JP3800338B2 - Communication control system - Google Patents

Description

  The present invention relates to a communication control system in which a communication station that performs communication in accordance with a TCP / IP protocol is connected to an E bus.

  TCP / IP (Transfer Control Protocol / Internet Protocol) is a widely used protocol for computer networks. TCP / IP is a combination of two protocols, TCP and IP. TCP is a protocol for accurately transmitting data, and IP is a protocol for transferring data between networks in a packet system.

  Ethernet is a LAN (Local Area Network) often used for computer networks. “Ethernet” is a registered trademark of Xerox Corporation.

FIG. 11 is a conceptual configuration diagram when a communication station that performs communication according to the TCP / IP protocol is mounted on an Ethernet bus (hereinafter referred to as an E bus). This figure conceptually shows the configuration of software and hardware.
The communication station ST is connected on the E bus B.
The communication station ST is connected to the E bus B via the network adapter NA.

TCP and UDP are on top of IP. The communication station ST performs communication according to these protocols.
UDP (User Data Protocol) is used to transmit a unit of data. The communication between the transmitting side and the receiving side is only one data transmission, and there is no recovery function for communication failure. In TCP, a connection is established between a transmission side and a reception side, and recovery after communication failure can be performed. In both protocols, the other party is recognized by the IP address and the protocol port number.
There is a user application program AP on top of TCP and UDP.

FIG. 12 is a diagram showing the structure of a communication frame in the case of TCP.
The network adapter NA receives only the frame addressed to itself in order to reduce the burden on the software on it. Therefore, a 6-byte destination address (referred to as a MAC address) is added to the Ethernet header shown in FIG. The MAC address is unique to the network adapter and is specified so that there is no duplication throughout the world. The TCP / IP transmission unit is called an IP packet and is stored in the data portion of the Ethernet frame. TCP / IP communication is designed to be transmitted through media other than Ethernet, such as ATM (Asynchronous Transfer) and WAN (Wide Area Network). Therefore, the IP packet has an Internet address (referred to as an IP address) as a destination.

  As described above, when a communication station that performs communication according to the TCP / IP protocol is connected to the E bus through the network adapter, the communication station has a MAC address and an IP address. IP recognizes only one MAC address for one communication station. This has been a bottleneck in making the communication system redundant.

  Some distributed control systems have an E-bus. Some distributed control systems have a redundant configuration in order to improve reliability. The fact that IP recognizes only one MAC address for one communication station has been an obstacle to redundancy in a distributed control system with an E bus.

Conventionally, when redundancy is performed in a communication control system in which communication stations that perform communication using the TCP / IP protocol are connected on the E bus, the configuration shown in FIG. 13 has been adopted. FIG. 13A shows an example using TCP, and FIG. 13B shows an example using UDP.
In FIG. 13, the E bus has B1 and B2 drawn. Redundancy from the network adapter NA to TCP or UDP is performed on the communication transmission side and reception side. The user application program AP selects which of the redundant paths is used. Since the IP address and the MAC address are in a 1: 1 relationship, the IP and the TCP or UDP on the IP address must be made redundant. The redundancy control is performed by the user application program AP.

JP-A-4-27239 JP-A-9-261235

However, this conventional example has the following problems.
(1) The redundancy control must be performed by the user application program, which is a load.
(2) When TCP is duplicated, resources (memory) such as connection management are required twice.
(3) TCP is a stream-type communication, and is communicated as a simple flow of data. Therefore, when the TCP connection is duplicated, it is difficult to correlate switching and data breaks.
(4) Communication for maintaining the TCP connection is necessary, and duplication increases the load on the network.
(5) When UDP is duplicated, the above problem is eliminated, but the order of data cannot be understood.
(6) With UDP duplication, data cannot be recovered even if it is lost.
(7) In order to solve the above problem, the user application program requires a recovery process, and the load is large.
(8) A general TCP / IP application does not assume such redundancy, and only the special application can enjoy the effect of redundancy.

  The present invention has been made to solve the above-described problems. By switching the redundancy portion, the MAC address recognized by the IP is transferred to the IP. An object of the present invention is to realize a communication control system in which a user application program does not need to be aware of redundancy even when a redundancy configuration is adopted when a communication station that performs communication using the TCP / IP protocol is connected on an E bus.

  In order to achieve such a subject, the present invention is configured as follows.

(1) A communication station is connected on the network, and this communication station is provided with a communication control means. The communication control means performs communication according to the TCP / IP protocol, and only one MAC address for one partner communication station. In a communication control system that recognizes
Only one communication control means is mounted on one communication station,
One communication station is provided with redundancy, one is a normal side, the other is a standby side ,
A network controller that is interposed between this port and the communication control means, has a MAC address, and transmits / receives a communication frame input / output to / from the port;
A port control means for connecting a network controller with a MAC address recognized by the communication control means to a new normal-side port when the regular-side port is switched;
Diagnostic means for transmitting a diagnostic frame for diagnosing the network from both the regular side and standby side ports at a predetermined cycle;
A communication control system comprising:

(2) The diagnostic means transmits an ICMP echo as a diagnostic frame at a predetermined cycle, and determines that an abnormality has occurred if no response is returned even after a time longer than the predetermined cycle. (1) The communication control system according to (1).

(3) The communication control system according to (1), wherein the diagnosis unit has different periods for transmitting diagnosis frames between the regular port and another port.

As is apparent from the above description, the present invention has the following effects.
According to invention of Claim 1, the following effect is acquired.
(1) Do not use special hardware. Commercially available network adapters can be used as they are for personal computers, etc. (2) Non-redundant stations can also coexist in this system (becomes the regular side), so the overall cost can be reduced.
(3) Since no special function is required above IP, it does not become a load on the user application program.
(4) Since the user application program is unaware of redundancy, other conventional applications can enjoy the advantages of redundancy without modification.
(5) Since the TCP connection is not redundant, a large amount of memory is not required. For example, several tens of bytes are enough for the MAC address management table per station.
(6) The TCP connection is maintained even when the path is switched. The effect of the failure does not reach the user application program.
(7) Network communication load does not increase. In the redundancy algorithm, the cables are often independent.
(8) It can coexist with a conventional redundant network.
(9) Switching from a failed port (or route) can be performed smoothly.

Hereinafter, the present invention will be described in detail with reference to the drawings.
FIG. 1 is a conceptual block diagram showing an embodiment of the present invention. In FIG. 1, the same parts as those in the previous figure are given the same reference numerals.

In FIG. 1, a communication station STN is connected on E buses B1 and B2.
In FIG. 1, the communication station STN has two network adapters NA1, NA2 and a single TCP / IP, and redundant software S is provided between them. The redundancy software S has a MAC address conversion table T inside. On top of TCP / IP, there is a normal user application program AP that uses TCP / IP functions.

FIG. 2 is a diagram showing a specific configuration example of FIG.
The network adapters NA1 and NA2 are redundantly provided in one communication station, and a unique MAC address is assigned to each.
One communication control unit 1 is installed in one communication station, and performs communication according to a TCP / IP communication protocol.

The conversion table memory 2 stores an address conversion table.
The redundancy control means 3 is interposed between the network adapters NA1, NA2 and the communication control means 1, and when there is a transmission request from the communication control means 1, the destination MAC address in the transmission request is set as the communication partner. In addition to the MAC address of the network adapter to be received, when receiving, the MAC address of the transmission source attached to the received frame is converted to the MAC address recognized by the communication control means 1 of the own station. The redundancy control means 3 executes the redundancy software S.

The AP execution means 4 executes the user application program AP using the TCP / IP function.
The diagnosis unit 5 diagnoses whether communication is normally performed.

The operation of this embodiment will be described.
FIG. 3 shows a network in which three redundant communication stations are provided. Stations A, B, and C each have two network adapters NA1 and NA2, and their MAC addresses are MA1, MA2, MB1, MB2, MC1, and MC2. The three stations are connected by redundant E buses B1 and B2, but do not depend on the bus connection. Works even when connected to a single bus.

The MAC address conversion table shown in FIG. 4 is defined in the redundancy software for three stations. With this conversion table, each network adapter determines the communication partner.
The communication control means 1 that performs communication using the IP protocol recognizes the MAC address (MA1, MB1, MC1, in this example) of the partner station on the side to which “1” is added. In order to know the MAC address of the other station, it broadcasts an Ethernet frame to know the IP address, and the station with that IP address responds (ARP: Address Resolution Protocol) and receives it. Sometimes this can be done by passing the MAC address of one side to the IP.

When there is a transmission request from the station A to the station B, the IP of the station A passes the IP packet to be transmitted to the redundancy software and the other party's MAC address (MB1). Communication is performed in the following procedure.
(1) When the redundancy software of the station A transmits to the 1 side, it transmits to MB1.
(2) When the redundant software of station A transmits to the second side, it transmits to MB2 according to the address conversion table.
(3) When the station B receives a frame from the MB1 side, the transmission source network adapter attaches its own MAC address (MA1) to the frame. Station B uses MA1 as it is from the conversion table.
(4) When a frame is received from the MB2 side at the station B, the source network adapter attaches its own MAC address (MA2) to the frame. Station B uses MA1 as the source MAC address from the conversion table.
Through the above operation, the transmission data from the station A reaches the station B. Even if the network adapter used for communication is switched, the MAC address on the one side is always handled in the IP on the transmission side and the reception side.

Which network adapter is selected on the transmission side and which network adapter is sent to the IP on the reception side depends on the following redundancy algorithm.
In the first redundancy algorithm, the transmission side always transmits to one regular network adapter. When communicating between two stations, the service side may be different in each direction. Since the same communication frame is not received redundantly, the redundant software on the receiving side performs only the MAC address conversion and sends all received frames to the IP.

  The redundancy software may alternately use transmitting network adapters. If an abnormality is found in the route, do not use it. This will propagate the frame except for one time immediately after the failure.

  The transmission side may be used with the regular side fixed. Switch to the standby side only when an abnormality is found on the regular path.

  The network adapters for the regular side and the standby side may be switched and used. The switching may not be performed every time, but may be performed every appropriate cycle or number of transmissions.

  In order to actively investigate path failures, Ethernet frames may be exchanged with the other party at appropriate intervals and used for path diagnosis. The diagnosis may be performed on both the regular side and the standby side.

  For the diagnosis, ICMP echo included in the Internet standard protocol may be used. In this case, when an echo request is transmitted to the partner station IP, the partner returns an echo response. A diagnosis is made by using the return of this response.

  Diagnosis may be performed for all MAC addresses on Ethernet. Since all cases are exhausted, normal transmission is always guaranteed.

  You may send a diagnostic frame only to the communication station required for route reservation. For example, if there is a router or switch having a unique IP address on the network, a diagnostic frame is transmitted only to the router or switch in order to secure a route to the router or switch. In this way, the diagnostic frame is not unnecessarily sent over the network, and network congestion can be alleviated.

  In the second redundancy algorithm, when transmitting, both network adapters are requested to transmit after changing the MAC address of the transmission destination. Since the same frame is received redundantly at the receiving station, redundant frames may be removed at the receiving station.

  The receiving side may use the fixed side fixedly, pass only the frame from the network adapter to the IP, and receive from the other may be used only for the purpose of diagnosis. Switch to the standby side only when an abnormality is found on the regular path.

  The network adapter to receive is switched alternately or at an appropriate cycle. Receive from others is used for diagnostic purposes only. When an abnormality is found in the receiving side route, it switches to the standby side.

  If there is no reception in the regular reception adapter but reception is in another adapter, it may be switched because there is an abnormality in the path of the regular adapter.

  Only frames from the adapter that received the frame first may be passed to the IP. Receive from others is used for diagnostic purposes only.

  In addition to the MAC address described above, a virtual MAC address may be defined. The use of the virtual MAC address is a function independent of the above-described redundancy algorithm and can coexist with these. The MAC address conversion table includes virtual MAC addresses. The virtual MAC address is the MAC address of the communication partner used by IP, and does not need to be correlated with the actual MAC address of the partner or the partner does not need to know the virtual MAC address.

FIG. 5 shows an example of a MAC address conversion table on which virtual MAC addresses are placed.
IP is only informed of the virtual MAC address. Informing the IP of the virtual MAC address is performed using the ARP described above. When a transmission request to station B occurs at station A, the redundancy software of station A is given an IP packet to be sent to MAC-B. At this time, when sending to the 1 side, the MAC address is changed to MB1, and when sending to the 2 side, the MAC address is changed to MB2. On the station B side, the MAC address is changed to MAC-A and transmitted to IP regardless of whether the source MAC address is MA1 or MA2.

  By doing so, the MAC address used by the IP does not depend on the hardware, and recovery from a failure becomes easy. As an example, let us consider a case where the network adapter on the one side of the station A is failed and replaced. The path 1 side is out of order, but communication continues. If one side network adapter is replaced here, the new adapter will have a different MAC address, eg, MA3. In the address conversion table of FIG. 4, when MA1 of the MAC address conversion table of the station B is changed to MA3, the MAC address MA1 used by the IP does not exist in the MAC address conversion table, and problems such as disconnection occur. To do. When the virtual MAC address is used, the specific MAC address of the other party is simply rewritten, and can be exchanged without affecting the IP operation.

  As the selection of the virtual MAC address, the MAC address that was used first may be used. For example, in the above table, the MAC address conversion table of the station B is initialized as MAC-A = MA1, MAC-C = MC1. This can save time and labor associated with determining the MAC address. After the actual MAC address changes due to a failure, only the virtual MAC address has the original information.

  The MAC address conversion table may be updated using the MAC address received from each network adapter. Since the ARP address is always determined when the system is started up, the MAC address conversion table can be automatically generated. In addition, when some network adapters are replaced, new adapters can be incorporated into the redundancy mechanism by receiving frames from the adapters.

FIG. 6 is a conceptual block diagram of another embodiment of the present invention.
In FIG. 6, there are two ports PA1, PA2 and a single TCP / IP, and there is a port management means PT between them. The port management means PT can be implemented as a part of a device driver on a normal operating system. On top of TCP / IP, there is a normal user application program AP that uses TCP / IP functions.

FIG. 7 is a diagram showing a specific configuration example of FIG.
Ports PA1 and PA2 only input and output communication frames and do not have MAC addresses. Network controllers NC1 and NC2 connected to the higher level of ports PA1 and PA2 have MAC addresses.
The port control means 6 connects the network controller with the MAC address recognized by the communication control means 1 to the new normal side port when the normal side port is switched.

The operation of this embodiment will be described.
FIG. 8 shows a network in which three redundant communication stations are connected. Stations A, B, and C each have two ports, and the MAC address of the network controller connected to each port is MA1, MA2, MB1, MB2, MC1, MC2, and the MAC address on the regular side is 1 (this In the example, it is MA1, MB1, MC1,). In the figure, three stations are connected to a single network, but this is not a limitation. If a route can be set between arbitrary ports, a separate network connected by the bridge B may be used.

  As an example, consider the case where station A and station B communicate. Both regular ports are PA1, and the MAC addresses of the network controllers connected to the regular ports are MA1 and MB1, respectively. The TCP connection is specified by four parameters: the IP address of station A, MA1, the IP address of station B, and MB1. Consider a case where an abnormality is detected in the regular side port of station A. Since the station A has detected an abnormality on the service side, the port management means PT connects the network controller whose MAC address is MA1 to the port PA2 and continues communication. As a result, when viewed from the station B, neither the other party's IP address nor the MAC address changes, so communication is continued. The MAC address of the network controller connected to the port PA1 of the station A is MA2 in this example, and recovery is awaited by continuing diagnosis using this MAC address.

Elements that exist in the route between two specific stations A and B that communicate with the port on the service side are:
(A) Transmission circuit of station A (b) Network element (cable etc.) between regular ports between two stations AB
(C) Station B receiving circuit (d) Station B transmitting circuit (e) Station A receiving circuit. In this embodiment, the factors of port switching at the station A are (a), (e), and (b), and other factors are recovered by the switching at the station B side.
In the E bus, communication between specific stations occurs only when necessary. For diagnosis, the station A sends an ICMP echo to another station. If a response is returned, it can be seen that the network between the transmitting / receiving circuit of station A and the partner station is normal.

There may be a plurality of abnormality detection algorithms.
(1) It is assumed that there is an abnormality in (a) or (e) when reception is not performed for a certain time longer than the diagnosis cycle.
(2) There may be an abnormality in (b) to (d) in a station where the echo response for diagnosis does not return. If it does not recover after waiting for a certain period of time, there is a possibility of (b), so the local station side switches.
A similar diagnosis is necessary on the standby side, but the cycle may be longer than that on the regular side.

  Diagnosis requires only sending a diagnostic frame to the devices necessary for securing the route. For example, if there is a router or switch having a unique IP address on the network, a diagnostic frame is transmitted only to the router or switch in order to secure a route to the router or switch. Of course, all the other stations may be diagnosed.

  If there is a bridge or switch that operates according to the destination MAC address on the network, the destination must be changed when the port is switched. Immediately after the port is switched, echo should be sent to the other station to forcibly switch the destinations of these bridges and switches.

9 and 10 are diagrams showing a configuration example of the port management means PT.
In FIG. 9, a MAC address is designated for each of the network controllers NC1 and NC2, and which port is used is switched by the switches SW1 and SW2. The switch SW1 and SW2 are switched by the port control means 6. In this configuration example, the time required for switching is short. The network controllers NC1 and NC2 are for transferring communication frames between the communication control means 1 and the E buses B1 and B2.

In FIG. 10, the network controllers NC1 and NC2 and the ports PA1 and PA2 are combined on a one-to-one basis, and the MAC address is changed by the port controller 6 setting an address in each of the controllers NC1 and NC2. In this configuration example, the switching time becomes longer, but it can cope with the failure of the regular controller.
Note that both of FIG. 9 and FIG. 10 may be realized simultaneously.

It is a conceptual block diagram which shows one Example of this invention. It is the figure which showed the specific structural example of FIG. It is operation | movement explanatory drawing of this invention. It is operation | movement explanatory drawing of this invention. It is operation | movement explanatory drawing of this invention. It is a notional block diagram of the other Example of this invention. It is operation | movement explanatory drawing of this invention. It is operation | movement explanatory drawing of this invention. It is the figure which showed the structural example of the port management means. It is the figure which showed the structural example of the port management means. It is a conceptual block diagram at the time of mounting the E bus | bath on the E bus | bath the communication station which communicates according to the protocol of TCP / IP. It is the figure which showed the structure of the communication frame. It is a block diagram of the prior art example of the communication control system which took the redundancy structure.

Explanation of symbols

STN station B1, B2 bus NA1, NA2 network adapter PA1, PA2 port NC1, NC2 network controller 1 communication control means 2 conversion table memory 3 redundancy control means 5 diagnostic means 6 port control means

Claims (3)

A communication station is connected on the network, and this communication station is provided with a communication control means. The communication control means performs communication according to the TCP / IP protocol and recognizes only one MAC address for one communication station. Communication control system
Only one communication control means is mounted on one communication station,
One communication station is provided with redundancy, one is a normal side, the other is a standby side ,
A network controller that is interposed between this port and the communication control means, has a MAC address, and transmits / receives a communication frame input / output to / from the port;
A port control means for connecting a network controller with a MAC address recognized by the communication control means to a new normal-side port when the regular-side port is switched;
Diagnostic means for transmitting a diagnostic frame for diagnosing the network from both the regular side and standby side ports at a predetermined cycle;
A communication control system comprising: The diagnostic means transmits an ICMP echo as a diagnostic frame at a predetermined cycle, and determines that an abnormality has occurred if no response is returned after a time longer than the predetermined cycle. The communication control system according to claim 1. The communication control system according to claim 1, wherein the diagnostic unit has different periods for transmitting diagnostic frames between the regular port and another port.

Images