JPS60119150A - Data transfer control method in computer network system - Google Patents

Abstract

PURPOSE:To prevent the generation of dead lock or the reduction of the transfer efficiency of a data bus due to unsuitable data transfer to an unreceivable computer by recognizing the status of each computer in a network through a circuit interface. CONSTITUTION:Each circuit interface Ii monitors the receiving ability of a computer Ai included in the interface Ii itself, and when deciding that the computer is disabled to receive, turns on FDi display on a control loop and informs the disabled result to other circuit interfaces. Each circuit interface Ii checks the receiving ability of the computer Ai on the data receiving side on the basis of the FDi display on the control loop, and if the computer is disabled to receive, sends no packet to a data bus 80. Only when the reception is abled, the packet is sent to the data bus 80.

Description

[Detailed Description of the Invention] [Technical Field of the Invention] The present invention relates to a data transfer control system in a star network system in which a plurality of computers transfer data via a communication line and an exchange having a data storage function, and in particular, A data transfer control method that prevents the occurrence of digital lock and a decrease in data transfer efficiency by providing a switch with a function that stops data transfer to a computer that cannot receive data in the network when it detects a computer that cannot receive data. Regarding.

[Technology background]

First, conventional technology will be explained.

FIG. 1 shows a conventional example of a star computer network system to which the present invention is directed. In the figure, 1 is a computer Ao, and 2 is a computer Az, a.

4 is a line, 5 is a data transfer network, and 6 is a line interface. , 7 is the line interface IL, 8 is the data bus, 9 is the sending buffer SB0, 10 is the receiving buffer R
Bo, 11 is a communication control unit CC (+, 12 is a control loop control unit Go, and 13 is a transmission buffer SBj.

14 is a reception buffer RBt, 15 is a communication control unit 00t
, 16 is Control Roof IIJ Ot-1
7" control loop, and 18 indicates a control loop return control unit. In addition, O and t are 2 at both ends when the numbers of t+1 network terminals are 0.1, 2, . . . , t. Note that when representing any one terminal, a number such that 0 i < 1 is representative.

For example, data transferred between computers (Ao, Az) is transferred to the transmission buffer (SBO) of the line interface.
, 5Bt) and receive buffers (RBo, RBz)
The packets are assembled into packets with a length that can be stored in the data bus, and are transferred via the data bus by the communication control unit (COo, OOz).

The packet has the format shown in Figure 2, and the flag is a special bit string that represents the node delimiter, and is set to “01”.
111110'' is used.The address is the address of the calculation or line interface of the destination of the mother packet.The data is the data to be transferred.

Between computer AL and line interface Xi (7) =
Kerato transfer is performed by transmitting the following three control signals through a line to control the reception buffer RBi and transmission buffer SBi.

■ ・Can send kerato (S, CrrarLt)
fF3Bi notifies Ai that there is no packet in signal F3Bi and it is possible to send packets from AL to SBi 0 Notifies Ai from cRBi that 74 packets have arrived at RBi, and receives AiK packet 'fc seek

(2) A'ket reception preparation completion (R, Grant) signal ki notifies RBi that preparation for packet reception has been completed and requests RBi to send the packet.

Next, during packet transfer between the line interface on the sending computer side and the line interface on the receiving computer side, all tokens shown in FIG. 3 are transmitted to the control loop 17 and controlled. The F in the illustrated token is a special pip) four turn indicating the frame delimiter, and the DXDB
indicates that the data bus 8 is in use. Also R
Bo to RBA are each line interface Xo to I
t's received data RB, ) to RBt's respective B
Displays the status of U8Y/IDLE. For convenience, the ODB display uses the same symbol as the RB o to RBtB4 display in the token and the reception buffer to prevent collision of thousands of packets on the data bus, and the RBo to RBA display uses the same symbols as the reception buffer and the mother packet in the reception buffer. Used to prevent disappearance.

Monitoring and control of each display in the token is performed by the communication control section 00i and the control loop control section Ci in each line interface Ii.

FIG. 4 shows a data transfer procedure in the conventional network system shown in FIG. 1 described above.

The figure shows an example in which a packet is transferred to either Ao or At.

Engineering. When RBo sends a 7Q packet sending ready (3, Graalt) signal to Ao, Ao learns that RBo is empty and sends a data packet. The data packet is line 3
and then stored in RB.

Here, CCo checks the free status of each of the data bus 8 and the destination RBt using the token via co,
If both are empty, the data packet is transferred to RBt. As a result, engineering A sends a packet reception request (R, 1g9) signal to AA'), causing A1 to prepare for reception.

Next, At is ready to receive (R,
C, rctnt ) signal to cause RBtK to send out a certain data packet. At receives this and the transfer of the packet is completed.

Such a conventional system has the advantage that computers can be easily connected by simply installing a line interface in the data transfer network 5, and that packets are not lost during transfer or unnecessary transfer is performed. Although I have
On the other hand, if hl is unreceivable for some reason,
When attempting to send hl heno or kerat from AQ, there is a problem that the following deadlock or reduction in transfer efficiency occurs.

FIG. 5 shows an example of a deadlock situation.

Assume that AO has a request to transfer 9 kets P, , P2 to At, and then transfer them to Pa k A77Z, and the other At is unable to receive them. In this case, Ao
The packet P1 sent out first is stored in RBt, but is not received by At, and the next packet P2 remains in SBo because RB7 is not available. Therefore, Ps, which should be transferred from Ao to jbx, cannot be sent from Ao, resulting in a deadlock.

In addition, even if At becomes unreceivable, other computers or line interfaces will not be able to know this, so packets destined for AA will still be sent to the data transfer network9. This results in a decrease in efficiency and, therefore, a decrease in the overall system/mother packet transfer efficiency. Such a state in which a computer cannot receive data is caused by a power outage, an overload state, a state in which data transfer software is not running, or the like.

[Important and configuration of the invention]

An object of the present invention is to prevent deadlocks and deterioration of data transfer efficiency by preventing data from being sent to an unreceivable computer from an exchange when transferring data from one computer to another.

Therefore, in the present invention, the line interface on the sending side is
It provides a means to identify the status of each computer, and its configuration is such that multiple computers are connected via communication lines to an exchange with a data storage function, and data is transferred between the multiple computers. In the computer network system, the switching equipment is composed of a plurality of line interfaces having a data storage function corresponding to each computer, and a data transfer network that transfers data between the plurality of line interfaces. At least some of them include means for detecting the data reception function status of the corresponding computer, means for notifying the detected result to another line interface, and when transferring data to another line interface, It is characterized by comprising means for identifying the reception function status of the corresponding computer notified from the other line interface, and means for controlling the entire execution of data transfer based on the identified result.

[Embodiments of the invention]

The details of the present invention will be explained below based on examples.

FIG. 6 is a block diagram of a system according to an embodiment of the present invention. Elements designated by reference numbers 1 to 18 in this figure are equivalent to elements with the same numbers shown in the conventional system of FIG. 1, and therefore, description thereof will be omitted here.

New requirements added to the present invention are 19 and 20.
flag detection unit hD,.

The FDt is provided within each line interface.

FIG. 7 shows the control loop 17 used in this embodiment.
Shows the format of the above token. This differs from the conventional token format shown in FIG. 3 in that it has an FDi status indication paired with each RBi status indication. Nao'yD
i indicates the reception function state of the corresponding computer Ai, and is indicated using the same symbol as the flag detection unit FDi for convenience.

The HDi display is reset by the flag detection unit FDt. Computer hi is normally ready for reception (R.

Grant) signals, packets, or flags are always sent on the line. The flag detection unit FDi can utilize the above characteristics, for example,
The continuity of the flag output is monitored under the following conditions, and when it is detected that the computer ki is unable to receive data, the FDi display of the control loop is set to ON.

In other words, if the signal sent from At is as shown in Figure 8, the flag output will be interrupted at ■, but if the packet is within the maximum allowed packet time, the FD
Do not set the L display to ON. It is assumed that the power to Ai- is turned off in the next step (2). Because of this, the flag is cut off, but
Do not set the FDL display to ON because there is a possibility that it is a packet.If the flag is still not detected even after the above maximum node time has elapsed, set the FDi display to ON for the first time. do.

In this way, each line interface A monitors the reception capability of the computer A2 to which it belongs, and when it determines that reception is not possible, the FDi display i01 on the control loop
Notify other line interfaces by setting it to +1. Each line interface Ij is connected to the data receiving side.

Check the reception ability of the FD on the control loop and the display, and if reception is not possible, send the A-button to data bus 8.
r Take data path 8 when sending is not possible and reception is possible.
Send 9 keets to.

The data packet transfer procedure in this embodiment is basically the same as that in the conventional example explained in FIG.

When the sending side line interface Ez receives the packet from Az, it checks the FD in addition to the DB and RB on the control loop, and as mentioned above, it checks if the DB and RBj are empty and FDjo: OFF. Then Aj
determines that the packet can be received and forwards the packet.
FD: If ON, it is determined that A is unreceivable, the transmission buffer SBif is reset, and no packet transfer is performed.

In the above-described embodiment, all line interfaces were provided with the function of detecting and notifying the reception capability of the corresponding computer, but it is also possible to provide only some line interfaces with this function. In this case, the functionality is lower than when all line interfaces have the same function, but it is possible to suppress an increase in the cost of the entire system and an increase in the hardware scale.

〔Effect of the invention〕

As described above, according to the present invention, since the state of each computer in the network is recognized by the line interface, deadlocks may occur due to inappropriate data transfer to a computer that cannot receive data, and data bus It is possible to prevent a decrease in transfer efficiency.

[Brief explanation of the drawing]

Figure 1 is a diagram showing a conventional example of a star computer network system, and Figure 2 is A? Figure 3 is an explanatory diagram of the token format on the control loop, Figure 4 is an explanatory diagram of the data transfer procedure of the conventional system shown in Figure 1, and Figure 5
Figure 6 is an explanatory diagram of a deadlock state, Figure 6 is a block diagram of a system according to an embodiment of the present invention, Figure 7 is a diagram of a token used in the system of the embodiment shown in Figure 6, and Figure 8 is a diagram showing a failure. 0 is an explanatory diagram of signals sent from the generated computer ki. In the figure, 1.2 is the computer, 3.4 is the line, 5 is the data transfer network, 6.7 is the line interface, 8 is the data bus,
9 is a transmission buffer, 1o is a reception buffer, 11 is a communication control section, 12 is a control loop control section, 17 is a control loop, and 19 is a flag detection section. Patent applicant Fujitsu Ltd. Representative Patent Attorney Fumihiro Hase (1 other person)

Claims (1)

[Claims] In a computer network system in which a plurality of computers are connected via communication lines to an exchange having a data storage function, and data is transferred between the plurality of computers, the above-mentioned exchange is connected to a plurality of exchanges having a data storage function corresponding to each computer. and a data transfer network that transfers data between the plurality of line interfaces, and at least some of the plurality of line interfaces detect the data reception function status of the corresponding computer. means for notifying the detected result to another line interface; and identifying the receiving function state of the corresponding computer notified from the other line interface when data is transferred to the other line interface. 1. A data transfer control method in a computer network system, comprising: means for controlling execution of data transfer based on the identified result.