WO2005103901A1 - Computer system - Google Patents

Abstract

[PROBLEMS] To improve performance and trouble resistance of a computer system. [MEANS FOR SOLVING PROBLEMS] An accelerator system holds a copy of a location table and an alternative key location table and a so-called index processing is executed in the accelerator system, there by distributing the load. This copy is realized with a low load by using broadcast communication. A large amount of communication between the accelerator systems (1-9) and a primary system (0) is performed smoothly by performing uplink communication by transmission urging. Thus, it is possible to improve the performance of the entire computer system.

Description

 Specification

 Computer '' System Technology

 [0001] The present invention relates to a computer system, and improves the performance of the computer.

 Background art

 [0002] Conventional computer systems improve the performance of a single CPU in order to improve performance, increase the main memory, and install multiple CPUs to increase the processing capacity. Have been. However, the performance improvement of a single CPU was slower than the increase in the processing amount, and the main factor was the performance improvement by installing multiple CPUs. This tendency was particularly remarkable in UNIX (registered trademark) servers. However, even if the contention of the main memory, the CPU, and the external storage device occurred and the resources were increased, the processing capacity did not increase proportionally. In particular, when the number of CPUs in a computer reached about 100, the resources required for its control increased, and it was almost impossible to improve the processing capacity. In addition, in order to improve the capacity of the computer, it is also necessary to separate the computer for the application and the computer for the database. The power database needs to be unified, and the database has become a bottleneck. .

[0003] The present inventor has proposed a location 'table and an alternate key instead of the conventional hierarchical index.

• Introduces the concept of a table, simplifies the complicated processing involved in index processing, and uses a method such as binary search to search the table itself, thereby ensuring high speed and easy maintenance. Invented the "data storage and retrieval method" (Patent Documents 1 and 2). Furthermore, in the "database reorganization system and database (hereinafter, referred to as" database reorganization system ") (Patent Document 3), the database is operated with respect to the database proposed in the" data retrieval storage method ". It proposes a mechanism that allows reorganization while doing so, and further invents that efficient reorganization is possible by adding an alternative key 'location' table to an alternative key 'table'. Also, "Day In the database accelerator function (hereinafter, referred to as “accelerator”) (Patent Document 4), when the “accelerator 1” system holds a copy of the location “table or alternate key” location “table” and searches for records. In addition, the present invention invents that the access can be processed in parallel by using the location table of the accelerator system and the location table of the alternative key. In the "data storage and retrieval system (Patent Document 5)", the connection from the primary 'block to the overflow' block and the overflow 'block power to the overflow ^ ~ · · block is performed using the overflow ^ ~ · block management table. Inventing the scheme. In the “data storage and retrieval system”, the alternate key block and the alternate key · overflow · block concatenation, the alternate key · overflow · block and the alternate key 'overflow' block are similarly connected. This is a method that uses a flow 'block management table. Furthermore, in the “communication system (Patent Document 6)”, the collision detection type communication, which is currently the mainstream of communication, has the following problems: The upstream communication is performed by transmitting the transmission reminder message from the rhythm node, and the mechanism is invented without collision.

[0004] Patent Document 1: Japanese Patent No. 3345628

 Patent Document 2: U.S. Patent No.6415375

 Patent Document 3: Patent application PCTZJP03Z11592

 Patent document 4: Patent application PCTZJP03Z13443

 Patent Document 5: # 112004 -020006

 Patent Document 6: Japanese Patent Application 2004-094628

 Disclosure of the invention

 Problems to be solved by the invention

The inventor of the present invention has invented an “accelerator” as a means for improving the processing capability of a computer system. This improves the processing capacity of the database processing computer, which has been a bottleneck in the past. In order to realize an accelerator system, it is necessary to have a copy of the location table and the alternate key location. Was only shown. The method is described in "Specification information transmission in this specification. Although the explanation is given in such a way that information is transmitted and received through the mechanism, the change information reflection mechanism, and the communication path, it can be realized by creating special hardware. When the information of a specific address is changed, the information of another address is automatically rewritten in the same manner, so that the location table of the primary system and the location table of the alternate key are changed. This is a mechanism (synchronization update mechanism) that updates the land location table and the land substitute key location table of the accelerator system when it is done. Was described.

 [0006] Implementing a mechanism for copying a location 'table or an alternate key' location 'table with low load as described here is not only a matter of improving the performance of an "accelerator" but also a computer. This is useful for improving the performance of the system. A large amount of communication between the accelerator system and the primary system (a system that processes data records). If this communication is not performed smoothly, the overall processing capacity will increase. Can't hope. In addition, efficient knock-up methods are important issues in operating a computer's system. In addition to efficient knock-up, the backup 'system is simply used as a backup' system. By using it as a reference system to improve the performance of the computer's overall system.

 Means for solving the problem

[0007] The present invention provides a data record having a data item including a primary key, a primary 'block for storing data records in the order of the primary key, and a location' table entry in which the address of each primary 'block is described. Maintains a primary system that has a location 'table' in a contiguous area, and a land location's table in which the address of each primary 'block is described. The system is equipped with an accelerator system and a communication channel. The primary system transmits change information by broadcast and receives the change information from the accelerator system. Against the limary system And transmitting the further completion notification, in Konbiyu data one 'system. [0008] The present invention also provides a data record having a data item including a primary key, a primary block storing data records in the order of the primary key, and a location table listing addresses of the primary blocks. · A primary system that has a location table with entries in a continuous area · A system with each primary block · A land with the address of the block · A location 'table · A land with entries in a continuous area · A mouth case The system has a 'accelerator with table' system and a communication path, and the rhythm node transmits a transmission reminder message by broadcast, and the accelerator system receives the reminder message for a predetermined waiting time. Later, Accelerator 'System Capra Primary It is in the 'combi-user' system, which is characterized by transmitting to the system.

 [0009] The present invention also provides a data record having a data item including a primary key and a substitute key, a substitute key entry including a substitute key and a primary key, a substitute key 'block including a substitute key entry, Alternate key, location, table.Alternate key with entries in contiguous area.Location 'Primary one' system with table and Fland's alternate key 'mouth Case Fland alternative key' location with 'table' entry in contiguous area The primary 'system sends the alternate key' location 'table change information, and each accelerator' system sends the alternate key 'location' table change information. Change the alternate key 'Fland.Location' table and receive the alternate One. Location 'change information table in response to received, after a predetermined waiting time, Akuserareta' and transmits a change completion notification for ply Marie system from the system, in a computer 'system.

[0010] The present invention also provides a data record having a data item including a primary key and a substitute key, a substitute key entry including a substitute key and a primary key, a substitute key 'block including a substitute key entry, Alternate key, location, table.Alternate key with entries in contiguous area.Location 'Primary one' system with table and Fland's alternate key 'mouth Case Fland alternative key' location with 'table' entry in contiguous area 'Axelator with table' system and communication channel, rhythm node Sending a transmission reminder message by broadcast, and receiving a transmission reminder message from the accelerator system, triggers transmission from the accelerator system to the primary system after a predetermined waiting time. , In computer systems.

 The present invention also provides a data record having a data item including a primary key, a primary block storing the data records in the order of the primary key, and a location table entry listing the address of each primary block. A primary system having a location table in a continuous area and a land location table in which the address of each primary block is described.Fland having an entry in the continuous area. It has a plurality of accelerator accelerator systems, a communication channel, and a rhythm node that sends a transmission reminder message.The rhythm node sends a transmission reminder message to each of the accelerator systems, and each of the accelerator `` systems '' Send Contract to receive reminder message As, after a predetermined waiting time, and transmitting the processing request for the primary 'system, simple drawings in a computer system described

FIG. 1 is an example of a database to which the present invention is applied.

 FIG. 2 is an example in which an overflow block management table is used in a database to which the present invention is applied.

 FIG. 3 is a diagram illustrating the principle of an accelerator.

 FIG. 4 is a diagram in the case of one accelerator system.

 [FIG. 5] This is a diagram in a case where there is one accelerator system and an overflow block management table is used.

 FIG. 6 is a time chart in the case of performing uplink communication using a transmission reminder message by broadcast communication.

 FIG. 7 is a typical configuration in a communication system.

[FIG. 8] A configuration of a computer using a communication system and a plurality of accelerator systems. 圆 9] This is a time chart for performing update processing and notification of change completion in response to transmission of change information and a transmission reminder message.

 FIG. 10 is a diagram of a format of change notification information.

 [FIG. 11] This is a diagram in which a single accelerator system is configured to belong to one group in a way of deciding a group.

 [FIG. 12] This is a diagram in which a single accelerator system is configured to belong to a plurality of groups according to how groups are determined.

 FIG. 13 is a diagram of a table for confirming that a change completion notification has been received for the primary system power and all applicable accelerators and system powers.

[Figure 14] Program logic in the accelerator system, which relates to processing requests to the primary system.

 [FIG. 15] This is related to receiving change information of the primary system and performing the change process in the program 'logic' in the accelerator's system.

[FIG. 16] This is related to the processing request of the program 'logic' in the primary 'system' and the accelerator 'system' system.

 [FIG. 17] Program logic in the primary system, relating to transmission of change information to the accelerator system.

 FIG. 18 is a diagram showing the configuration of a computer system, in which the final stage communication device has the functions of a tonic node, a dominant node, and a rhythm node.

FIG. 19 is a diagram of a BUS type computer system using an accelerator system.

 FIG. 20 is a time chart when a BUS type computer performs uplink communication using a transmission prompt message by broadcast communication in a system.

 FIG. 21 is a diagram of a computer system using a ring-type accelerator system.

Explanation of symbols

 0 · · 'Primary one' system (tonic 'node')

01, 02 · 'Secondary one' system 1,2,3,4,5 ... Accelerator one's system (dominant 'node)

 6, 7, 8, 9 ··· Accelerator 1 'system (dominant' node ')

 10 ... location 'table

 101 ... last pointer

 11 ... primary ^ ~ block

 12 "'Overflow' block

 15 · 'Overflow' block management table

 151 ··· 'Overflow ^ ~ · Block management table' pointer

 15A, 15B, 15C '.' Alternative key 'Overflow' block management table

 15A1, 15B1, 15C1 '' 'Alternate key' 'Overflow' 'Block management table' Pointer

 16 ... 'Fland' location 'table

161 · 'Fland's Last Pointer

 16A1, 16B1, 16C1 '' 'Fland alternate key' table last pointer

19 · 'Fland' Overflow ^ ~ · Block management tape 'Nore

 19A, 19B, 19C '.' Fland alternative key 'Overflow' block management table

19A1, 19B1, 19C1 '.' Fland alternative key'Overflow'block'Pointer

20, 21, 22, 23 ... Rhythm 'node,

 41, 42

 431, 432, 433

 441, 442, 443

 49 · "Switch

 50 Intermediate communication device

 55, 56, 57

 61, 64, 65, 66 ... Communication channel

 641, 642, 643… Communication channel

 71, 74, 75, 76

741, 742, 743… Communication channel 110, 120, 130, 140, 150 · · · Block · Element device

 160, 170, 180, 190. 'Block' element device

 111, 121, 131, 141, 151 · "Program

 161, 171, 181, 191 · Program

 112, 122, 132, 142, 152

 162, 172, 182, 192 • Block or alternate key block

 200, 210, 220

 201, 211, 221

 202, 212, 222 • π case table or alternate key block

 300Communication processing device

 S1… Send dunning

 S2, S3, S4, S5 ... Uplink communication

 S6, S7, S8, S9 ... Downlink communication

BEST MODE FOR CARRYING OUT THE INVENTION

[Data storage search method]

Here, the contents of the data storage and retrieval system invented by the inventor will be briefly described. The data storage search method of the present invention uses a location 'table and an alternative key table, and performs a binary search on them to search for a target record. Data records are stored in the primary 'block in primary key order. When adding a data record to the primary block, if the primary block is full, connect the overflow block to the primary block and store the data record. It is possible to further connect an overflow block to the overflow block. It has a location 'table' which has a location 'table' record (or location 'table' entry 1) in which the address of each primary block is described in a continuous area. The location 'table is reserved in a continuous area in advance. Here, the continuous areas are in a logical order, and may be separated in a physical area. In such a case, it can be treated as logically continuous using the address conversion table. This is the same in the following description. . Location 'Use the last pointer to indicate the end of the used area of the table.

 [0014] Records are stored in fixed-length storage areas called blocks. The block consists of a primary block and an overflow block. If a record cannot be added to the last primary block, add a primary block and store the record.

 [0015] Here, the term "connection" does not mean that the physical connection is established.

 'The block holds the address of the first overflow ^ ~ · The address of the first overflow block; the state where the block holds the address of the second overflow block' Such a representation is used because it can be treated as if the blocks are connected to each other (the same applies hereinafter). Because of this storage, the location 'table' entries are ordered by primary key. A primary key search finds a block by performing a nourish 'search between the location' table 'entry pointed to by the location' table 'first address and the final pointer, and finds the target in that block. Find a record. If the overflow block is linked to the block, the overflow block is also searched. Here, the same logic can be used to update, add, and delete force records mentioned in the search.

[0016] A substitute key 'record (or a substitute key' entry) that also has a substitute key value and a primary key value is stored in the substitute key * block in the order of the substitute key value. Alternate key * Alternate key to block.When adding an entry, if the Alternate key 'block is full, the Alternate key, overflow, block is linked to the alternate key block, and the alternate key Is stored. It is possible to connect an alternative key 'overflow block to the alternative key' overflow block. It has an alternate key 'location' table that has an alternate key 'location' table containing the address of each alternate key block (or an alternate key 'location' table. Entry) in a continuous area. Substitute key · The location 'table is reserved in a continuous area in advance. Alternate Key Location Uses the Alternate Key Last Pointer to indicate the end of the used area of the table. Addition of a substitute key entry, which is larger than the substitute key value of the existing substitute key entry. ヽ The substitute key entry with the substitute key value is the last substitute key. If the key cannot be stored in the substitute key block, create a new substitute key block and store the record in the substitute key block. The substitute key / location 'table and the substitute key / block are paired and called the substitute key' table.

 [0017] The substitute key is a non-unique key in the database, such as a name and a date of birth in the employee database. There may be no substitute key or a plurality of substitute keys for a certain type of database. To search for a record with a certain alternate key, the alternate key, location, first entry in the table and the alternate key last pointer point to the alternate key location 'table' entry 'binary' search. Then, find the target alternate key block and search in the alternative key block to find the alternate key entry with the target alternate key. If the substitute key / overflow / block is concatenated with the substitute key / block, the substitute key 'overflow' block is also searched. Next, by using the primary key of the alternate key entry, a binary search of the location tape is performed to find the target block, and the target record within the block. If an overflow block is linked to the block, the overflow 'block is also searched. Since the substitute key is non-unique, there may be multiple records with the same substitute key value. In this case, if the substitute key 'next substitute key in block' record has the same substitute key value, the same operation as above is repeated. Here, search is described, but record update, addition, and deletion can be realized by the same logic. If there are multiple alternative keys, the same number of alternative key tables as the types of alternative keys are created and used. This concludes the description of the “data storage search method”.

[0018] [record]

A record always has one unique primary key and zero or more non-unique keys (alternate keys, which can be unique as a result). In addition, there can be items (columns) that are not keys. The primary key is, for example, in the case of an employee database, a code that can identify an employee, such as an employee code.The substitute key is a name, a date of birth, and the like. No problem. For records where there is no item that should be a meaningful primary key, a sequential number in the storage order may be assigned and used as the primary key. Items (columns) are units of information, some of which are key and some of which are not. One or more occurrences in the record. The columns may be in fixed-length format, but may also be in variable-length format. In the case of the variable length format, no column information exists !, and a column can be recognized as a column.

 [0019] In the database 'system, expressions such as a subschema and a schema are generally used, but in this specification, description will be made irrespective. In the description of this specification, expressions such as "add column to database" and "access to database" refer to operations on a particular type of database file (eg, employee files). It is not for the whole database file. In addition, when a specific type of database 'file power' is also configured, the term database is used for each database 'file'.

 FIG. 1 shows a database structure invented by the “data storage and retrieval method”. Here, the overflow ^ ~ · block and the substitute key ^ ~ · overflow ^ ~ · block are omitted. Fig. 2 shows the "data storage and retrieval system", which manages the overflow and blocks and the alternate keys and blocks using the overflow block management table for the database of Fig. 1. In FIG.

 [0021] [Principle of the Accelerator 1 'system]

Accelerators are literally accelerators. In the conventional database system, the method of improving the processing capacity of the database was extremely limited. On the other hand, it is possible to effectively increase the processing capacity of the database without limitation. The principle of the accelerator system is as follows. Perform a binary search on the location table and alternate key location table to find the primary block containing the target record and the overflow block connected to it. In addition, find the medium record. When doing a binary 'search for a location' table ', you will often find the bisection point, often more than searching for records in a block. . Ma It is also very unlikely that multiple processes will request records in the same block at the same time. Therefore, if a plurality of copies of the location 'table and the alternate key' location table are held and a binary search can be performed on each of them in parallel, it is possible to execute many access requests for records. Figure 3 illustrates the principle of an “accelerator”. FIG. 4 illustrates a case of a method without using an overflow / block management table with respect to a basic configuration of an accelerator 1 'system. FIG. 5 illustrates the basic configuration of the accelerator system using a method using an overflow block management table.

[0022] The primary system includes a primary block and an overflow block, and an alternate key block. An alternate key and an overflow block are used to perform record update and deletion of records. It is a system. In contrast, the Accelerator 'System' has a 'Fland' location 'table and a Fland alternative key' location 'table, and performs a binary search on them. The target primary block and its associated overflow key; block, alternate key; block and its associated alternate key 'overflow

• After finding the block, perform the operation on the primary 'block and its associated overflow block, alternate key block and its alternate key and overflow key' block on the primary system.

FIG. 4 shows a case where there is one accelerator system. The Accelerator system has a location table (Fland's location table) and an alternate key case table (Fland's alternative key location table). , Alternate Key. Block, Alternate Key 'Ono One Flow ^ ~ · Blocks are not available. Accelerator ^ ~ The system's land location table is functionally equivalent to the primary 'system location' table. Similarly, the Fland / Alternate Keys location table for the Accelerator-System is functionally equivalent to the Primary-Alternate Keys Location table for the system. Accelerator ^ ~ · The system's land · Each record in the location 'table points to the same block as each location' table 'record in the primary' system. [0024] [Access in Accelerator 1 'system]

 In the accelerator system, when there is an access request, in the case of the primary key, a binary search is performed on the Fland's 'mouth case table' to find the target block, and the record search in that block is performed by the primary system. Ask. In the case of an alternate key, the Fland 'Alternate Key ^ ~ · performs a binary search of the location table ^ ~, finds the target block, and finds the target alternate key from the primary' Alternate Key 'block held by the system. Find the record and use the alternative key record to perform a binary search of the Fland 'location table' to find the desired record. The search method has been described here, but by applying this method, records can be updated, added, and deleted. Also, in the case of an alternate key, a binary 'search of the location table' is performed based on the 'alternate key' record, but the address of the block and the block number are stored in the 'alternate key' record. Not necessary in some cases. In this way, it is possible to increase the amount of processing by searching and updating records in parallel by a plurality of accelerator systems.

 [0025] In FIG. 4, the accelerator system has one location table and three alternative key location tables, which are the same as the number of primary systems. This is called "symmetric system" in "Accelerator". On the other hand, for example, it is assumed that an accelerator-type system has only two types of location tables, the location table and the alternate key. It is possible to create an accelerator system. This is called an asymmetric system. In the case of the accelerator 'system', the primary 'block and overflow' block and the alternate key 'block and alternate key' overflow.

 FIG. 5 shows an example in which an overflow block management table is added to the above-described accelerator system.

[0027] [Synchronization of information between the primary system and the accelerator system]

Next, a method for synchronizing the information of the primary system and the information of the accelerator system in the system using the “accelerator” will be described with reference to FIGS. 4 and 5. FIG. ply The Mari 'system has a location' table 10, an alternate key 'location table 10A, 10B, 10C. In addition, it has the final pointer (101, 10A1, 10B1, 10C1). If the database uses the overflow block management table (in the case of Fig. 5), the overflow block management table 15 and the alternate key overflow block management tables 15A, 15B, and 15C are used. ,have. Also, an overflow key ~ block management table 'pointer 151 is provided in the block management table, and an alternative key' overflow ''Block management table' Pointer 15A1 is provided. Similarly, substitute key 'Overflow' block management table · Pointers 15B1 and 15C1 are provided! / ヽ

 [0028] The accelerator 1 'system 3 has a Fland' location table 16, a Fland alternative key 1 location table (16A, 16B, 16C), and a final pointer (161, 16A1, 16B1, 16C1). If the database is in a database format using the overflow block management table, the Fland 'overflow' block management table 19, the Fland alternative key 'overflow' block management table 19Α, 19B, 19C is provided. For each of the land replacement keys' overflow 'block management tables of 19A, 19B and 19C, a land replacement key' overflow 'block management table' pointers 19A1, 19B1 and 19C1 are provided.

[0029] In the accelerator system, when a change occurs in the location "tape location" or the alternate key location table in the primary system, the change is notified to the accelerator system, and in the accelerator system, Change the location table or the replacement table. Change to one of location 'table 10, last pointer 101, alternate key' location table (10A, 10B, 10C), alternate key 'location table last pointer (10A1, 10B1, 10C1) in primary' system ' When an error occurs, the changed part is notified to the accelerator system. Based on the notification, the accelerator system will determine the location of the relevant location table 16, the alternative location key location table (16A, 16B, 16C) and the alternative location key location table. For the end pointer (16A1, 16B1, 16C1), change the relevant changed part.

 When the database uses the overflow block management table, in addition to the above, in addition to the above, in the primary system, the overflow block management table 15 and the overflow block management table 15 are used. Pointer 151, alternative key 'overflow' block. Block management table (15A, 15B, 15C), alternative key 'overflow' block management table. When the last pointer (15A1, 15B1, 15C1) is changed Informs the Accelerator-1 'system of the change, and in the Accelerator-1' system, the Fland's' Overflow 'block management table 19, the Fland's last pointer 161, the Fland's' Overflow' block management table 'Pointer 191, Fland substitution key “Overflow” block management table (19A, 19B, 19C), Fland substitution Against any one 'overflow one' block management table 'final pointer (19A1, 19B1, 19C1), and changes the corresponding portion.

 [0031] In this way, the primary system notifies the accelerator system to the changed unit, and immediately applies the change in the accelerator system, so that the location table 16, the location table 16 of the accelerator system can be applied. Overflow ^ ~ Block management table 19, Land final pointer 161, Land 'Overflow' block management table Final pointer 191, Land alternative key 'location table (16Α, 16B, 16C), Land alternative key 'Last pointer of location table (16Α1, 16B1, 16C1), Fland alternative key' Overflow 'block management table (19Α, 19B, 19C), Fland alternative key' Overflow '' Block management table ' Last pointer one (19A1, 19B1, 19C1) kept the same as the primary one's system It is. When the change of the relevant location is completed, the fax monitor system sends a change completion notification to the primary system. The primary 'system' waits for the exclusion of that part until a notification of the completion of the change of all the accelerator's system capabilities arrives.

[0032] In the above, the application of the basic case to the accelerator system has been described. However, in the case of direct column addition, direct column deletion, and direct column change in the "database system", the following is further required. Condition is added. In addition to the above conditions, the primary Location The column operation pointer for table, the column operation completion pointer, and the new location table, the new column operation pointer are increased. To deal with this, the accelerator must be able to use the column working pointer for the table.

(Fand column operation pointer), Fland column operation completion pointer, Fland new location 'table, and Fland new column operation pointer. Database overflow Overflow. If the format uses the block management table, a new location is added to the new location table.The block management table and a new overflow block management table 'pointer are added. . If any of the above elements are changed in the primary system, the change is notified to the accelerator system, and the accelerator system changes the location. In addition, when performing reorganization, the new location 'table' and the new alternative key 'location' table, and the final pointer of each, are further reduced.

 [0033] In the accelerator system, the access to the block is changed to the primary system, and the others are described in the column addition / deletion change described above and in the accelerator system. It can be realized by combining the methods described above.

 [0034] In the above description, the description has been made assuming a symmetric system of the accelerator system. However, in the case of an asymmetric system, only the accelerator system that retains the changed part in the primary system receives the change and is updated. Will do. Above

"The explanation was made using a citation from the" database 'system "by the present inventor.

[Communication system]

Next, the “communication system” will be briefly described. The current mainstream communication is collision detection. This means that nodes connected to the communication channel (generally including communication devices, but in this case targeting computers such as personal computers and servers, excluding communication devices) transmit information at arbitrary timing, If two or more nodes transmit at almost the same time, a collision is detected, the information is canceled, and transmission from the node is performed again. The retransmission is a line in which a collision occurs at an accelerated rate when the amount of communication on the force communication path that shifts the timing of each other increases so that a collision does not occur. [0036] On the other hand, in the "communication system", nodes are classified into tonic 'nodes and dominant' nodes, and hosts 'computers and servers are basically tonic' nodes. In addition, the communication direction is separated from the tonic 'node to the dominant' node (downlink communication) and from the dominant 'node to the tonic' node (uplink communication). Furthermore, in order to avoid collision of uplink communication, the rhythm node sends a transmission reminder message to each dominant 'node' by broadcasting, and each dominant 'node waits for a predetermined time after receiving the transmission reminder. After the elapse, uplink communication is performed. This predetermined waiting time is quantitatively determined according to the communication sequence number, and according to the communication sequence number, the dominant node has a communication path number, a communication path speed, and a communication message length. The node with the communication sequence number n performs transmission at the same time as the transmission of the node with the communication sequence number n-1 ends (strictly speaking, the timing differs depending on the length of the communication path). The formula for calculating the predetermined waiting time is described in detail in “Communication System”, and is as follows.

 [0037] It is assumed that the transmission message length is L (byte) and the transmission speed of the communication path is S (bit Z seconds). On the other hand, let the waiting time be T (nanosecond). The length of the communication path (62, 63, 64, 72, 73, 74) is calculated as follows: Length of communication path (62) = Length of communication path (72), Length of communication path (63) = Length of communication path It is assumed that the length (73) and the length of the communication path (64) = the length of the communication path (74). It is also assumed that the length of the communication channel (62) = the length of the communication channel (63) = the length of the communication channel (64). First, consider the case where the lengths of the communication paths (621, 622, 623) are the same. Let the length of each of these be D (meters). It is also assumed that the length of the communication path (62) = the length of the communication path (63) = the length of the communication path (64). Accelerator 1 System Assume that uplink communication is performed in the order of 1, 2, and 3. Then, at the moment (T1) at which the accelerator 1 system 1 completes the transmission, the end of the transmission information passes through the portion where the communication path 621 is connected to the accelerator system 1. If a transmission message is transmitted from the accelerator 1 'system 2 at the time of T1, collision with the transmission message of the accelerator 1' system 1 will not occur. The waiting time Tn of the accelerator system 'n' is expressed by the following equation (1).

 [0038] [number 1]

Τ "-(«-1) (81/5) · '· Equation 1 [0039] Other cases where the length of the communication path is different are described in “Communication System”. In downlink communication, a communication right is assigned from a rhythm node to a tonic node, and downlink communication is performed from a tonic node having the communication right. Further, if the communication path is separated into uplink and downlink, more efficient communication can be performed.

 FIG. 6 shows a time chart of the transmission reminder and the uplink communication when the lengths of the communication paths are equal. In this figure, it is assumed that the tonic 'rhythm' node is included in the node. Dominant 'The number of nodes is four. The distance between the tonic 'node and each dominant' node is equal. Send a reminder from rhythm node 0 (tonic node) (S 1

). Since the lengths of the communication channels are equal, the dominant node (1, 2, 3, 4) receives a transmission reminder at the same time. Dominant 'node 1 immediately performs uplink communication. Dominant Nodes 2, 3, and 4 perform uplink communication after a predetermined waiting time. When one cycle of uplink communication is completed, downlink communication is performed from the tonic node. In this figure, the figure up to the drop communication for the dominant node 3 is shown.

 [Method of Realizing High-Performance Computer]

 By using the accelerator system, it is possible to realize high-performance computers and systems. In conventional computer systems, high performance was achieved by installing a large amount of CPU and main memory. However, as explained in the background art, even if they are mounted in large quantities, the load for controlling them increases, making it difficult to increase the processing capacity. By using an accelerator 'computer with system' system, it is possible to dramatically increase the processing performance of the computer. The primary system and the accelerator system may be housed in a single housing, or they may be independent and connected. However, in order to achieve high performance while simply putting them together, simply putting them in one housing or connecting them is not enough.

[0043] In the "Accelerator", as a method of notifying the change information of the primary system, a method of individually transmitting information to the primary system and each of the accelerator systems is described. In the description, the information is transmitted and received through the change information transmission mechanism, the change information reflection mechanism, and the communication mechanism. It can also be realized by creating air. When the information of a specific address is changed, the information of another address is automatically rewritten in the same manner, so that the location table of the primary system and the location table of the alternate key are changed. This is a mechanism (synchronous update mechanism) that updates the land 'location' table and the land substitution key 'location' table of the accelerator-'system when it is done. However, the specific means of implementation were clearly clear. In addition, a large amount of communication from each accelerator system to the primary system occurs, but the means to realize this communication has been clearly identified.

 FIG. 7 shows a typical configuration of the “communication system”. Node 0 (0) is a tonic 'node and incorporates a rhythm' node 20. Nodes 1, 2, 3, · · · 9 are dominant 'nodes. Both are connected by a communication path via an intermediate communication device 50 and final stage communication devices 55, 56, 57. The communication path is separated for upstream communication and downstream communication. Figure 8 shows the application of this mechanism to the “Accelerator”. The tonic 'node becomes the primary system 1 and the dominant' node becomes the accelerator 1 'system (1, 2, 3, · · · 9).

 [0045] [Change information transmission]

With reference to FIG. 8, a description will be given of how to realize an “accelerator” in such a configuration. In this figure, the communication paths (61, 64, 65, 66, 641, etc., 71, 74, 75, 76, 741, etc.) and the communication device may be serial transmission or parallel transmission. In FIG. 8, the communication paths 61 and 71, which use two communication paths for upstream communication and downstream communication, such as 61 and 71, are combined into one communication path in FIG. There is also a method of realizing uplink and downlink communication by time division. The time-chart in FIG. 9 shows such time-division communication. When two communication paths are used for uplink and downlink, it is possible to perform uplink and downlink communication simultaneously in parallel. In the following description, communication from the primary system to the accelerator system will be referred to as downstream communication, and communication from the accelerator system to the primary system will be referred to as upstream communication. First, the types of uplink communication and downlink communication will be described. If there is any change in the primary 'system, and the change is related to the accelerator' system, the primary 'system sends the information to the accelerator' system as downlink communication (change Transmission of information). In response to this change information, each accelerator's system applies the change and, upon completion of the change, sends a change completion notification to the primary system as upstream communication (transmission of the change completion notification). In addition, in the accelerator's system, the application searches for locations, tables and alternate key locations, and tables based on requests from computers, and searches blocks and alternate keys based on the search results. Request the system (send processing request). The primary system transmits a notification of the content of the block or the substitute key block or a change result as downlink communication based on a request from the accelerator system (transmission of the processing result). Here, a method for efficiently transmitting the change information will be described. When there is only one accelerator system or several accelerator systems, it is unlikely that any particular problem will occur if individual change information is sent to each accelerator system. When the number increases, the time required for transmitting the change information increases, and there is a possibility that other downlink communication cannot be performed. In order to avoid such a situation, the transmission of the change information is broadcast to the Accelerator's system. It is very efficient to use broadcast since changes in the primary 'system need to be sent to all the accelerator' systems in the case of symmetric systems. FIG. 9 is a time chart when the broadcast is used for transmitting the change information. Here, too, the rhythm node is built into the primary system. After the change information is broadcast from the primary system, the rhythm node broadcasts a transmission instigation message to the accelerator systems (1, 2, 3, 4) (Sl). The change information can also be used as a reminding message. In this case, the relationship between the rhythm node and the primary system is that, when one cycle of the rhythm node upstream communication is completed, the primary system is given the right to transmit change information to the primary system. However, it must be separated from normal downlink communication. Each accelerator / system performs update processing immediately after receiving the change information. Thereafter, the accelerator systems 2, 3, and 4 send an update completion notification to the primary system after a predetermined waiting time (S2, S3, S4, S5). FIG. 9 shows the timing for the case where each accelerator 'system has the same performance (the time required for the update processing is equal). [0047] [Contents of change information]

 The details of the change information will be described in detail. FIG. 10 is an example of change information. The contents of the change information include information such as destination address 80, source address 81, serial number 82, change target 83, change target address 84, change length 85, and changed content 86. Destination address 80 is the address of the node that should receive the information. In this case, since the broadcast is the change information, the group ID of the accelerator system is the destination address. The source address 81 is the address of the primary system. The serial number 82 is used to distinguish between different information when multiple change notifications are sent and when a change completion notification is received by the primary system, and the serial number used in the change information is used. This is used to identify which change information the change completion notification is sent to by sending the accelerator 'system power to the primary' system in the serial number of the change completion notification, which is commonly used in information communication. Is the way you are. The 83 to be changed is the location 'table' or multiple alternate key 'location' tables, such as the final pointer, overflow, block management table, multiple alternative keys, overflow, block 'management table, etc. Accelerator · Indicates which component of the system is changed. This will ensure that the primary 'system and accelerator' systems maintain the same mapping table and can be undoubtedly changed. The address 84 to be changed may be specified physically or logically.

The term “physical” is used when the physical configuration of the components of the accelerator system in the primary system and the physical configuration of the components in the accelerator system are exactly the same. The physical configuration is exactly the same, for example, if the location 'table's structural power is exactly the same in the primary system and the accelerator system, and the location' table in the primary 'system' The address is exactly the same as the address on the storage device of the location 'table in the accelerator' system. As a specific example, if the location 'table' entry 5 of the primary 'system' is, for example, 50 bytes from address 1050 of the storage device, all accelerators ~ ~ Five It is 50 bytes from address 1050 on the 1S storage device. In this case, the primary system sends a change notification indicating that any number of pieces of information on the storage device has been changed. Upon receiving the change notification, the accelerator system changes the address of the storage device. In this case, it will operate normally even if there is no information on the change target 83.

 Logical means location 'table, alternate key' location 'table, overflow, block management table, alternate key overflow, block management table It is. The other components are not structured to have multiple entries, so if you specify, for example, the last pointer, the target is determined.

 To further expand this logical identity, the structure of the location 'table and land.location.table, alternate key' location 'table and the land and alternate key' location 'table may differ. good. The difference in structure is as follows. The information needed for the location 'table' is the number of the entry and the address of the primary 'block' pointed to by that entry. In addition to this, it is useful to extend the entry information in the primary 'system location' table to hold the statistics of each block. On the other hand, the accelerator 'system land' location table has only the entry number and the address of the primary block without statistics. In this way, the amount of information stored in the xereleter system can be saved and changes can be reduced.

Further, the accelerator can be extended as follows. The Accelerator's system is to maintain a location 'table and an alternate key' location table. However, if the load on some specific substitute keys is high, the substitute key block is retained in the accelerator. The alternate key block is, of course, the same as the one held by the primary 'system, and is synchronized with the primary' system by the synchronous tight coupling method shown in “Backup / Recovery. System”. Take. By doing so, it is possible to reduce the load on the primary system when the load on some of the alternate key 'location' tables is high. [0050] The shorter the change information is, the shorter the time required for transmission and updating is. For example, if the update is for a location 'table' entry, the block number and address are not changed, so the case where the minimum or maximum value of the primary key value is changed is the target. Even when either the minimum value or the maximum value is changed, if the entire primary key value is changed, the lower digit may be changed. In such a case, it is possible to shorten the change information by transmitting only the difference using the change target address and the change length 85. In this case, if the address to be changed is logical, it must be in the form of an entry number and a displacement within the entry. Even when the difference is not used, the length to be changed is indispensable when the change target address 84 is physically specified. The change target address 84 is specified logically, and even if it is not transmission of only the difference, transmitting the length to be changed is useful for ensuring the change.

 [0051] The post-change content is the content itself to be changed. For example, if the maximum value of the primary key value of the location 'table' entry is changed from “56789” to “56790”, the content will be “5 6790” after the change. Also, as described above, only the last two digits “90” can be transmitted as a difference.

 [0052] [Broadcast]

 It is described that the communication information is performed by using the broadcast communication. This will be described in more detail. In the case of broadcast communication, there are simultaneous broadcast and group broadcast. There are symmetric and asymmetric systems in the accelerator system. A symmetric system is a form in which all of the accelerator and system have the location 'table' and the alternate key 'location' table held by the primary 'system'. In contrast, an asymmetric system refers to a form in which a particular accelerator's system has only a part of the location 'table and the alternate key's location' table held by the primary 'system. An asymmetric system is effective in saving storage space when the access frequency differs for each key type.

If all of the accelerator systems are symmetric systems, simultaneous broadcasting may be used. On the other hand, in the case of an asymmetric system, group broadcasting is used. this Although there are various methods for defining the group, the following method shown in FIG. 11 is typical. Fig. 11 shows a method of patterning whether an accelerator system has a location table or an alternate key location table, and assigning one group to one accelerator system. . Determining the group in this way is a method generally used in conventional communication. This method has the following features. Each accelerator system has a group ID for each system, making it easier to respond

. On the other hand, when the types of substitute keys increase and the combination becomes complicated, the types of groups may increase exponentially. In the primary 'system', it is necessary to analyze which groups will be affected if changes are made to the 'location' table or the alternate key 'location. Also, in the case of FIG. 11, when the location 'table is changed, the group group 2, the group 3, the group

Broadcasting must be performed for each of the four, and the amount of information for downlink communication increases. On the other hand, FIG. 12 shows an example in which the accelerator system is allowed to belong to a plurality of groups. Figure 97 in Figure 12 shows whether each accelerator system has a location table or an alternate key.location table. On the other hand, show how to determine the group name depending on whether you have the location 'table or the alternate key ^ ~ · location' table! /! This is shown at 98 in Figure 12. At 99 in FIG. 12, a list of the group IDs determined in this way is shown. The accelerator system 1 belongs to any of groups 1, 2, 3, and 4. Similarly, the accelerator systems 2, 3, and 4 also belong to groups 1, 2, 3, and 4. Accelerator system 5 does not belong to force group 4, which belongs to groups 1, 2, and 3. Accelerator system 6 does not belong to force groups 3 and 4 that belong to groups 1 and 2. Accelerator system 7 belongs to Group 1 only. If the method shown in Figure 12 is adopted, changes to the location 'table or alternate key' location 'table should be broadcast to the group corresponding to the location' table or alternate key 'location' table. Only one downlink information communication is required. In addition, the destination of the downstream information communication of the primary system can be easily determined. As the number of alternatives increases, the types of groups increase only proportionately. Meanwhile, Axellare It is necessary for a single-user system to keep track of which group it belongs to, so that it can support broadcasting for multiple group IDs.

 In FIG. 9, after receiving the change completion notification, downlink communication is performed from the primary system (S6, S7, S8, S9), but actually, communication paths for uplink communication and downlink communication are provided. Are provided separately, so that the downlink communication may be performed when the transmission of the change information and the transmission of the transmission reminder have been completed. In addition, downlink communication need not be performed equally for each accelerator system, but may be performed according to the traffic.

 [Accelerator · When the system performance is uneven]

 If the performance of each accelerator's system is not the same, it receives the change information and the transmission reminder, and measures the waiting time. The update processing time can be set as a waiting time by transmitting the time required by the primary system in the change information and adding the update processing time to a predetermined waiting time. In this case, the performance of Accelerator 'System 1' (the first Accelerator 'system with the communication sequence number) should be the same as that of the Primary' System '. It is necessary to have the performance enough to finish the update processing within the time. If the update process does not end within the predetermined waiting time for some reason, the change completion notification is not transmitted. If a change completion notification is sent at an arbitrary timing, a communication collision will occur.

 [0057] [Change completion notification]

The broadcast is performed to the group in this way, the necessary changes are made in the accelerator's system, and a change completion notification is sent to the accelerator's system primary system. In view of the primary system power, it is necessary to receive and confirm numerous change completion notifications without omission. It is effective to use the change completion confirmation table shown in Fig. 13 as a method for checking without omission. The node in the change completion confirmation table is each of the accelerator systems. The status is incomplete or completed. In Figure 13, incomplete is indicated by a blank. The count in the top row means that the number of nodes in the group is entered first, and each time the primary system receives a change completion notification from each node, the status of the node is set to completion, and 1 is counted from the count. Reduce. If already When a change completion notification is received again from a node whose status is completed, it is preferable to inquire the node. In this case, do not subtract 1 from the count. When the count becomes 0, it can be confirmed that all the change completion notifications for the group have been received. It is convenient to create a prototype of this change completion confirmation table for each group, and to make a copy of the primary system power when sending a change notification.

 [0058] The primary system transmits a change notification to the accelerator system, and the primary system receives a change completion notification from all the target accelerator systems (the accelerator systems belonging to the change notification target group). Until the Primary 'system, the changed part is excluded.

 The transmission of the change information and the transmission of the change completion notification have been described above. Next, the accelerator. The power of the system, which is the transmission method of the processing request. This is to make each accelerator's system power also be transmitted using the transmission reminder shown in “Communication system”. In this case, there is a high possibility that the length of the information of the change completion notification and the processing request is different. The reason is that the change completion notification only needs to include the serial number of the change information and information indicating that the change has been completed, so that a length of about 20 bytes is sufficient even if the destination address and the source address are included. On the other hand, in the case of a search for a processing request, a block address, a key type, a key value, and the like are required. Updates and writes require the contents of the record and may be longer.

As described above, if information of different lengths is to be handled by a packet of a single length, a useless space in the packet increases. In order to avoid such a situation, as described in the “communication system”, when transmitting the transmission reminder, it is preferable to specify the type of transmission information of the accelerator system. The transmission reminder in the case of the change completion notification has a packet length of, for example, 64 bytes, and a normal processing request has a packet length of, for example, 256 bytes. Naturally, if the packet lengths are different, the required predetermined waiting time varies for the same accelerator system, so when using multiple packet lengths, the predetermined It is necessary to calculate the waiting time. As described above, by applying the “communication system” to the “accelerator” and transmitting the change information by broadcasting, efficient downlink communication can be performed.

 [Transmission of processing request]

 The processing request for uplink communication has already been described briefly, but will be described again. Each accelerator system accesses the database based on an access request to the database from the application program in the accelerator system or the application processing computer. For example, if the access is by primary key, the Accelerator-1 'system searches its own land' location 'table to find the target block. Blocks and overflow blocks do not exist in the accelerator 'system, but only in the primary' system. In the case of a search request, specify the search DB type, search target block type, block number or block 'address of the target block, and primary key value. It is sent to the primary system as a processing request record. This processing request occurs at an arbitrary timing when each accelerator's system is operating, and there is a possibility of collision if these are sent without permission. In addition, since collisions are likely to occur when communication is crowded, performance deteriorates when many processes must be performed.

 [0063] In order to avoid such a collision, in the "communication system", a transmission reminder is transmitted from the rhythm node in uplink communication, and when the communication reminder is received by the dominant node (accelerator system), a predetermined waiting time is received. After a time, the uplink communication is to be performed. By doing so, collision does not occur in uplink communication. The method of transmitting the transmission reminder and performing the upstream communication with a predetermined waiting time after the reception of the transmission reminder is also applied to the “axelator”. The timing chart in this case is the same as in FIG.

 [0064] [Accelerator ^ ~ · System logic]

Accelerator 1. The power required for a program in a system The logic of the program will be explained again. FIGS. 14 and 15 are flowcharts of an accelerator system program (accelerator system program). In Fig. 14, first, access is accepted from the application program. The application program may be run in the accelerator system or in an external system. It may be configured to run and accept database access. The accelerator system 'system' program then identifies the type of access. Which database (ski) is targeted, how many key types are, what access is search, addition, update, or deletion, and how many key values are. Next, based on the key type and the key value, a binary 'search of the location' table or the alternative key 'location' table is performed. There, find the number or address of the desired block or alternate key block. Therefore, the accelerator 'system' program creates and stores outgoing messages. Thereafter, upon receiving the transmission reminder message, after a predetermined waiting time, a processing request is made to the primary's system by uplink communication. The processing is performed on the primary system, and the result is returned to the accelerator system. The result is returned to the application program. FIG. 15 is a flowchart in the case of receiving change information of the primary system. Change information is also received from the primary system. Make necessary changes depending on the target of the change. After that, a change completion notification is sent after a predetermined waiting time.

[Logic in the primary system]

Next, the processing contents of the program of the primary system (primary system program) will be described. FIG. 16 and FIG. 17 are flow charts thereof. FIG. 16 relates to a processing request from the accelerator 1 'system. First, the processing of the accelerator system is accepted. Next, the type of access is identified. For which database (key), what kind of key is, what kind of access is search, add, update, delete, how many key values, block or alternate key block number It is. Next, find the entry in the location 'table or the alternate key location table by the number of the block or alternate key' block. If the processing request is for an update, the entry is excluded. The exclusion method will be described later. When the processing request is based on the primary key, the primary 'block' and the overflow connected to the primary 'block are detected.If the target record is detected from the block and the processing request is a search, Sets that record. If the processing request is an update, update the record. In case of addition, it is assumed that no record with that primary key value exists To determine the record insertion location and insert the record. If the processing request is to be deleted, delete the record. If the processing request is search, update, or delete, and there is no record with the primary key value, an error occurs. If a processing request is added and a record with that primary key value already exists, an error will occur. If it is normal, set necessary information such as a record in addition to the return code. Then pass the information to the Accelerator system. This is transmitted as downlink communication. Figure 17 is a flow chart for sending change information to the accelerator system.

[0066] Other methods]

 Next, the exclusion method will be described. The exclusion is most simply done by flagging the desired entry in the location 'table or alternate key location.table. In this way, it is possible to reliably detect which block or substitute key 'block is exclusive. Ending a transaction, deadlocking a transaction; canceling a program or canceling an abnormal termination of a program can release the exclusion related to a certain transaction. Location 'Table · Write out the entry on another table and refer to it.

[0067] [Special Accelerator 'System]

Above, a description will be given of the force variation which has been basically explained about the accelerator system. If the accelerator's system is powerful, the primary's system accepts processing requests from the accelerator's system and performs processing, and if changes occur, sends change information to the accelerator's system. Only doing so will help improve overall efficiency. However, if the number of Accelerator 'systems is not very large, the primary' System 'can also be loaded with the functions performed by the Accelerator' system to improve overall processing efficiency. In such a case, for example, the accelerator 1 'system 1 in Fig. 8 is set as a dummy, and the accelerator 1' system 1 does not perform any processing on the processing request that also accepts the application 'program power. This is a method that sends it to the primary system as it is. Accelerator 1 'system 2, 3, 4, 9 Performs the processing as shown in FIG. In this case, the primary system program has the logic shown in FIG. 14, and if the processing is requested from the accelerator system 1, the logic shown in FIG. 14 and the logic shown in FIG. 16 are used. Execute.

[0068] [Accelerator I. When there are many systems]

 The example in the case of adopting the basic configuration has been described above. If the number of Accelerators 'systems is relatively small, no particular problem is expected to occur even if a transmission reminder message is sent to all Accelerators' systems. On a communication path with a speed of 10 Gbps, if 8 bits are 1 byte, communication of 10 bits per Ins (nanosecond) is possible, and 0.8 ns is required for 1 byte communication . Assuming that the length of the upstream communication message of the accelerator / system is constant, for example, 256 bytes, in this case, the time for transmitting 256 bytes is 204.8 ns. If the number of Accelerator's systems is set to 100, a transmission reminder message is sent approximately every 20 seconds, and processing requests from the Accelerator's systems can be sent at that interval. This means that the number of times that one axelator's system can perform uplink communication per second is 50,000 times.

[0069] However, if the number of accelerator / systems becomes 1000, the number of uplink communications per second becomes 5000. It is also possible that not all accelerator systems do not have information on upstream communications. An effective method in such a case is a multi-level rhythm 'node shown in "communication system". Figure 18 shows an example. In FIG. 18, the basic configuration is the same as that of FIG. 8, but the configuration of the final stage communication device is different. The final stage communication device (55, 56, 57) in FIG. 18 has the function of a dominant 'node and a tonic' node, and further has a rhythm 'node (21, 22, 23). The function of this final stage communication device is as follows. The last-stage communication device 55 will be described as an example. The final stage communication device 55 functions as a tonic 'node for the accelerators' systems 1, 2, and 3. It also has a rhythm 'node. Send a transmission reminder message to each of the rhythm 'nodecards'accelerators' systems 1, 2, and 3. After receiving the transmission reminder message, the Accelerator's system performs uplink communication after a predetermined waiting time. In this case, if there is no uplink information, for example, NACK is transmitted. The final stage communication device 55 is thus connected to each of the accelerators' systems 1, 2, 3 These upstream communications are collected and stored.

 [0070] In response to such an operation, the rhythm node 20 controls the final stage communication devices 55, 56, and 57 to dominant.

 'Send a reminder message as a node. This transmission reminder message is sent to a group that targets the multi-stage communication devices 55, 56, and 57. It is preferable that the final stage communication devices 55, 56, and 57 do not transmit the transmission reminder message to the accelerator system. Even if it is transmitted, it is not addressed to the Accelerator's system, so it is ignored. This will reduce unnecessary traffic. The last stage communication device, as a dominant 'node', performs a rising communication after a predetermined waiting time after receiving the transmission instigation message from the rhythm 'node 20. Uplink communication may be performed one message at a time from each terminal communication device. For example, ten messages may be transmitted at one time. Naturally, the predetermined waiting time is longer because the total length of the message is longer.

 [0071] By using such a method, it is possible to reduce empty uplink communication from the accelerator system. Furthermore, as shown in the "communication system", the number of uplink communications allocated to each dominant node is included in the transmission prompt message so that the uplink communication of a variable number of messages is performed from the last communication device. Then, communication efficiency can be further improved.

 [Accelerator I. System failure]

 By the way, if a large number of accelerator systems are connected in this way, some accelerator systems may not be able to operate for some reason. Some possible reasons include failure of the Accelerator's system, failure of the branch line communication path, maintenance of the Accelerator's system, and so on. The following describes how to deal with cases where a certain accelerator system does not operate.

[0073] A method of detecting when an axelator's system stops operating is to use a method of periodically exchanging signals with the primary system and deeming a failure if the signal is lost. I have. In the present invention, the presence / absence of uplink communication in response to a transmission reminder can be used for a signal, and it is reasonable to use this. In this way, if the primary system detects that the accelerator system is inactive, the primary system disconnects the accelerator system. Primary one's cis The system performs backout processing (returns the data contents to the contents at the start of the transaction) when the data in the database has been changed in the transaction that was processed by the inactive accelerator system. . If the contents of the backout affect the accelerator's system or the secondary 'system described later, the change information is transmitted to the accelerator's system or the secondary' system. In order to perform backout in this way, it is necessary to obtain the pre-change information (B log) of the relevant part when changing data in the primary system. This B-log is no longer needed once the change has been completed on the Primary 'system and the change has been completed on the Accelerator's system and the Secondary system. In some cases, you can use it effectively by saving it for knockup. This is described in the Secondary 'System section. The inoperable accelerator system is excluded from the change completion confirmation table.

Next, the primary system stores all of the subsequent change information including the change completion notification that has not been received from the accelerator system. In addition, the application processing computer does not transmit a processing request to the inactive accelerator system, and distributes the transaction at the time of inactivity to another accelerator system. This completes the disconnection of the disabled accelerator system. From then on, the cause of the inoperability will be manually searched for and restored to work.

[0075] The accelerator that was not in operation 'system power When the system can be operated, the following recovery measures are taken. The primary 'system' sends the saved change information to the inactive accelerator 'system. This start time is defined as tl. In the primary system, the ability to transmit change information addressed to other axelator systems after the tl point is disabled. The axcelerator system, which is inactive, only receives and does not apply the changes. The failed accelerator system executes the changes sequentially based on the change information stored in the primary system and sends a change completion notification to the primary system. The change completion notification may be sent each time one change information change is completed, but the number of change completion notifications increases. You may go together. The time when the saved change information is lost is defined as t2. After t2, based on the change information saved by the inactive accelerator system, the change is executed sequentially and a change completion notification is sent to the primary system. The change completion notification may be made each time one change information change is completed, but may be made collectively because the change completion notification increases. The time when the change of the change information saved by the inactive accelerator system is completed is defined as t3. At time t3, the synchronization of the non-operating accelerator system and the other accelerator system has been completed. Therefore, the inactive accelerator system after t3 can be treated as an active accelerator system. In the primary system, the inactive accelerator system is removed from the change completion confirmation table, and the system is restored.

 [0076] Next, a description of the case where the primary system becomes inoperative will be described in detail in the case of using the secondary system.

 [0077] [BUS type configuration]

 Although the case of the cascade connection has been described in FIG. 8, the configuration is possible even with the BUS type. FIG. 19 shows an example in which a BUS type configuration is used. The primary system and the accelerator system are connected by a BUS-type communication channel (41, 42). The communication path 41 is for downlink communication, and the communication path 42 is for uplink communication. In addition, the force communication path 41, in which both the communication paths 41 and 42 have a single branch, may have two branches. The reason why the communication path 42 is one-sided is that the upstream communication power from the accelerator system does not flow to the downstream side. When the upstream communication power from the accelerator's system flows to the downstream side, for example, the upstream communication power from the accelerator 1 'system 3 also flows to the accelerator 4' system 4 side, so it flows from the accelerator's systems 4, 5, 6, 7 This means that there is a possibility of collision on the communication path 42. Such a collision can be avoided by setting the communication path 42 to one-branch and transmitting the upstream communication of each accelerator's system at appropriate intervals. A reasonable interval is nothing more than taking a predetermined waiting time.

FIG. 20 is a diagram relating to the predetermined waiting time. Here, to keep the drawing simple Here, there are four cases where the accelerator system is 1 to 4. Rhythm 'node (= primary' system 0) sends a transmission reminder message by broadcast. Then, unlike the case of the cascade type, the outgoing reminder message is delivered to each accelerator's system with a delay according to the distance since each accelerator's system is gradually increasing in distance from the primary's system. (S l). After receiving the transmission reminder message, the “Accelerator's system” performs uplink communication after a predetermined waiting time (S2, S3, S4, S5).

[0079] In this case, the predetermined waiting time is equal to the distance between each accelerator and the system described in detail in the "communication system", and the equation in the case is as follows. The distance between each accelerator and the system is D (meters), the message length is L (bytes), and the transmission speed of the communication path is S (bit Z seconds). On the other hand, the waiting time is T seconds. The number of the primary 'system power closest to the accelerator' system is 1, and the nth dominant 'node is the nth node. Assuming that the waiting time of the nth dominant 'node is Tn (nanoseconds), Equation 2 expressing the waiting time Τη is as follows.

[0080] [Equation 2]

T _n = {n-1) {(8 / S)-(D / (3 x \ 0 ^s ))}

[0081] The case where the distance between each accelerator 'system is different is described in "Communication system".

[0082] [Ring type]

FIG. 21 is a diagram showing a case where the accelerator 'system is configured in a ring shape. Here, in the communication path 41, for example, the branch line communication paths 331 and 312 are not disconnected but are connected. This enables broadcast communication and eliminates the problem that the entire network stops even if one of the nodes fails. Here, it is composed of the primary system 0 and the accelerators' systems 1, 2, 3, ... 9. Further, since the communication path is a ring type, the communication path 41 is a single line. For this reason, the communication path 41 does not separate upstream communication and downstream communication. For convenience, the primary 'system power, the accelerator' and the communication in the direction of the system are downlinked. The communication in the upstream direction is called uplink communication, and whether the communication flowing through the communication path 41 is uplink communication or downlink communication depends on its destination. Each accelerator's system is connected to a communication channel 41 via branch communication channels 311, 312, and the like. The branch communication paths are one-branch, and the branch communication paths 311 and 321 are used for receiving downlink communication, and the branch communication paths 312 and 322 are used for transmitting uplink communication.

[0083] [Uplink communication]

 Here, it is assumed that a rhythm node 20 is provided in the primary 'system 0'. First, for uplink communication, a rhythm 'node 20 transmits a transmission reminder message by broadcast. After receiving the transmission reminder message, the accelerator's system transmits an upstream message with a predetermined waiting time depending on the distance between each accelerator's system. In this case, the predetermined waiting time uses an expression in the BUS type communication network. In the case of the ring type, the message may be a message addressed to the primary system without being limited to uplink communication. The primary 'system' receives the message addressed to itself, and the other messages do not exist again in the power-accelerator 'system, which transmits the message again to the communication channel as downlink communication. The Primary 'system discards all received messages except those addressed to itself.

[Downlink communication]

 Next, the rhythm node assigns downlink communication to the primary system 0. Therefore, the primary system 0 performs downlink communication. All communication messages returned to the primary system after performing downstream communication are discarded.

Claims

The scope of the claims

 [1] Contiguous area with a data record that has a data item that includes the primary key, a primary block that stores the data records in primary key order, and a case table entry that contains the address of each primary 'block' A primary 'system with a location' table and

 An accelerator system that has a land location table that has a land location table table in the contiguous area in which the address of each primary block is described;

 Communication channel,

 When the primary system sends the change information by broadcast and the accelerator system receives the change information, it sends the change completion notification to the primary system after a predetermined waiting time. Characterized by the computer's system.

 [2] Contiguous area has a data record with data items including the primary key, a primary block storing the data records in the order of the primary key, and a case table entry listing the address of each primary 'block' A primary 'system with a location' table and

 An accelerator system that has a land location table that has a land location table table in the contiguous area in which the address of each primary block is described;

 Communication channel,

 Rhythm ・ The node sends a transmission reminder message by broadcast, and after receiving a transmission reminder message by the accelerator system, it transmits to the accelerator system 'system-primary' system after a predetermined waiting time. A combination computer system.

[3] A data record with a data item that contains a primary key and a substitute key, a substitute key 'entry consisting of the substitute key and the primary key, a substitute key' block containing the substitute key 'entry, and a substitute key location table .Alternate key location with entries in contiguous area Primary 'system with tables,

 Fland substitution key A Fland substitution key that has a 'location' table 'entry in a continuous area.

 Communication channel,

 The primary system sends the alternate key location table change information, each accelerator 'system receives the alternate key' location 'table change information, changes the alternate key' land location 'table, and replaces it. A computer-based system characterized in that after receiving a change information of a key 'location' table, after a predetermined waiting time, an accelerator-system power is transmitted to the primary 'system' to notify a change completion.

 [4] A data record with a data item that contains a primary key and a substitute key, a substitute key 'entry consisting of a substitute key and a primary key, a substitute key' block containing a substitute key 'entry, and a substitute key.location.table. A primary 'system having an alternate key location' table with entries in a contiguous area;

 Fland substitution key A Fland substitution key that has a 'location' table 'entry in a continuous area.

 Communication channel,

 Rhythm ・ The node sends a transmission reminder message by broadcast, and after receiving a transmission reminder message by the accelerator system, it transmits to the accelerator system 'system-primary' system after a predetermined waiting time. A combination computer system.

 [5] Contiguous area with a data record that has data items including the primary key, a primary block that stores the data records in primary key order, and a case table entry that contains the address of each primary 'block' A primary 'system with a location' table and

A plurality of accelerators each having a land location table having a land location table table in a continuous area in which addresses of each primary block are described; Communication channel,

 A rhythm node that sends a transmission dunning message,

 Rhythm.The node sends a transmission reminder message to each accelerator's system, and each accelerator's system sends a processing request to the primary's system after a predetermined wait time upon receiving the transmission reminder message. A computer 'system.