equation. That is, the value at the time of the simulation pattern 1 for each number of networks is d
1, when the value of pattern 2 is d2, the ratio of the difference between the two is determined by the following equation. Table 5 shows the results. Here, m () represents an average. m ((d2-d1) / d1)

[0111]

[Table 5]

According to this, two simulations
The difference between the patterns is 6% or less, and it can be determined that there is almost no difference between the two. Thus, it can be concluded that each measurement and the conclusions derived therefrom are not affected by the number of control elements per physical network.

Assuming that the route selection address should be short in order to shorten the packet in consideration of the result shown above and a low-speed network, the method 3 is adopted, and B = 16,
A good value is 24. The above results show that the number of networks is 2,000, the number of terminals per physical network is 200
This shows that each parameter is within a practical range even in a network system with 400,000 terminals.
Also, it can be seen that nodes can be attached to and detached from a moving network. Further, according to the usage status of the group addresses shown in FIG. 13, 1.2 group addresses are used for control elements in one physical network. No problem.

[0114]

According to the first aspect of the present invention, a method for automatically setting a route selection address includes the following steps. (A) a step of assigning an arbitrary route selection address to a plurality of nodes; (b) a step of assigning priority information to the arbitrary route selection address; and (c) based on the priority information, Setting a unique routing address by prioritizing one of the plurality of routing addresses and changing the other simultaneously with or before or after this.

According to this method, it is possible to automatically set a route selection address over a plurality of physical networks.

In the automatic setting method of the route selection address according to the second aspect, the route selection address in the node is hierarchized in a plurality of levels in the automatic setting method of the route selection address according to the first aspect. It can be small. Then, the amount of information stored in each node can be reduced.

A third aspect of the present invention provides the automatic route selection address setting method according to the first or second aspect, wherein the route selection address includes a group address and a physical address. The group address is assigned to a group obtained by dividing the node so as to satisfy the following conditions. (Condition) For an arbitrary physical network, whether the group is completely included or not included at all in the set of all nodes connected thereto.

Therefore, there is an effect that it is easy to automatically assign a route selection address across a plurality of physical networks.

According to a fourth aspect of the present invention, there is provided a method for automatically setting a route selection address, wherein a route selection table having a plurality of entries is prepared for each node, and the plurality of entries are used as candidates for the route selection address. And the configuration further includes the following steps. (A) a step of giving priority order information to each entry; (b) a step of storing the number of hops from each node to the node in the entry; and (c) a step of storing information in the entry from the node. Broadcasting the hop count and replacing the hop count of the lower priority entry with the hop count of the higher priority entry according to the priority order of the entries in the table; (d) then, based on the hop count in said entry,
Assigning a unique routing address to the node.

As a result, addresses of the entire network can be efficiently and automatically set.

According to a fifth aspect of the present invention, in the method of automatically setting a route selection address according to the fourth aspect, when the information of the route selection table in the control element is broadcast, the amount of information to be broadcast is set. Is reduced as compared with the amount of information in the case where the information of the route selection table in the router is broadcast, so that the required communication capacity can be suppressed low. Also, unlike the router, the information of the route selection table in the control element indicates that
Since there is only one, the information in the route selection table of the control element almost coincides with the information of the broadcast destination node. Therefore, even if the information amount is reduced, there is no problem in address setting.

A method for automatically setting a route selection address according to a seventh aspect is the method for automatically setting a route selection address according to any one of the first to sixth aspects, wherein the node is read as an interface and the physical network is replaced. After reading the internal bus of a physical network or a router, reading the group address as a network address, reading the interface as an interface connector, and setting the unique routing address, the routing address or the routing If the table has not changed for some time, the most used network address on the physical network to which each interface is connected is
Since the network address of the interface is rewritten, the automatic setting of the IP address can be performed.

The route selection method according to the eighth aspect is the fourth aspect.
And a group address cache set for each output interface, wherein the group address cache includes, as information, a group belonging to one physical network to which the output interface is connected. And includes the following steps: (A) After the packet arrives at the node, if the group information included in the packet matches the group information included in the group address cache, the packet is sent to the physical address included in the packet. If not, sending the packet to the next node according to the routing table.

With this configuration, a packet can be delivered to a target node according to a group address.

[Brief description of the drawings]

FIG. 1 is an explanatory diagram conceptually illustrating the configuration of a general network system.

FIG. 2 is an explanatory diagram of a protocol stack provided in the network system according to the embodiment of the present invention.

FIG. 3 is an explanatory diagram for explaining multi-stage group addresses according to the embodiment of the present invention.

FIG. 4 is an explanatory diagram for explaining a route selection table using multi-stage group addresses in the embodiment of the present invention.

FIG. 5 is an explanatory diagram for explaining a partial algorithm relationship relating to an automatic setting algorithm of a route selection address used in the embodiment of the present invention.

FIG. 6 is an explanatory diagram illustrating a configuration of a broadcast packet used in the embodiment of the present invention.

FIG. 7 is a flowchart illustrating a reception algorithm used in the embodiment of the present invention.

FIG. 8 is a flowchart illustrating an algorithm for determining a group address used in the embodiment of the present invention.

FIG. 9 is an explanatory diagram for explaining an example of a method of calculating the number of group addresses used in the embodiment of the present invention.

FIG. 10 is an explanatory diagram illustrating a route selection method using a group address used in the embodiment of the present invention.

FIG. 11 is a flowchart illustrating an algorithm of a route selection method using a group address used in an embodiment of the present invention.

FIG. 12 is an explanatory diagram illustrating replacement of a use concept when the present invention is applied to automatic setting of an IP address in the embodiment of the present invention.

FIG. 13 is a graph showing a relationship between the number of physical networks in the entire network system and the number of group addresses used in the embodiment of the present invention.

FIG. 14 is a graph showing the number of nodes in a depleted state in the example of the present invention.

FIG. 15 is a graph showing the number of depleted nodes in the example of the present invention.

FIG. 16 is a graph showing the number of depleted nodes in the example of the present invention.

FIG. 17 is a graph showing the automatic setting time of the group address when there is no depleted node in the embodiment of the present invention.

FIG. 18 is a graph showing an automatic setting time of a group address when there is no depleted node in the embodiment of the present invention.

FIG. 19 is a graph showing an automatic setting time of a group address when there is no depleted node in the embodiment of the present invention.

FIG. 20 is a graph showing an automatic setting time of a group address when there is no depleted node in the embodiment of the present invention.

FIG. 21 is a graph showing an automatic setting time of a group address when there is no depleted node in the embodiment of the present invention.

FIG. 22 is a graph showing a depletion degree of a group address after the network system is stabilized in the embodiment of the present invention.

FIG. 23 is a graph showing a depletion level of a group address after the network system is stabilized in the embodiment of the present invention.

Claims (8)

[Claims] 1. A method for automatically setting a route selection address, comprising the following steps. (A) a step of assigning an arbitrary route selection address to a plurality of nodes; (b) a step of assigning priority information to the arbitrary route selection address; and (c) based on the priority information, Setting a unique routing address by prioritizing one of the plurality of routing addresses and changing the other simultaneously with or before or after this. 2. The automatic setting method of a route selection address according to claim 1, wherein the route selection address in said node is hierarchized in a plurality of stages. 3. The route selection address includes a group address and a physical address, and the group address is assigned to a group obtained by dividing the node so as to satisfy the following condition. 3. The automatic setting method of a route selection address according to claim 1, wherein (Condition) For an arbitrary physical network, whether the group is completely included or not included at all in the set of all nodes connected thereto. 4. A route selection table having a plurality of entries is prepared for each node, and the plurality of entries are information serving as candidates for the route selection address, and further include the following steps: A method for automatically setting a route selection address. (A) a step of giving priority order information to each entry; (b) a step of storing the number of hops from each node to the node in the entry; and (c) a step of storing information in the entry from the node. Broadcasting the hop count and replacing the hop count of the lower priority entry with the hop count of the higher priority entry according to the priority order of the entries in the table; (d) then, based on the hop count in said entry,
Assigning a unique routing address to the node. 5. The method of automatically setting a route selection address according to claim 4, wherein when the information of the route selection table in the control element is broadcast, the amount of information to be broadcasted is broadcast by the information of the route selection table in the router. An automatic setting method of a route selection address, characterized in that it is reduced compared to the amount of information in the case. 6. The method according to claim 1, wherein, in the step of setting a unique address in the plurality of nodes, the address is shifted from a small value.
6. The method for automatically setting a route selection address according to any one of items 1 to 5. 7. The method according to claim 1, wherein the node is read as an interface, the physical network is read as a physical network or an internal bus of a router, and the group is read. After replacing the address with a network address, replacing the interface with an interface connector, and setting the unique routing address, if this routing address or the routing table has not been changed for a while, Is
A method for automatically setting a route selection address, comprising rewriting a network address of an interface to a network address most frequently used on a physical network to which each interface is connected. 8. A route selection table according to claim 4, further comprising a group address cache set for each output interface, wherein said group address cache belongs to one physical network to which said output interface is connected. A route selection method comprising the following groups as information and including the following steps: (A) After the packet arrives at the node, if the group information included in the packet matches the group information included in the group address cache, the packet is sent to the physical address included in the packet. If not, sending the packet to the next node according to the routing table.