JPH07182248A - Highly reliable information processor - Google Patents

Abstract

PURPOSE:To quickly execute updating processing for memory data and recoverying processing at the time of generating a fault in a highly reliable information processor provided with a duplicated memory unit. CONSTITUTION:A data updating request is simultaneously sent from a data processing unit to main memory and sub-memory units 2a, 2b. Each of the units 2a, 2b is provided with an auxiliary storage part 12 for storing the data updating request sent from the data processing unit and an updating processing part 11 for executing the processing of the data updating request. The unit 2a immediately updates storage data stored in a memory part 13 by an updating control part 11 based upon the data updating request sent from the data processing unit. On the other hand, the unit 2b stores the data updating request from the data processing unit in the storage part 13, and when it is instructed to end the data updating request, it updates the data stored in the memory part 13 by the data updating request stored in the storage part 12.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a highly reliable information processing device having a duplicated memory unit.

[0002]

2. Description of the Related Art Conventionally, an information processing apparatus which is required to have high reliability is provided with two memory units in order to prevent data loss and system stoppage due to a memory unit failure. One of them is a main memory unit and the other is a sub memory unit to hold the same data. Here, the updating of the data held in the memory unit is conventionally serialized such that the updating of the main memory unit is completed before the updating of the sub memory unit. This ensures that if there is a failure during the update process and the update of the data is incomplete, that there is normal data in either the main memory unit or the sub memory unit. Can be recovered. If a failure occurs during the updating of the data in the main memory unit, the state of the data can be returned to the state before the update by copying all the data in the sub memory unit to the main memory unit.
Further, if a failure occurs during the updating of the data in the sub memory unit, the main memory unit has the updated data. Therefore, the update processing can be completed by copying this to the sub memory unit.

[0003]

SUMMARY OF THE INVENTION In the above conventional method,
Since the data update processing must be serialized in the main memory unit and the sub memory unit, the data processing unit must repeatedly send the data update request twice, which hinders the processing of the data processing unit.
When the occupancy of the communication path for sending the data update request becomes high and the communication path is shared by a plurality of units, the communication of other units is hindered.

Further, when there are a plurality of data processing units, the data update requests from different data processing units cannot be distinguished. Therefore, in order to be able to recover from a failure, the data update processing by these data processing units must also be performed. Must be serialized. Therefore, when the number of data processing units increases, the data update processing becomes a major obstacle to improving the performance. Also, when recovering from a failure, since it is not known at which position the data was destroyed, the entire contents of the memory unit must be copied, which takes a long time and delays the restart of normal processing. .

The present invention has been made in view of the above circumstances, and is provided with a main memory unit and a sub memory unit capable of quickly performing a memory data update process and a recovery process when a failure occurs and improving the processing efficiency. An object is to provide a highly reliable information processing device.

[0006]

A highly reliable information processing apparatus according to the present invention is a highly reliable information in which data sent simultaneously from a data processing unit is redundantly held in first and second memory units. In the processing device, each of the first and second memory units has an auxiliary storage unit for storing a data update request from the data processing unit and an update processing unit for processing the data update request added to the memory unit. , The first memory unit immediately updates the stored data in the memory unit by the update control unit based on the data update request sent from the data processing unit, and the second memory unit outputs the data from the data processing unit. Data update request stored in the auxiliary storage unit, and the data update stored in the auxiliary storage unit when the end of the data update request is instructed. Characterized by updating the data stored in the memory unit by seeking.

[0007]

The data processing unit sends data update requests simultaneously from the start command section to the first and second memory units. The update control unit of the first memory unit stores the received packet of the data update request in the auxiliary storage unit and at the same time reflects the update request in the stored data of the memory unit, and when the end command unit of the data update request is received. The data update request packet stored in the auxiliary storage unit is discarded.

On the other hand, the update control unit of the second memory unit stores the received data update request packet in the auxiliary storage unit without reflecting it on the held data, and collects it when the end command unit is received. The update request is reflected in the stored data in the memory unit, and the stored data update request packet is discarded.

Processing for returning the data state of the first memory unit to the state before update when a failure occurs while the data update request is being transmitted and the data update request is sent only halfway Do. That is, when a failure occurs, the data processing unit starts failure recovery processing,
Inspect the auxiliary storage of the first memory unit. When an uncompleted data update request is found by this, the data processing unit fetches the address at which the data is updated from the data part of the data update request, and obtains the pre-update data of the same address from the second memory unit. By reading and writing to the first memory unit, the first
Recover data in memory unit. The uncompleted data update request stored in the auxiliary storage unit is discarded after the address is fetched.

By performing the above processing, the data update request of the first memory unit and the second memory unit can be completed at one time while the recovery processing at the time of occurrence of failure can be performed. The data update request can be released early and the data processing can be restarted.

[0011]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a configuration example of a high reliability information processing apparatus according to an embodiment of the present invention.

In FIG. 1, reference numerals 1a and 1b are data processing units, and memory units 3a and 3b are connected via a communication path 2.
Connected to. The data processing units 1a and 1b read data from the memory units 3a and 3b, process the data, and request data updating to the memory units 3a and 3b.

The memory units 3a and 3b are made into a set of a main memory unit and a sub memory unit according to the designation from the data processing units 1a and 1b, and hold data of the same contents.

The memory units 1a and 1b are constructed as shown in FIG. In FIG. 2, reference numeral 11 denotes an update control unit, which is set by the data processing units 1a and 1b to control the memory unit to operate as either the main memory unit or the sub memory unit. Reference numeral 12 denotes an auxiliary storage unit, which is a data processing unit 1
This is for storing the data update request sent from a and 1b. A memory unit 13 is a unit for storing data.

FIG. 3 shows the structure of a data update request sent from the data processing units 1a and 1b to the memory units 3a and 3b. This data update request comprises a start command section 21, a data section 22, and an end command section 23. Data part 2 in this data update request
Reference numeral 2 denotes an update address 24 indicating a position to update the data stored in the memory unit 13 of the memory units 3a and 3b, and update data 2 newly written by the update.
This is a set of 5 groups, which is necessary for the update process.
The update processing request includes a start command section 21, a data section 22,
Data is sent from the data processing units 1a and 1b to the memory units 3a and 3b in the order of the end instruction unit 23.

When a data update request is sent from the data processing units 1a and 1b, the memory units 3a and 3b receive the start command section 21 from the update control section 11 to start the update processing, and the data section 22. Update processing by. Then, when the update control unit 11 receives the end instruction unit 23, the update control unit 11 performs a process for ending the update process.

FIG. 4 shows another example of the structure of the data update request. In order to identify the data processing unit that has transmitted the data update request to the start command section 21, the data section 22, and the end command section 23. Identification information part (PE-ID: Proc
essing Element-ID) 26 is added.

In this way, the identification information 26 is added to the data update request.
By adding the, for example, the data processing unit 1
Data update requests sent from a plurality of data processing units such as a, 1b, ... Can be distinguished, and data update requests from different data processing units can be processed in parallel.

Next, the operation of the above embodiment will be described with reference to the flowchart shown in FIG. FIG. 5 shows a memory unit 3
The operations of the update control unit 11 of a and 3b represent the operation of the update control unit 11 of the memory unit that has become the main memory unit after the main mode, and the operation of the sub memory unit after the sub mode. The operation of the update control unit 11 of the failed memory unit is shown.

When the information processing apparatus is activated, one of the data processing units 1a and 1b performs a mode setting process on the set of memory units 3a and 3b as shown in FIG. 5 (step A0). That is, with respect to the update control unit 11 of each memory unit, one of the memory units becomes the main memory unit (main mode) (step A
1) Make settings so that the other memory unit becomes the sub memory unit (sub mode) (step B1).
). Now, the memory unit 3a is set as the main memory unit and the memory unit 3b is set as the sub memory unit, and the memory units 3a and 3b are set by the data processing unit 1a.
A case of performing the data update process of is described.

When the data processing unit 1a needs to update the data held in the memory units 3a and 3b during the operation of the information processing apparatus, the data processing unit 1a issues a data update request as shown in FIG. make,
Through the communication path 2, the start command unit 21, the data unit 22, and the end command unit 23 are sequentially sent to both the main memory unit 3a and the sub memory unit 3b holding the data to be updated. In the memory units 3a and 3b, the update control unit 11 receives the start command unit 21 of the data update request to start the update process.

First, the update control unit 11 of the main memory unit 3a stores the received data update request packet in the auxiliary storage unit 12 and, at the same time, processes the data update request according to the flow after the main mode in FIG. That is, when the start command section 21 is received in step A2, it is judged whether or not the data section 22 is sent subsequently (step A2).
3), when the data part 22 is received, its update address 2
4 and the update data 25, the stored data in the memory unit 13 is updated (step A4). That is, the update processing is performed by writing the update data 25 in the memory unit 13 at the address position designated by the update address 24. This update process is continuously performed while the data section 22 is being sent from the data processing unit 1a. Then, when the end command section 23 is sent from the data processing unit 1a after the data section 22 (step A5), the update control section 11 sends the data update request packet stored in the auxiliary storage section 12 to the data update request packet. At the same time as discarding, it waits until the next data update request start command unit 21 is sent.

On the other hand, the update control unit 11 of the sub memory unit 3b processes the data update request in accordance with the flow shown in the sub mode of FIG. That is, when the update control unit 11 receives the start command unit 21 in step B2,
After that, it is judged whether or not the data section 22 is sent (step B3), and when the data section 22 is sent,
The data section 22 is stored in the auxiliary storage section 12 (step B4). That is, until the end command part 23 of the data update request is received, the data part 22 following the start command part 21.
Is stored in the auxiliary storage unit 12, and the data in the data unit 22 is not written in the memory unit 13. And
When the end command section 23 is sent from the data processing unit 1a after the data section 22 (step B5), the update control section 11 causes the data section 22 stored in the auxiliary storage section 12 to store the data in the memory section 13. While updating the memory content,
The data update request packet stored in the auxiliary storage unit 12 is discarded (step B6). After that, the process waits until the next data update request start command unit 21 is sent.

If a failure occurs while the data processing unit 1a is transmitting a data update request and the data update request is sent only halfway, the data state of the main memory unit 3a is updated. Perform processing to return to the previous state. That is, the data processing unit 1a starts the failure recovery process at the time when a failure occurs, and inspects the auxiliary storage unit 12 of the main memory unit 3a. When an uncompleted data update request is found by this, the data processing unit 1a takes out the address at which the data is updated from the data part 22 of the data update request stored in the auxiliary storage part 12, and stores it in the sub memory. The pre-update data of the same address is read from the unit 3b to read the main memory unit 3a.
Send to. Update controller 11 of main memory unit 3a
Uses the address and data sent from the data processing unit 1a to rewrite the stored contents of the memory unit 13 and restore the state before the update process is started. The uncompleted data update request stored in the auxiliary storage unit 12 is discarded after the address is fetched. This realizes the failure recovery processing.

Therefore, the main memory unit 3a and the sub memory unit 3b can be processed while the recovery processing when a failure occurs is still possible.
The data update requests can be completed at once, and the data processing units 1a and 1b can be quickly released from the data update request and resume the data processing. Also,
When a communication mechanism capable of sending a data update request to the main and sub memory units 3a and 3b at the same time by one communication operation as in the bus connection method shown in FIG. The amount of communication for updating data can be halved as compared with the conventional processing method that performs processing, and the load on the communication path can be significantly reduced.

Further, in the recovery processing when a failure occurs,
Since all the data held in the memory units 3a and 3b is not copied, but only the updated portion is copied, the time required for the recovery processing can be shortened.

FIG. 6 shows another control example of the update controller 11 in the main memory unit 3a. The process shown in FIG. 6 is different only in the process of step A4,
Others are the same as in the case of FIG. That is, when the data section 22 of the data update request is reflected in the memory section 13,
The pre-update data stored in the memory unit 13 is taken out and moved to the auxiliary storage unit 12, and then the data update process is performed. For example, for the packet of the data update request received by the main memory unit 3a, the pre-update data stored in the memory unit 13 is read and written in the update data 25 of the data unit 22, and then the auxiliary storage unit 13 is used. To store.

When a failure occurs and the data update request has not been transmitted to the end and the update process must be canceled, the data processing unit 1a instructs the main memory unit 3a to recover the data. . When the update control unit 11 of the main memory unit 3a receives the data recovery instruction from the data processing unit 1a, the data unit previously transferred to the auxiliary storage unit 12 is reflected in the memory unit 13 so that the memory unit The stored data of 13 is returned to the state immediately before the start of the update process.

As described above, by transferring the data stored in the memory unit 13 in the main memory unit 3a to the auxiliary storage unit 12 during the data updating process, normal operation from the auxiliary storage unit 12 in the main memory unit 3a can be performed when a failure occurs. Data can be read. Therefore, when returning the main memory unit 3a to the state before the start of the data update process due to a failure, the data processing unit 1a sends the update control unit 11 of the main memory unit 3a to the auxiliary storage unit
The recovery process is performed only by instructing the recovery of data from 2. Therefore, the data transfer from the sub memory unit 3b becomes unnecessary, and the time required for the failure recovery processing can be further shortened.

In the above embodiment, one of the data processing units 1a and 1b sets the main mode and the sub mode for the set of memory units 3a and 3b when the information processing apparatus is started. Alternatively, the main mode and the sub mode of the memory units 3a and 3b may be set by the operating system when the information processing apparatus is activated.

[0031]

As described in detail above, according to the present invention, 2
In a high reliability information processing apparatus having a duplicated memory unit, a memory unit includes an auxiliary storage unit for storing a data update request sent from a data processing unit and an update processing unit for processing the data update request. Since the data update request is added to the main memory unit and the sub memory unit, the data update request for the main and sub memory units can be processed at once with the recovery processing when a failure occurs. It becomes possible to restart the processing.

Further, by adding the identification information to the data update request, it becomes possible to distinguish the data update requests from different data processing units, and the data update requests from a plurality of data processing units are processed in parallel. You can
The efficiency of the data update process can be improved.

[Brief description of drawings]

FIG. 1 is a system configuration diagram of a highly reliable information processing apparatus according to an embodiment of the present invention.

FIG. 2 is a configuration diagram of a memory unit in the embodiment.

FIG. 3 is a diagram showing a configuration of a data update request in the embodiment.

FIG. 4 is a diagram showing another configuration example of a data update request.

FIG. 5 is a flowchart showing a processing operation of an update control unit in the memory unit.

FIG. 6 is a flowchart showing another processing operation of the update control unit.

[Explanation of symbols]

 1a, 1b Data processing unit 2 Communication path 3a, 3b Memory unit 11 Update control part 12 Auxiliary storage part 13 Memory part 21 Start command part 22 Data part 23 End command part 24 Update address 25 Update data 26 Identification information part

Claims (4)

[Claims] 1. A high reliability information processing apparatus for redundantly holding data sent simultaneously from a data processing unit in a first and a second memory unit, wherein the first and second memory units are memories respectively. An auxiliary storage unit for storing a data update request from the data processing unit and an update processing unit for processing the data update request are added to the unit, and the first memory unit is sent from the data processing unit. The update control unit immediately updates the stored data in the memory unit based on the data update request, and the second memory unit stores the data update request from the data processing unit in the auxiliary storage unit. When the end is instructed, the stored data in the memory unit is updated according to the data update request stored in the auxiliary storage unit. Highly reliable information processing device. 2. A high reliability information processing apparatus for overlappingly holding data sent from a data processing unit in first and second memory units, wherein each of the first and second memory units is a memory. An auxiliary storage unit that stores a data update request from the data processing unit and an update processing unit that processes the data update request are added to the unit, and the first memory unit is a data update request from the data processing unit. When the data is sent, the update control unit transfers the data before update in the memory unit to the auxiliary storage unit, and then updates the stored data in the memory unit, and the second memory unit receives the data from the data processing unit. The data update request is stored in the auxiliary storage unit, and when the end of the data update request is instructed, the data update request stored in the auxiliary storage unit is executed. A highly reliable information processing device characterized by updating stored data in a memory unit. 3. The high reliability information processing apparatus according to claim 1, wherein identification information for identifying the data processing unit that issued the data updating request is added to the data updating request, and a plurality of data processing units are provided. A highly reliable information processing apparatus, characterized in that the data update request of the above can be identified. 4. The high reliability information processing apparatus according to claim 2, wherein the first memory unit, when a failure occurs in the data update processing, transfers the data before update to the auxiliary storage unit in advance. A high-reliability information processing apparatus, characterized in that recovery processing from a failure is performed by returning the information processing unit to the unit.