JP6536237B2 - Control information setting apparatus, control information setting method, and control information setting program - Google Patents

Description

  The present invention relates to, for example, a control information setting device that manages a log buffer shared by a plurality of devices.

  Among information systems that write information in files, databases (hereinafter, may be referred to as "DB" in the present application), etc., an information system requiring high reliability uses information about write operations as log information, In many cases, a method of recording in a log buffer different from files and databases is adopted. For example, an information system requiring high reliability is a system in which it is not permissible to lose information that has been written. A database management system is an example of an information system that requires high reliability.

  For example, Patent Documents 1 to 3 disclose an example of a database management system that requires high reliability.

  The database management system disclosed in Patent Document 1 includes a shared DB data management unit, a data processing unit, and a prefetch unit. The shared DB data management unit executes the process of acquiring the data after executing exclusive control on the data in response to a request for the data processing unit to acquire the data. The shared DB data management unit further reads data to be prefetched in response to a request that the prefetch unit prefetches data.

  The database management system disclosed in Patent Document 2 includes a shared DB data management unit, a data processing unit, and a prefetch unit. The shared DB data management unit specifies the process associated with the request and the lower limit value associated with the request in response to the request for the prefetching unit to refer to the data. If the period between the current time and the time at which the latest input / output processing was performed is shorter than the specified lower limit, the shared DB data management unit continues until the period reaches the specified lower limit , Interrupt the identified process. The shared DB data management unit further executes exclusive control on the data and acquisition processing of the data in response to a request for the data processing unit to acquire the data.

  The database management apparatus disclosed in Patent Document 3 has a plurality of computers that manage data sets including a plurality of data. Among the plurality of computers, the first computer stores the first data in the memory in response to a request for storing the first data, and further creates first copy data which is a copy of the first data. The first computer calculates the number of times of transmitting the first copy data to the second computer based on the success probability that the copy data can be stored in the second computer. The first computer transmits the first duplicate data to the second computer for the calculated number of times.

  Hereinafter, for convenience of description, in the specification of the present application, an operation of storing (saving) information in a file, a database, or the like will be referred to as “writing”. An operation of storing (saving) log information regarding a writing operation in a log recording medium different from a database or the like is referred to as "recording". In addition, it is assumed that “parallel” includes “parallel” and “pseudo-parallel”.

  In database systems, processing requests (transactions) for a plurality of databases are often executed in parallel. For example, in the database system, a plurality of processes performing processing related to transactions operate in parallel. Even when multiple processes operate in parallel, the database system needs to properly execute the processing for transactions. Therefore, in the database system, a plurality of processes need to execute write operations to files and databases, and record log information regarding the write operations without contradiction.

  An example for realizing this process will be described. The database system has a log buffer capable of recording log information and its control information in a shared memory accessible to all processes that may record log information. Each process executing processing in the database system records log information regarding the write operation in the log buffer while maintaining the consistency of the processes with each other.

  Next, a memory device that realizes the log buffer will be described.

A memory device different from a DRAM is provided as a memory device constituting a computer system. Note that DRAM represents an abbreviation of Dynamic_Random_Access_Memory. The memory element has the following characteristics:
Even if power is not supplied to the computer system, the characteristic (nonvolatile) that the data held in the period when the power is supplied is not lost,
A feature (byte addressable) that data can be read from or written to the memory element directly by a machine language instruction to a central processing unit (CPU).

  The memory device can be realized, for example, by using a memory device having characteristics such as non-volatility, such as PRAM and MRAM. The memory device can be realized, for example, by combining a volatile memory device and a non-volatile but not byte addressable block access device such as a battery and an SSD. A computer system adopting the combination aspect uses non-volatile block access type data recorded in volatile memory elements using power supplied from a battery when power is not supplied. Store in the device. The computer system stores data stored in the block access type element in the volatile memory element when the power is restored. According to this combination aspect, even a volatile memory element can be treated as a non-volatile memory in actuality, so non-volatility can be realized.

  In the above description, PRAM represents an abbreviation of Parameter_Random_Access_Memory. MRAM stands for Magnetoresistive_Random_Access_Memory. SSD stands for Solid_State_Drive.

  Hereinafter, for convenience of description, the memory element having nonvolatileity is referred to as “nonvolatile memory (NVRAM)” regardless of the aspect of realizing nonvolatileity.

  Next, an example of using the non-volatile memory as described above as the log buffer in the database system will be described.

  Non-Patent Document 1 discloses a method using nonvolatile memory as a log buffer. According to Non-Patent Document 1, a plurality of processes acquire exclusive control of the log buffer by exclusive control of the log buffer, and the end of the log buffer is acquired within a period in which the acquired exclusive control is acquired. An operation (hereinafter, referred to as “log area securing operation”) for securing a storage area (log storage area) following “b” is executed. The storage area secured by the log area securing operation can store, for example, data representing the size (length, size) of log information to be stored (hereinafter referred to as “length”). It includes a length area, and a flag area capable of storing a flag representing a state in which storage processing related to the log information is being performed. Hereinafter, the flag is referred to as "copying flag".

  The process that has acquired control regarding the log buffer records log information in the secured storage area.

The log area securing operation can be realized, for example, by the following function 1 and function 2. That is,
Function 1: A function of updating control information on the log buffer (for example, a pointer indicating the end of the log buffer),
Function 2: The size of the log information is recorded in the length area of the secured storage area, and a value indicating that storage processing is in progress is recorded in the flag area of the secured storage area.

  According to Non-Patent Document 1, when a failure occurs in the own system, the database management system records a file storing contents at a certain temporary point (check point) of the database and an operation history after the check point Refer to log information. The database management system restores data up to the checkpoint based on the referenced file, and then performs a write operation on the referenced log information on the file to immediately before the failure occurs. Restore the contents of the database in.

  Non Patent Literature 2 discloses a method of reducing contention by using a plurality of log buffers.

Patent No. 5261809 Patent No. 5098120 Patent No. 5548829

Fang, Ru and Hsiao, Hui-I and He, Bin and Mohan, C and Wang, Yun, "High_performance_database_logging_using_storage_class_memory", Data_Engineering (ICDE), 2011, IEEE_27th_International_Conference, p1221-1231 Wang, Tianzheng and Johnson, Ryan, “Scalable logging through emerging non-volatile memory,” Very Large Data Bases (VLDB) 2014, p865-876 Intel 64 and IA-32 Architectures Software Developers Manual Volume 2A: Instruction Set Reference, AM, (http://www.tel.com/Assets/PDF/manual/253666.pdf)

  As described above, the database management system that manages a plurality of processing requests for the database performs exclusive control on log buffers such as log information shared with a plurality of information processing apparatuses. This is because the database management system needs to correctly execute processing regarding the log buffer shared by a plurality of information processing apparatuses.

  Even if the database management system has a plurality of arithmetic units that execute processing according to the processing request, the database management system must execute the processing sequentially while the exclusive control is being performed. It does not. For example, the devices (systems and systems) disclosed in Patent Document 1 to Patent Document 3, Non-Patent Document 1 and Non-Patent Document 2 are shared during the period in which the exclusive control right is held. Performs many processes sequentially on the log buffer.

  However, since the database management system has to execute processing sequentially while having the exclusive control right, for example, patent documents 1 to 3, non-patent documents 1 and non-patent documents The apparatus disclosed in Document 2 has a reduced processing efficiency.

  Therefore, a main object of the present invention is to provide a control information setting apparatus and the like for efficiently setting control information in a storage area, which can process a processing request for a log buffer in a short period of time.

In order to achieve the above object, in one aspect of the present invention, a control information setting device includes:
Whether the log information and the reachable state representing the state where it is possible to refer to the log information that follows the log information in a predetermined order, or the unreachable state representing the state where it is impossible to refer A log buffer in which storage areas that can be stored are logically arranged in the above order in a state in which they are associated with reachable information representing H, and control information including size information representing the size of the log information;
In the log buffer, the first control information is recorded from a first control information recorded in a storage area representing a boundary between a storage area to which the recording process for recording the log information has been allocated and a storage area to which the recording process is not allocated. The delivery information is read, the unreachable state is set to the reachable information included in the first control information, and a size is set to the size information included in the first control information, When the reachable information is in the reachable state, the second control information specifying that the size information satisfies the predetermined condition is specified according to the order in the log buffer, and the specified second control information is specified. And a reachable information setting unit that sets the included reachable information to the reachable state.

Further, as another aspect of the present invention, a control information setting method is:
Whether the log information and the reachable state representing the state where it is possible to refer to the log information that follows the log information in a predetermined order, or the unreachable state representing the state where it is impossible to refer With regard to a log buffer in which storage areas that can be stored are logically arranged in the above-described order in association with reachable information representing the control information and control information including size information representing the size of the log information.
In the log buffer, the first control information is recorded from a first control information recorded in a storage area representing a boundary between a storage area to which the recording process for recording the log information has been allocated and a storage area to which the recording process is not allocated. The delivery information is read, the unreachable state is set to the reachable information included in the first control information, and a size is set to the size information included in the first control information, When the reachable information is in the reachable state, the second control information specifying that the size information satisfies the predetermined condition is specified according to the order in the log buffer, and the specified second control information is specified. The included reachable information is set to the reachable state.

  Furthermore, the same object is also realized by the control information setting program and a computer readable recording medium for recording the program.

  According to the control information setting apparatus and the like according to the present invention, it is possible to efficiently set, in the storage area, control information that enables processing of a processing request to the log buffer in a short period of time.

It is a block diagram which shows the structure which the database management apparatus which concerns on the 1st Embodiment of this invention has. It is a figure which shows notionally the structure which the log buffer which concerns on 1st Embodiment has. It is a block diagram which shows the structure which the log information which concerns on 1st Embodiment has. It is a block diagram which shows the structure which a log buffer information storage part has. It is a flowchart (1/2) which shows the flow of the process in the database management apparatus which concerns on 1st Embodiment. It is a flowchart (2/2) which shows the flow of the process in the database management apparatus which concerns on 1st Embodiment. It is a figure which represents an example of control information notionally. Fig. 7 shows a flowchart showing the flow of processing in the database management device. It is a flowchart showing the flow of the process in the head position calculation part which concerns on 1st Embodiment. It is a flowchart showing the flow of the process in the log process part (1/3) which concerns on 1st Embodiment. It is a flowchart which shows the flow of the process in the log process part (2/3) which concerns on 1st Embodiment. It is a flowchart which shows the flow of the process in the log process part (3/3) which concerns on 1st Embodiment. It is a flowchart showing the flow of the process in the delivery location information update part which concerns on 1st Embodiment. It is a flowchart which shows the flow of the process in the delivery location information update part which concerns on 1st Embodiment. It is a flowchart showing the flow of the process in the completion positional information update part which concerns on 1st Embodiment. It is a block diagram which shows the structure which the database management apparatus which concerns on the 2nd Embodiment of this invention has. It is a flow chart which shows a flow of processing in control information setting device 151 concerning a 2nd embodiment. It is a block diagram showing roughly an example of hardware constitutions of a calculation processing device which can realize a database management device concerning each embodiment of the present invention.

  First, multiprocessor instructions capable of realizing part of the functions in the database management apparatus according to each embodiment of the present invention will be described. As a multiprocessor instruction, for example, Non-Patent Document 3 discloses a "cmpxchg" instruction which is an instruction related to an Intel x86 processor. The "cmpxchg" instruction represents a Compare_and_Exchange instruction. The "cmpxchg" instruction may also be referred to as a Compare_and_Swap (Cas) instruction. The multiprocessor instruction capable of realizing a part of the function is not limited to the above-described example.

  The "cmpxchg" instruction includes, for example, three operands of position information indicating the position of a certain storage area, a first value, and a second value. The position information is, for example, a name representing a register (eax register, rax register in the case of 64 bit_mode, and so on) in the database system. For convenience of explanation, the position information is assumed to be an eax register. The “cmpxchg” instruction is an instruction that atomically executes a series of operations shown in process 1 and process 3 (described later).

Process 1: Read the first value,
Process 2: When the read first value matches the value stored in the eax register, the second value is stored in the memory area storing the first value.
Process 3: When the read first value does not match the value stored in the eax register, the read first value is stored in the eax register.

  Next, atomic will be described. In atomic processing, hardware guarantees that a processor different from the one processor can not execute the processing for the storage area while the processor performs a series of processing for the storage area. Indicates that it is a process. For example, the “cmpxchg” instruction ensures by hardware that one processor does not access the storage area while another processor is executing the processing 1 to the processing 3 while the processing is executed to the storage area. Indicates that the process is

  Next, embodiments of the present invention will be described in detail with reference to the drawings.

First Embodiment
The configuration of the database management apparatus 101 according to the first embodiment of the present invention will be described in detail with reference to FIG. FIG. 1 is a block diagram showing a configuration of the database management apparatus 101 according to the first embodiment of the present invention.

  The database management apparatus 101 according to the first embodiment of the present invention roughly includes a plurality of transaction execution units, a log buffer 112, and a log buffer information storage unit 111. The plurality of transaction execution units are, for example, a transaction execution unit 107, a transaction execution unit 109, and a transaction execution unit 110. The transaction execution unit 107 has a log processing unit 102 and a memory 108. The log processing unit 102 includes a head position calculation unit 103 and a log information storage unit 104. The log information storage unit 104 has a delivery position information update unit 105 and a completion position information update unit 106. The transaction execution unit 109 and the transaction execution unit 110 have the same configuration as the transaction execution unit 107.

  Although the database management apparatus 101 illustrated in FIG. 1 has three transaction execution units, it does not have to be three.

  The log buffer 112 is implemented using, for example, an NVRAM that represents a non-volatile memory element. The configuration of the log buffer 112 will be described in detail with reference to FIG. FIG. 2 is a diagram conceptually showing the structure of the log buffer 112 according to the first embodiment.

  The log buffer 112 can store, for example, at least one log information set in which log information output in a write operation related to log information and control information related to the log information are associated. In the example illustrated in FIG. 2, the log buffer 112 includes a first log information set representing information associated with the log information 1 and the control information 1, and second log information representing information associated with the log information 2 and the control information 2. Includes sets, etc.

  The log information set may be recorded in the log buffer 112 in the order of control information and log information associated with the control information. In the present embodiment, it is assumed that the control information and the log information associated with the control information are stored in the log buffer 112 at least logically in the order of recording. For example, the control information is a size to which the above-described "cas" instruction can be applied, for example, information having a size of one word.

  Control information related to log information will be described in detail with reference to FIG. FIG. 3 is a block diagram showing a configuration of log information according to the first embodiment.

  The control information includes reachable information, recording operation completion information, recording operation state information, and size (length, length, size) information. That is, the control information has a data structure including reachable information.

  The reachable information is information indicating whether or not it is possible to refer to log information sets located in a later order with respect to each log information set in the log buffer 112. That is, in the log buffer 112, the reachable information may be in a reachable state representing a state where it is possible to refer to the log information set following a certain log information set in a predetermined order, or may be referred to. It is information indicating whether it is an unreachable state representing an impossible state. For example, the state where it is possible to refer to the log information set located in back order is 1 (valid), and the value where it is impossible to refer to the log information set located in back order is 0 (invalid) . The reachable information is information (for example, a flag) that is referred to when it is determined whether to update the later-described reachable position information (described later). In the following description, for convenience of explanation, it is assumed that a storage area capable of storing reachable information is represented as a variable “rflag”. The state where it is possible to refer to the log information set located in the later order is referred to as the “delivery state”. In addition, a state in which it is impossible to refer to the log information sets located in the later order is referred to as an “unreachable state”. The variable “rflag” represents the log buffer 112 or a specific storage area in the memory 108. The reachable information may be expressed as, for example, a reachable state control flag. For example, the reachable information is in the reachable state before the process related to the log information associated with the control information including the reachable information is assigned, and is in the unreachable state after the process is assigned.

  The recording operation completion information is information indicating whether or not the process of recording log information in the log buffer 112 is completed. For example, the state before the process of recording the log information in the log buffer 112 is complete is 1 (valid), and the value is 0 (invalid) when the process is complete. The recording operation completion information is information (for example, a flag) that is referred to when it is determined whether to update completion position information (described later) described later. In the following description, a storage area capable of storing the recording operation completion information is represented as a variable "cflag". It is assumed that a state before the process is completed is referred to as an “incomplete state”. Further, a state after the completion of the process is referred to as a "completed state". For example, the variable "cflag" represents the log buffer 112 or a specific storage area in the memory 108. The recording operation completion information may be expressed as, for example, a completion control flag. For example, the recording operation completion information indicates the uncompleted state when the process related to the log information associated with the control information including the recording operation completion information is before completion, and indicates the completion state after the completion.

  The recording operation state information is information indicating whether or not processing for recording log information in the log buffer 112 is being performed. For example, the processing state is 1 and the non-processing state is 0. In the following description, the storage area capable of storing the recording operation state information is represented as a variable “inCopy”. For example, the variable "inCopy" represents the log buffer 112 or a specific storage area in the memory 108. The recording operation state information may be expressed as an in-process flag. Furthermore, in the following description, the state in which the process is being executed is referred to as the “in-process state”. A state in which the process is not executed is represented as a "non-processing state".

  The size information indicates the size (size, length, length) of the log information. In the following description, a storage area capable of storing size information is also referred to as a variable “length”. For example, the variable "length" represents the log buffer 112 or a specific storage area in the memory 108.

  Next, the configuration of the log buffer information storage unit 111 will be described with reference to FIG. FIG. 4 is a block diagram showing a configuration of the log buffer information storage unit 111.

  The log buffer information storage unit 111 includes reachable position information, complete position information, and head position information.

  The reachable position information stores, in the log buffer 112, among the log information sets including the reachable state representing the state where the process of recording the log information has been started, the log information set at which the position is located in the rearmost order is stored. Represents a position. Hereinafter, for convenience of explanation, it is assumed that a storage area in the log buffer information storage unit 111 capable of storing reachable position information is represented as a variable “ReachableUpTo”.

  The completion position information indicates the position in the log buffer 112 in which the processing for recording the log information is completed, and among the log information sets in the completion state, the position at which the log information set is located at the rearmost position is stored. . Hereinafter, for convenience of explanation, a storage area in the log buffer information storage unit 111 capable of storing the completed position information is represented as a variable “FinishedUpTo”.

  The head position information indicates a position where the log buffer 112 can store a log information set that has not yet been assigned a process of recording log information. Hereinafter, for convenience of explanation, it is assumed that a storage area of the log buffer information storage unit 111 which can store the head position information is represented as a variable “TAIL”.

  For convenience of explanation, in the present embodiment, it is assumed that the log buffer 112 stores N log information sets. Also, it is assumed that the log identifier that can uniquely identify the log information is the order of the log information set in the log buffer 112. For example, in the example shown in FIG. 2, it is represented as log information I (however, I represents a natural number), and an I-th log information set representing information associated with control information I. Also, the head position information is the position closest to the storage area where N log information sets are stored among the areas capable of storing a new log information set (in the case of this example, the (N + 1) th log Let's say that it represents an information set.

  In the following description, further, regarding control information included in the (N + 1) th log information set, the variable “rflag” stores 1 (that is, the reachable state), and the variable “cflag” is 1 ((1) That is, it is assumed that the uncompleted state is stored. Furthermore, regarding the control information, it is assumed that the variable “inCopy” stores 0 (that is, a non-processing state) and the variable “length” stores 0.

  Next, processing in the database management apparatus 101 according to the present embodiment will be described with reference to FIGS. 5 and 6. 5 and 6 are flowcharts showing the flow of processing in the database management apparatus 101 according to the first embodiment.

  The transaction execution unit 107 executes a process related to the transaction in response to a request for instructing a transaction related to a search process or the like to be executed on the database (step S101).

  In the process shown in step S101, for example, the transaction execution unit 107 starts position information indicating position information indicating a storage area in which log information related to a certain write operation is stored and size information indicating the size of the log information. Output (step S102).

  The start position calculation unit 103 inputs the size information output by the transaction execution unit 107. The start position calculation unit 103 gives priority to the process related to the storage area (for example, variable “TAIL”) storing the start position information in the log buffer information storage unit 111 shared by the plurality of transaction execution units. The exclusive control right that can be executed is acquired (step S103). The start position calculation unit 103 reads the start position information stored in the variable “TAIL” within a period in which the acquired exclusive control right is held (step S104).

  The start position calculation unit 103 adds the size represented by the input size information and the read start position information (step S105), and has the acquired exclusive control right as a value calculated as a result of addition. Within the period, it is stored in the variable “TAIL” in the log buffer information storage unit 111 (step S106). The start position calculation unit 103 releases the acquired exclusive control right (step S107). The start position calculation unit 103 outputs the start position information read in step S104 to the log information storage unit 104 as first start information (step S108).

  The information acquired by the start position calculation unit 103 is different among the plurality of transaction execution units under the control using the exclusive control right. In addition, the head position calculation unit in another transaction execution unit (for example, the transaction execution unit 109 or the transaction execution unit 110) acquires head position information by releasing the exclusive control right by the head position calculation unit 103. be able to.

  The log information storage unit 104 inputs the first head information output by the head position calculation unit 103, the size of log information output by the transaction execution unit 107, and the position information output by the transaction execution unit 107. The log information storage unit 104 stores the control information stored in the storage area starting from the read head position information in the log buffer 112 (that is, the position where the (N + 1) th log information set is stored). The readability information included in the read and read control information is stored in the memory 108.

  The log information storage unit 104 is control information including the unreachable state, the recording operation completion information included in the input control information, the processing in progress state, and the size of the input log information (hereinafter referred to as “second control information”). Is represented (step S111). The log information storage unit 104 records the created second control information in the storage area starting from the input first head information (step S112).

  When the deliverable information stored in the memory 108 is in the deliverable state (YES in step S113), the log information storage unit 104 performs an operation (described later) for updating the deliverable position information to the deliverable position information update unit 105. (Step S114). When the reachable information stored in the memory 108 is not in the reachable state (NO in step S113), the log information storage unit 104 does not instruct the reachable position information update unit 105 to update the reachable position information.

  The log information storage unit 104 is stored in the storage area starting from the first head information input prior to step S111 in the log buffer 112 and in the storage area starting from position information input prior to step S111. Log information is recorded (step S115). The log information storage unit 104 reads the control information stored in the storage area starting from the read first head information, and stores the recording operation completion information included in the read control information in the memory 108 (see FIG. Step S116). The log information storage unit 104 controls the control information including the reachable information in the control information read in step S116, the completed state, the non-processing state, and the size of the log information in the control information (hereinafter referred to as “third control information Is represented (step S117).

  When the recording operation completion information stored in the memory 108 in step S116 is in the uncompleted state (YES in step S118), the log information storage unit 104 completes the processing (described later) of updating the completion position information. It instructs the updating unit 106 (step S119). When the recording operation completion information stored in the memory 108 in step S116 is not in the uncompleted state (NO in step S118), the log information storage unit 104 causes the completion position information updating unit 106 to update the completion position information. I do not instruct.

  The log information storage unit 104 compares the value obtained by adding the input first head information and the input size (that is, the second head information) with the reachable position information stored in the log buffer information storage unit 111. . If the reachable position information is equal to or greater than the second head information (YES in step S120), the log information storage unit 104 ends the operation of recording the log information. On the other hand, when the reachable position information is less than the second head information (NO in step S120), the log information storage unit 104 waits until the reachable position information becomes equal to or more than the second head information. A process is performed (step S121).

  Next, processing according to the present embodiment will be described with reference to an example of an information processing apparatus having a plurality of processors.

  For convenience of explanation, control information represents data that can be handled by atomic operations by multiprocessor instructions. Also, atomic operations by multiprocessor instructions can be implemented, for example, using the "cmpxchg" instruction described above. In this example, the process of acquiring the exclusive control right is represented as "LockAcquire", and the process of releasing the acquired exclusive control right is represented as "LockRelease".

  The process of acquiring the exclusive control right and the process of releasing the exclusive control right are often implemented using atomic operations via multiprocessor instructions. There are various methods for realizing the process of acquiring the exclusive control right and the process of releasing the acquired exclusive control right. Here, the detailed description of the method for realizing the exclusive control right is omitted.

  Furthermore, the database management apparatus 101 completes the process of recording in the log buffer 112 up to the N-th log information set, and then refers to the database according to the present embodiment with reference to an example in the case where a new transaction is not processed. The process in the management apparatus 101 will be described. Further, in this example, it is assumed that the variable “ReachableUpTo”, the variable “FinishedUpTo”, and the variable “TAIL” first store a position representing a storage area where the (N + 1) th log information set is recorded. .

  Also, in the example, as shown in FIG. 7, regarding the control information included in the (N + 1) th log information set, the variable “rflag” capable of recording the reachable information is in the reachable state (ie, Suppose that 1) is recorded. FIG. 7 is a diagram conceptually illustrating an example of control information. It is assumed that the variable “cflag” capable of recording the recording operation completion information records the uncompleted state (ie, 1). Regarding the control information, it is assumed that the variable “inCopy” capable of recording the recording operation state information records the non-processing state (ie, 0). It is assumed that the variable “length” capable of recording size information records 0. That is, in the control information included in the (N + 1) th log information set, the reachable information is in the reachable state, and the recording operation completion information is in the uncompleted state. Further, in the control information included in the (N + 1) th log information set, the recording operation state information is in the non-processing state, and the size information is 0.

  Each process is assumed to be allocated to an area in a shared memory (not shown in FIG. 1). That is, the shared memory includes a storage area for the I-th (where 1 ≦ I ≦ n) process. The process I stores log information generated along with execution of transaction processing by the transaction execution unit 107 in the log information area in the storage area allocated to the process I.

  The process in the database management apparatus 101 will be described with reference to FIG. FIG. 8 shows a flowchart showing the flow of processing in the database management apparatus 101.

  The process in the database management apparatus 101 is roughly divided into a process of calculating first head information indicating a storage area for recording a log information set (step S125), and a log information set is recorded in an area indicated by the calculated first head information. And (step S126).

  An outline of the process will be described for the processes shown in step S125 and step S126.

Step S125:
The start position calculation unit 103 receives, for example, the size information output from the transaction execution unit 107. The head position calculation unit 103 reads the head position information in the log buffer information storage unit 111, and outputs the read head position information (that is, the first head information described above) to the log information storage unit 104. The process performed by the start position calculation unit 103 can be realized, for example, using a “reserve” function that can reserve a storage area. The "reserve" function represents, for example, a function that receives information indicating the size of log information and reserves a storage area for the input size.

Step S126:
The log information storage unit 104 inputs the first head information output by the head position calculation unit 103, the size of log information output by the transaction execution unit 107, and the position information output by the transaction execution unit 107. The log information storage unit 104 records the log information set related to the log information stored in the position information in the storage area represented by the first head information input in the log buffer 112.

  The processing executed by the log information storage unit 104 can be realized, for example, using a copy function capable of storing data in a storage area. The log information storage unit 104 inputs three pieces of information shown in the following input 1 to input 3 by processing according to the copy function, for example.

Input 1: Information indicating a position (first position) in the process-specific memory where log information to be recorded in the storage area is stored,
Input 2: Size information indicating the size of the log information
Input 3: Information indicating the position (second position) of the storage area secured by the log area securing operation or the like.

  The log information storage unit 104 processes the log information stored in the storage area starting from the first position into the storage area starting from the second position, for example, by processing according to the copy function. Record.

  Next, before describing the processes in steps S125 and S126 in detail, variables representing a storage area in the memory 108 or the like will be described.

  For convenience of explanation, in each relevant transaction execution unit, the variables "myLSN", "myLog", "myLength", "nextLSN", "myTurn", "oldv", and "newv" It is assumed that a specific storage area in the memory in the transaction execution unit is represented. Also, buffer represents the log buffer 112. Furthermore, buffer [myLSN] represents a storage area of the log buffer 112 indicated by the value stored in the variable “myLSN”.

  Regarding control information, assume that reachability information is stored, for example, in a variable "rflag" that represents a storage area in the log buffer 112. It is assumed that the recording operation completion information is stored, for example, in a variable "cflag" that represents a storage area in the log buffer 112. For example, it is assumed that the recording operation state information is stored in a variable “inCopy” that represents a storage area in the log buffer 112. The size information is assumed to be stored, for example, in a variable "length" that represents a storage area in the log buffer 112.

  It is assumed that the reachable position information is stored in a variable “ReachableUpTo” that represents a storage area in the log buffer information storage unit 111. It is assumed that the completed position information is stored in a variable “FinishedUpTo” representing a storage area in the log buffer information storage unit 111. It is assumed that the head position information is stored in a variable “TAIL” that represents a storage area in the log buffer information storage unit 111.

  The process shown in step S125 (FIG. 8) performed by the start position calculation unit 103 will be described in more detail with reference to FIG. FIG. 9 is a flowchart showing processing in the head position calculation unit 103 according to the first embodiment.

  The start position calculation unit 103 acquires the exclusive control right regarding the variable “TAIL” in which the start position information is stored in the log buffer information storage unit 111 (step S131). The start position calculation unit 103 reads the start position information stored in the variable “TAIL” within a period in which the acquired exclusive control right is held (step S132). The start position calculation unit 103 stores the read start position information in the variable “myLSN” (step S133). The start position calculation unit 103 adds the size included in the input size information to the read start position information (step S134), and the calculated value is stored in the log during the period in which the acquired exclusive control right is held. The variable “TAIL” in the buffer information storage unit 111 is stored (step S135). The start position calculation unit 103 releases the acquired exclusive control right (step S136), and outputs the value stored in the variable “myLSN” (that is, the first start information) to the log information storage unit 104 (step S137).

  That is, by the processing shown in steps S131 to S137, the head position calculation unit 103 outputs the first head information indicating the storage area to which the log information is not allocated in the log buffer 112, and further relates to the head position information. Update the storage area. In the processes shown in steps S131 to S137, for example, the start position calculation unit 103 reads the start position information by executing the same process as the process instructed by the "reserve" function. The start position calculation unit 103 acquires the exclusive control right, for example, by executing the same process as the process instructed by the function “Lock_Acquire”.

  The process executed by the log information storage unit 104 in step S126 (FIG. 8) will be described in more detail with reference to FIG. FIG. 10 is a flowchart (1/3) showing the process in the log processing unit 102 according to the first embodiment.

  The log information storage unit 104 inputs the position information output by the transaction execution unit 107, the size information output by the transaction execution unit 107, and the first head information output by the head position calculation unit 103. More specifically, the log information storage unit 104 stores the input position information in the variable "myLog", stores the input size information in the variable "myLength", and inputs the first head information input, the variable "myLSN" Store in

  The log information storage unit 104 adds the input first head information and the input size information, and then adds head position information (that is, “second head information”) to which the log information set can be recorded. calculate. More specifically, the log information storage unit 104 stores the value stored in the variable "myLength" storing the size information and the variable "myLSN" storing the first head information. Are added (step S141), and the calculated result (that is, "the second head information") is stored in the variable "nextLSN" (step S142).

  Next, the log information storage unit 104 sets the log information set recorded in the storage area represented by the input first head information in the log buffer 112, and the control information included in the log information set (hereinafter referred to as “the log information set”). Read “1st control information”. More specifically, the log information storage unit 104 reads the log information set stored in the storage area buffer [myLSN] (step S143), and among the read log information set, sets the first control information to the variable “oldv”. (Step S144).

  Next, the log information storage unit 104 reads the reachable information included in the read first control information, and stores the read reachability information in the memory 108. More specifically, the log information storage unit 104 reads the reachable information stored in the variable "rflag" among the first control information stored in the variable "oldv", and reads the read deliverable information as a variable. It stores in "myTurn" (step S145).

  Next, the log information storage unit 104 includes control information including the unreachable state, the recording operation completion information included in the input control information, the processing in progress state, and the size represented by the size information of the input log information , “2nd control information” is created. More specifically, the log information storage unit 104 stores the first control information stored in the variable "oldv" in the variable "newv" (step S146). The log information storage unit 104 stores the size information of the variable "myLength" in the variable "length" included in the variable "newv" (step S147). Furthermore, the log information storage unit 104 stores the in-processing state (for example, 1) in the variable “inCopy” included in the variable “newv”. Furthermore, the log information storage unit 104 stores the unreachable state (for example, 0) in the variable "rflag" included in the variable "newv" (step S148).

  Next, the log information storage unit 104 stores the second control information in the storage area represented by the input first head information by executing the same process as the process executed by the “cas” instruction.

  The process of storing the second control information will be described in detail. The log information storage unit 104 reads control information (hereinafter referred to as “third control information”) in the log information set recorded in the storage area represented by the input first head information, and the read third control information and , It is determined whether or not the first control information matches.

  When the first control information and the third control information match, the log information storage unit 104 stores the second control information in the storage area represented by the input first head information. That is, in this case, the log information storage unit 104 performs control including the unreachable state, the recording operation completion information included in the input control information, the processing in progress state, and the size represented by the size information of the input log information. The information is stored in a storage area in which the log information set is stored.

  If the third control information does not match the first control information, the log information storage unit 104 sets the read third control information as the first control information. Thereafter, the log information storage unit 104 reads out the reachable information included in the read first control information (corresponding to step S145), or the third control information matches the first control information. A process (corresponding to step S149 (described later)) is executed to determine whether or not this is the case. That is, the log information storage unit 104 executes the processing shown in step S145 to step S150 (described later) until the created second control information is stored in the storage area represented by the input first head information.

  A process similar to the process executed by the “cas” instruction as described above is performed atomically. More specifically, a process of executing the same process as the process executed by the “cas” instruction will be described.

  The log information storage unit 104 determines whether the first control information stored in the variable “oldv” matches the control information stored in the storage area buffer [myLSN] (step S149). .

  When the first control information stored in the variable “oldv” matches the third control information stored in the storage area buffer [myLSN] (in step S149) YES), the second control information stored in the variable "newv" is recorded in the storage area buffer [myLSN] (step S151).

  If the third control information stored in storage area buffer [myLSN] does not match the first control information stored in variable “oldv” (NO in step S 149) And the third control information recorded in the storage area buffer [myLSN] is stored in the variable “oldv” (step S150). Thereafter, the log information storage unit 104 executes the processes shown in steps S145 to S149.

  That is, the processes shown in step S149 and step S150 are performed atomically in the same manner as the process executed by the "cas" instruction.

  Next, when the reachable information read in step S145 is in the reachable state, the log information storage unit 104 updates the reachable position information stored in the log buffer information storage unit 111 (described later). ) To the delivery location information updating unit 105. The log information storage unit 104 instructs, for example, the reachable position information update unit 105 to perform an update operation by outputting the second head information to the reachable position information update unit 105. The log information storage unit 104 updates the reachable position information stored in the log buffer information storage unit 111 when the reachable information read in step S145 is not in the reachable state. Do not instruct 105.

  More specifically, the log information storage unit 104 reads the reachable information stored in the variable "myTurn", and determines whether the read reachable information is in the reachable state (step S152). When the reachable information is in the reachable state (YES in step S152), the log information storage unit 104 sends the second head information stored in the variable “nextLSN” to the reachable position information update unit 105. By outputting the information, the delivery position information updating unit 105 is instructed to perform an update operation regarding the delivery position information (step S153). That is, the log information storage unit 104 instructs the reachable position information update unit 105 to update the reachable position information stored in the variable “ReachableUpTo”. On the other hand, when the reachable information is not in the reachable state (NO in step S152), the log information storage unit 104 does not instruct the reachable position information update unit 105 to perform an update operation regarding the reachable position information. .

  Next, the log information storage unit 104 records the log information stored in the storage area starting from the input position information in the storage area starting from the input first head information in the log buffer 112. . More specifically, the log information storage unit 104 stores the log information stored in the storage area represented by the position information stored in the variable "myLog" in the storage area buffer [myLSN] in the variable "myLength". Only the size of the value being recorded is recorded (step S154).

  The process following step S154 will be described in detail with reference to FIG. FIG. 11 is a flowchart (2/3) showing the flow of processing in the log processing unit 102 according to the first embodiment.

  In the log buffer 112, the log buffer information storage unit 111 stores a log information set recorded in a storage area represented by the input first head information, and control information included in the log information set (hereinafter referred to as “ 4) Read "Control information". More specifically, the log information storage unit 104 reads the log information set from the storage area buffer [myLSN] indicated by the first head information stored in the variable “myLSN”, and the fourth control information in the read log information set Are stored in the variable "oldv" (step S160).

  Next, the log buffer information storage unit 111 reads the recording operation completion information stored in the read fourth control information. More specifically, the log information storage unit 104 reads the recording operation completion information from the variable "cflag" included in the variable "oldv" (step S161), and, for example, the variable "myTurn" (Step S162).

  Next, the log information storage unit 104 represents control information including the unreachable state, the complete state, the non-processing state, and the size information stored in the third control information (hereinafter referred to as “fifth control information”). Create). More specifically, the log information storage unit 104 stores the control information stored in the variable "oldv" in the variable "newv" (step S163). The log information storage unit 104 stores the non-processing state (for example, 0) representing not being processed in the variable “inCopy” in the variable “newv” (step S164), and the variable “cflag” in the variable “newv” Stores a completion status (for example, 0) indicating completion (step S165). Furthermore, the log information storage unit 104 stores the unreachable state (for example, 0) indicating that the state is not the reachable state in the variable “rflag” of the variable “newv” (step S166).

  Next, the log information storage unit 104 records the fifth control information in the storage area represented by the input first head information by executing the same process as the process executed by the “cas” instruction.

  The process of recording the fifth control information will be described in detail. The log information storage unit 104 reads control information (hereinafter referred to as “sixth control information”) in the log information set recorded in the storage area represented by the input first head information, and the read sixth control information , It is determined whether the fourth control information matches.

  When the sixth control information matches the fourth control information, the log information storage unit 104 records fifth control information in the storage area represented by the input first head information. That is, in this case, the log information storage unit 104 includes the fifth information including the information indicating that it is not in the reachable state, the completion state, the information indicating that the recording operation is not in progress, and the size indicated by the size information of the input log information. Control information is recorded in a storage area in which the log information set is stored.

  If the sixth control information does not match the fourth control information, the log information storage unit 104 sets the read sixth control information as the fourth control information, and then reads the recording operation completion information, or , And the process of comparing the sixth control information with the fourth control information. That is, the log information storage unit 104 repeats the process until the created control information is recorded in the storage area represented by the input first head information.

  The log information storage unit 104 atomically executes the same process as the process executed by the “cas” instruction as described above.

  More specifically, a process of executing the same process as the process executed by the “cas” instruction will be described. The log information storage unit 104 determines whether the fourth control information stored in the variable “oldv” matches the sixth control information stored in the storage area buffer [myLSN] (step S167). . If the sixth control information stored in the storage area buffer [myLSN] matches the fourth control information stored in the variable “oldv” (YES in step S 167), the log information storage unit 104, The fifth control information stored in the variable "newv" is recorded in the storage area buffer [myLSN] (step S169).

  If the sixth control information stored in the storage area buffer [myLSN] does not match the fourth control information stored in the variable “oldv” (NO in step S 167), the log information storage unit 104 The sixth control information stored in the storage area buffer [myLSN] is stored in the variable “oldv” (step S168). Thereafter, the log information storage unit 104 executes the processes shown in steps S162 to S167.

  The log information storage unit 104 atomically executes the same process as the process executed by the “cas” instruction as described above.

  Next, when the recording operation completion information read in step S162 is in an uncompleted state indicating before completion, the log information storage unit 104 updates the completion position information in the log buffer information storage unit 111 (described later) ) Is instructed to the completion position information updating unit 106. The log information storage unit 104 instructs the completion position information updating unit 106 to perform an update operation, for example, by outputting the second head information to the completion position information updating unit 106. The log information storage unit 104 does not instruct the completion position information updating unit 106 to update the completion position information when the recording operation completion information read in step S162 is not in an incomplete state.

  More specifically, the log information storage unit 104 reads the recording operation completion information stored in the variable "myTurn", and determines whether the read recording operation completion information is in an uncompleted state (step S170). ). When the recording operation completion information is in the uncompleted state (YES in step S170), the log information storage unit 104 sends the second head information stored in the variable “nextLSN” to the completion position information updating unit 106. By outputting the data, the completion position information updating unit 106 is instructed to perform an update operation regarding the completion position information (step S171). That is, the log information storage unit 104 instructs the completed position information updating unit 106 to update the completed position information stored in the variable “FinishedUpTo”. If the recording operation completion information is not in the uncompleted state (NO in step S170), the log information storage unit 104 does not instruct the process of updating the completion position information.

  Next, the process following step S170 (or step S171) will be described in detail with reference to FIG. FIG. 12 is a flowchart (3/3) showing the flow of processing in the log processing unit 102 according to the first embodiment.

  The log information storage unit 104 reads the reachable position information from the log buffer information storage unit 111, and compares the read reachable position information with the second head information calculated in step S141 (FIG. 10). If the read delivery position information is less than the second head information, the log information storage unit 104 continues until the reachable position information stored in the log buffer information storage unit 111 becomes equal to or more than the second head information. Execute processing to wait.

  More specifically, the log information storage unit 104 reads the reachable position information stored in the variable “ReachableUpTo” in the log buffer information storage unit 111 and stores the read reachable position information and the variable “nextLSN”. The second leading information is compared (step S181). When the reachable position information is less than the second head information (NO in step S181), the log information storage unit 104 stores the reachable position stored in the variable “ReachableUpTo” in the log buffer information storage unit 111. A process of waiting until the information reaches or exceeds the second head information is executed (step S182). For example, the process of waiting to be performed by the log information storage unit 104 can be realized by using a “Sleep” function having the value of the variable “nextLSN” as an input.

  A process performed by the reachable position information updating unit 105 will be described with reference to FIG. FIG. 13 is a flowchart showing processing in the reachable position information updating unit 105 according to the first embodiment.

  The delivered position information updating unit 105 receives the second head information, and stores the input second head information in the memory 108 (hereinafter referred to as “first area”). The reachable position information updating unit 105 reads the value stored in the memory 108, and stores the read value in the log buffer information storage unit 111 as reachable position information. More specifically, the delivered position information updating unit 105 inputs the second head information output from the log information storage unit 104, and stores the input second head information in the variable "nextLSN". The delivered position information updating unit 105 stores, in the variable “lsn”, the second head information stored in the variable “nextLSN” (step S191). The delivered position information updating unit 105 stores the value stored in the variable “lsn” in the variable “ReachableUpTo” in the log buffer information storage unit 111 (step S192).

  Next, in the log buffer 112, the reachable position information updating unit 105 sets the log information set stored in the storage area pointed to by the value stored in the first area, and the control information in the log information set , “7th control information”). More specifically, the delivered position information updating unit 105 reads the control information recorded in the storage area buffer [lsn] indicated by the value stored in the variable “lsn” (step S193). The delivered position information updating unit 105 stores the seventh control information included in the read log information set in the variable “oldv” (step S194).

  Next, the delivered position information updating unit 105 determines whether the size in the size information included in the read seventh control information is 0 or not. When the size is not 0, the reachable position information updating unit 105 adds the size included in the seventh control information to the value stored in the first area, and calculates the calculated value. Store in the area. Thereafter, the process is repeatedly executed until the size represented by the size information included in the control information becomes zero. On the other hand, when the size is 0, the reachable position information update unit 105 has a reachable state indicating that the process has not started yet, and the recording operation completion information included in the seventh control information. Control information (hereinafter referred to as “eighth control information”) including recording operation state information included in the seventh control information and size information included in the seventh control information is created. In this case, the eighth control information created in step S197 and step S198 is a process in which the process shown in step S195 is executed, and therefore, 0 is included as size information.

  More specifically, the delivered location information updating unit 105 determines whether the size stored in the variable “length” included in the seventh control information stored in the variable “oldv” is 0 or not. (Step S195). If the size is not 0 (NO in step S 195), the delivered position information updating unit 105 adds the value stored in the variable “lsn” and the value stored in the variable “length”. , And stores the calculated result in the variable "lsn" (step S196). Thereafter, the delivered position information updating unit 105 executes the process shown in step S192. On the other hand, when the size is 0 (YES in step S 195), the delivered position information updating unit 105 sets the seventh control information stored in the variable “oldv” to the variable “newv”. It stores it (step S197), and stores the reachable state indicating that the process has not started yet as the value of the reachable information in the variable "newv" (step S198). The variable “newv” stores eighth control information by the processes shown in steps S197 and S198.

  The delivered position information updating unit 105 executes the same process as the process executed by the “cas” instruction to store the generated eighth control information in the first area in the log buffer 112. Record in the storage area pointed to by.

  A process of recording the eighth control information in the storage area indicated by the value stored in the first area in the log buffer 112 will be described in detail. The delivered position information updating unit 105 reads control information (hereinafter, “ninth control information”) stored in the storage area indicated by the value stored in the first area in the log buffer 112. When the read ninth control information matches the read seventh control information, the delivered position information updating unit 105 stores the generated eighth control information in the first area in the log buffer 112. Record in the storage area pointed to by the value. If the read ninth control information and the read seventh control information do not match, the reachable position information updating unit 105 “whether or not the size in the size information included in the seventh control information is 0. Or the process of “determining whether the ninth control information matches the seventh control information” is executed.

  The reachable position information updating unit 105 atomically executes the same process as the process executed by the “cas” instruction as described above.

  More specifically, the delivered position information updating unit 105 reads the ninth control information stored in the storage area buffer [lsn], and reads the read ninth control information and the variable “oldv”. 7 Compare with the control information (step S199). When the seventh control information stored in the variable “oldv” matches the ninth control information (YES in step S199), the delivered position information updating unit 105 stores the information in the variable “newv”. The present eighth control information is recorded in the storage area buffer [lsn] (step S201). When the seventh control information stored in variable “oldv” does not match the ninth control information (NO in step S199), delivered position information updating unit 105 stores the information in storage area buffer [lsn]. The stored ninth control information is stored in a variable "oldv" (step S200). Thereafter, the delivered location information updating unit 105 executes the process shown in step S195. The log information storage unit 104 atomically executes the same process as the process executed by the “cas” instruction as described above.

  Next, the process following step S201 will be described in detail with reference to FIG. FIG. 14 is a flowchart showing a flow of processing in the reachable position information updating unit 105 according to the first embodiment.

  The reachable position information updating unit 105 compares the value stored in the first area with the reachable position information in the log buffer information storage unit 111. The reachable position information updating unit 105 instructs the log information storage unit 104 storing a value smaller than the reachable position information in the first area to end the queuing process. The reachable position information updating unit 105 that has received the instruction ends the waiting process.

  More specifically, the reachable position information updating unit 105 executes, for example, the same processing as the processing in the "WakeUp" function. The reachable position information updating unit 105 reads a value from the variable “lsn”, and compares the read value with the reachable position information in the log buffer information storage unit 111. The reachable position information updating unit 105 identifies the log information storage unit whose read value is less than the reachable position information, and executes an instruction to end the queuing process in the identified log information storage unit (step S211). The log information storage unit having received the instruction ends the waiting process.

  A process performed by the completion position information update unit 106 will be described with reference to FIG. FIG. 15 is a flowchart showing processing in the completion position information updating unit 106 according to the first embodiment.

  The completion position information updating unit 106 inputs the second head information, and stores the input second head information in the memory 108 (hereinafter referred to as “second area”). More specifically, the completion position information updating unit 106 inputs the second head information output from the log information storage unit 104, and stores the input second head information in the variable "nextLSN". The complete position information updating unit 106 stores the second head information stored in the variable “nextLSN” in the variable “lsn” (step S221).

  Next, the completion position information updating unit 106 stores the value stored in the second area in the log buffer information storage unit 111 as completion position information. More specifically, the completed position information updating unit 106 stores the value stored in the variable “lsn” in the variable “FinishedUpTo” in the log buffer information storage unit 111 (step S222).

  Next, the completion position information updating unit 106 reads control information (hereinafter referred to as “tenth control information”) stored in the storage area pointed to by the value stored in the second area in the log buffer 112. . The delivered position information updating unit 105 determines whether the size in the size information included in the read tenth control information is 0 or not, and the recording operation state information in the tenth control information is in the processing state. At least one is determined.

  When the size is not 0 and the recording operation state information is in the non-processing state, the completion position information updating unit 106 reads the tenth control information read to the value stored in the first area. Add the sizes contained in.

  When the size is 0 or the recording operation state information is in processing, the completion position information updating unit 106 includes control information including the following four pieces of information (hereinafter referred to as “the 11th control information”): Create).

· Reachable information included in the tenth control information,
An incomplete state representing a state before the process is completed,
Recording operation state information included in the tenth control information
Control information including size information included in the tenth control information.

  More specifically, the completion position information updating unit 106 reads the control information recorded in the storage area buffer [lsn] indicated by the value stored in the variable “lsn” (step S223). The complete position information updating unit 106 stores the tenth control information included in the read control information in the variable “oldv” (step S224). The delivered position information updating unit 105 executes at least one of the determination process 1 and the determination process 2 described below (step S225).

Determination process 1: Whether or not the size stored in the variable “length” included in the tenth control information stored in the variable “oldv” is 0,
Determination process 2: Whether or not the variable “inCopy” included in the tenth control information stored in the variable “oldv” is in the processing state.

  In the case where the size stored in variable “length” is not 0 and the value stored in variable “inCopy” is in the non-processing state (at step S 225) (NO), the value stored in the variable “lsn” and the value stored in the variable “length” included in the variable “oldv” are added (step S226). The delivered position information updating unit 105 stores the calculated result in the variable “lsn”. Thereafter, the reachable position information updating unit 105 executes the process shown in step S222.

  Note that if the size stored in the variable “length” is 0, or the value stored in the variable “inCopy” is in the processing state (step S225). YES), the tenth control information stored in the variable "oldv" is stored in the variable "newv" (step S227). The delivered position information updating unit 105 stores 1 (that is, an uncompleted state) in the variable “cflag” included in the variable “newv” (step S228). That is, the variable "newv" stores eleventh control information.

  Next, the completion position information updating unit 106 stores the created eleventh control information in the second area in the log buffer 112 by executing the same process as the process executed by the “cas” instruction. Record in the storage area pointed to by

  The process of recording the eleventh control information will be described in detail. The completion position information updating unit 106 reads control information (hereinafter referred to as “12th control information”) stored in the values stored in the second area in the log buffer 112. If the read twelfth control information matches the tenth control information, the completion position information updating unit 106 determines that the value stored in the second area in the log buffer 112 is the prepared eleventh control information. Record in the indicated storage area. If the read twelfth control information and the tenth control information do not match, the completion position information updating unit 106 sets the twelfth control information to the tenth control information, and then, the above-described determination on the tenth control information Execute the process

  The completion position information updating unit 106 atomically executes the same processing as the processing executed by the “cas” instruction as described above. More specifically, a process of executing the same process as the process executed by the “cas” instruction will be described.

  The completion position information updating unit 106 reads the twelfth control information stored in the storage area buffer [lsn]. The completed position information updating unit 106 stores the eleventh control information stored in the variable “newv” when the read twelfth control information matches the tenth control information (YES in step S229). Record in buffer [lsn] (step S231). If the read twelfth control information and the tenth control information do not match (NO in step S 229), the completion position information updating unit 106 sets the twelfth control information stored in the storage area buffer [lsn] to The variable "oldv" is stored (step S230). Thereafter, the completion position information updating unit 106 executes the process shown in step S225.

  The completion position information updating unit 106 atomically executes the same processing as the processing executed by the “cas” instruction as described above.

  That is, in the log information set included in the log buffer 112, the completion position information updating unit 106 performs the process shown in FIG. Find the log information set that is That is, among the log information sets whose identifiers have a larger value than the log information set being processed, the reachable position information updating unit 105 is the log information set being processed or the size is still 0. Search log information set. The delivered position information updating unit 105 sets the uncompleted state (or 1) in the control information included in the found log information set. That is, among the log information sets whose identifiers have a larger value than the log information set being processed, the reachable position information updating unit 105 is still in the processing state or the size is 0 in the log information set. Set to an uncompleted state (ie, representing that it is before completion).

  Next, the effects of the database management apparatus 101 according to the first embodiment will be described.

  According to the database management apparatus 101 according to the present embodiment, it is possible to efficiently set, in a storage area, control information that enables processing of a processing request to the log buffer 112 in a short time. The reason is that the log processing unit 102 does not have to execute the operation of setting the size in the log buffer 112 or the like while the process is being executed within the period of having the exclusive control right.

  The reason for achieving the effects described above will be described in more detail.

  In the operation of allocating the log buffer 112 by the database management apparatus 101 according to the present embodiment, the operation to be executed during the period in which the acquired exclusive control right is acquired reads the variable "TAIL" related to the log buffer 112, and then It only updates.

  On the other hand, in addition to the processing related to the variable "TAIL", the control information in the log buffer is an operation that needs to be executed during the period when the device described in Non-Patent Document 1 has the acquired exclusive control right. Processing for setting size information related to the size, and processing for setting recording operation state information.

  Furthermore, after recording log information in the log buffer, the device described in Non-Patent Document 1 securely stores data in a database by completing processing related to a transaction. Therefore, in the device, particularly when the information system is a system (many-core system) equipped with a large number of cores, contention occurs as the plurality of processes execute the operation of securing the log buffer simultaneously. As a result, the processing performance of the many core system is degraded.

  Therefore, comparing the database management apparatus 101 with the apparatus described in Non-Patent Document 1, the process executed by the database management apparatus 101 with respect to the log buffer 112 in the period in which the exclusive control right is held. It is less by the amount of processing. That is, since the amount of processing to be executed during the period of having the acquired exclusive control right is small, according to the database management apparatus 101 according to the present embodiment, for example, contention on the exclusive control is reduced in a many core system. can do.

Further, the apparatus disclosed in Non-Patent Document 2 has a problem that the control regarding the log buffer in the system is complicated as compared with the database management apparatus 101 according to the present embodiment. The reasons for this are, for example, the following two. That is,
○ When there are multiple log buffers 112 to be managed, each process needs to select the write destination log buffer 112.
○ When there are multiple targets (that is, the log buffer 112) for determining whether or not the recording process for the log information is completed, in the data recovery operation after the failure occurrence, the log information written to each log buffer 112 It is necessary to recognize the order and read the log information in the order.

  Therefore, according to the database management apparatus 101 according to the present embodiment, it is not necessary to execute complicated control.

  For example, the log buffer 112 can also be implemented using non-volatile memory, with log information representing change history for the database. In this case, according to the database management apparatus 101 according to the present embodiment, it is not necessary to record the change history in a device such as a disk or an SSD requiring an input / output operation while synchronizing with the occurrence of the change history. . Further, according to the database management apparatus 101 according to the present embodiment, it is possible to execute recording of log information at high speed, particularly in a many core system with a large number of cores.

  According to the recording operation completion information in the database management apparatus 101 according to the present embodiment, it can be determined in a short time whether or not the processing related to the log information set in the log buffer 112 is before completion. The reason is that the control information includes recording operation completion information indicating whether or not the process is completed. Since the control information has the recording operation completion information, the database management apparatus 101 refers to the recording operation completion information in a certain log information set to determine whether or not the input / output processing has been completed for the certain log information set. It can be determined between.

  According to the reachable position information in the database management apparatus 101 according to the present embodiment, it is possible to know in the log buffer 112 which log information set up to which processing has been started in a short time. The reason for this is that the reachable position information updating unit 105 can specify the position of the log information set located in the rearmost order among the log information sets in the reachable state.

  Further, according to the reachable position information in the database management apparatus 101 according to the present embodiment, it is possible to securely execute all the process of recording the log information set. The reason is that, based on the position specified by the reachable position information updating unit 105, the waiting process is executed until the process on all the log information sets is started. That is, the log processing unit 102 executes the synchronization processing at the timing at which the processing related to the log information set is all started. Therefore, according to the database management apparatus 101 according to the present embodiment, the next process can be performed earlier than when the synchronous process is executed at the timing when all the processes related to the log information set are completed. .

  Further, according to the completion position information in the database management apparatus 101 according to the present embodiment, it is possible to know in a short time whether or not the log information set in the log buffer 112 is in a state before the processing is completed. it can. The reason is that the completion position information updating unit 106 specifies the position where the control information whose size is 0 stored in the variable “length” is recorded.

  Also, by realizing the log buffer information storage unit 111 with the memory 108 that can be referred to by a plurality of transaction execution units, processing can be executed more efficiently. The reason is that a plurality of transaction executors can refer to the locations stored in the memory 108. A plurality of transaction executors can refer to the position without performing communication processing with each other. Furthermore, according to the database management apparatus 101 according to the present embodiment, as the processing performance of the memory 108 is higher, the position can be referred to in a shorter time.

Second Embodiment
The configuration of the control information setting device 151 according to the second embodiment of the present invention will be described in detail with reference to FIG. FIG. 16 is a block diagram showing a configuration of the control information setting device 151 according to the second embodiment of the present invention.

  The control information setting device 151 according to the second embodiment includes a deliverable information setting unit 152 and a log buffer 153.

  As described above with reference to FIG. 2, the log buffer 153 has a logically logical space for storing log information sets in which log information and control information to be referred to when executing processing relating to the log information are associated. In order (ie, logically ordered). The control information includes reachable information indicating whether or not it is possible to refer to the log information set located in a later order, and size information indicating the size (size, size) of the log information. Do.

  For convenience of explanation, in the log buffer 153, an area to which processing related to log information has been allocated (hereinafter referred to as "first area") and an area to which processing related to log information is not allocated (hereinafter referred to as "second area" It is assumed that it is separated from

  Next, processing in the control information setting device 151 according to the second embodiment will be described with reference to FIG. FIG. 17 is a flowchart showing the flow of processing in the control information setting device 151 according to the second embodiment.

  The reachable information setting unit 152 inputs the boundary value between the first area and the second area, for example, the size of the log information output by the external device, and the log information. Deliverable information setting unit 152 controls control information (hereinafter referred to as “first control”) recorded in storage area pointed to by the input boundary value (hereinafter referred to as “first storage area”) in log buffer 153. Information) is read (step S241). The delivery information setting unit 152 sets, in the control information recorded in the first storage area, an unreachable state indicating that it is impossible to refer to the log information set located in the later order (Step S242). The delivered information setting unit 152 sets the input size to the size information for the control information recorded in the first storage area (step S243).

  The delivery information setting unit 152 determines whether the delivery information included in the read first control information is in the delivery state (step S244). When the reachable information is in the reachable state (YES in step S244), the reachable information setting unit 152 causes the log buffer 153 to have the size information included in the control information in the order regarding the log information set. The second control information satisfying the predetermined condition is specified (step S245). For example, the predetermined condition is whether the size information is 0 or not. For example, the reachable information setting unit 152 specifies the second control information from the control information following the first control information in the process shown in step S245. In the storage area having the logical order, the position where the control information following the first control information is recorded is, for example, the position stored in the variable “nextLSN” in the first embodiment.

  The delivery information setting unit 152 sets the delivery state to the delivery information included in the identified second control information (step S246).

  On the other hand, if the reachable information included in the read first control information is in the unreachable state (NO in step S244), the reachable information setting unit 152 proceeds to step S245 and step S246. Do not execute the process shown.

  The boundary value is, for example, head position information stored in the log buffer information storage unit 111 (FIG. 4) described in the first embodiment. The delivery information setting unit 152 can be realized, for example, by the functions of the above-described log information storage unit 104 (FIG. 1) and the like.

  Next, the effects of the control information setting device 151 according to the second embodiment will be described.

  The control information setting device 151 according to the present embodiment can efficiently set, in the storage area, control information that enables processing of the processing request to the log buffer 153 in a short period of time. The reason for this is that the process of setting the size in the log buffer 153 can be realized using reachable information during a period in which the log buffer 153 does not have exclusive control right.

  On the other hand, for example, the device disclosed in Non-Patent Document 1 has size information in control information in a log buffer shared by a plurality of transaction executors with exclusive control over the log buffer. Set within a certain period. As a result, according to the control information setting device 151 of the second embodiment, the processing amount to be processed within the period of having the exclusive control right is reduced. Therefore, according to the control information setting device 151 according to the present embodiment, it is possible to set control information that makes it possible to process a processing request in a short period of time.

(Hardware configuration example)
A configuration example of hardware resources for realizing the database management device in each embodiment of the present invention described above using one calculation processing device (information processing device, computer) will be described. However, such a database management device may be realized physically or functionally using at least two calculation processing devices. Further, the database management device may be realized as a dedicated device.

  FIG. 18 is a view schematically showing an example of the hardware configuration of a calculation processing apparatus capable of realizing the database management apparatus according to the first embodiment and the control information setting apparatus according to the second embodiment. The calculation processing unit 20 includes a central processing unit (Central_Processing_Unit, hereinafter referred to as “CPU”) 21, a memory 22, a disk 23, a non-volatile recording medium 24, and a communication interface (hereinafter referred to as “communication IF”) 27 And a display 28. The calculation processing device 20 may further include an input device 25 and an output device 26. The calculation processing device 20 can transmit and receive information to and from other calculation processing devices and communication devices via the communication IF 27.

  The non-volatile recording medium 24 is, for example, a compact disc (Compact_Disc) or a digital versatile disc (Digital_Versatile_Disc) that can be read by a computer. In addition, the non-volatile storage medium 24 may be a universal serial bus memory (USB memory), a solid state drive (Solid_State_Drive) or the like. The non-volatile recording medium 24 holds such a program and enables portability without supplying power. The non-volatile recording medium 24 is not limited to the medium described above. Also, instead of the non-volatile recording medium 24, the program may be carried via the communication IF 27 and the communication network.

  That is, the CPU 21 copies a software program (computer program: hereinafter simply referred to as “program”) stored in the disk 23 to the memory 22 when executing it, and executes arithmetic processing. The CPU 21 reads from the memory 22 data required for program execution. When the display is required, the CPU 21 displays the output result on the display 28. When an output to the outside is required, the CPU 21 outputs an output result to the output device 26. When the program is input from the outside, the CPU 21 reads the program from the input device 25. The CPU 21 controls the database management program (FIG. 5, FIG. 6, FIG. 8 to FIG. 15) in the memory 22 corresponding to the function (process) represented by FIG. 1 or FIG. Interpret and execute the information setting program (FIG. 17). The CPU 21 sequentially executes the processing described in each embodiment of the present invention described above.

  That is, in such a case, it can be understood that the present invention can also be realized by the control information setting program or the database management program. Furthermore, it can be understood that the present invention can also be realized by a computer readable non-volatile recording medium in which such a database management program or such a control information setting program is recorded.

  The present invention has been described above by taking the above-described embodiment as an exemplary example. However, the present invention is not limited to the embodiments described above. That is, the present invention can apply various aspects that can be understood by those skilled in the art within the scope of the present invention.

DESCRIPTION OF SYMBOLS 101 database management apparatus 102 log processing part 103 head position calculation part 104 log information storage part 105 reachable position information update part 106 completion position information update part 107 transaction execution part 108 memory 109 transaction execution part 110 transaction execution part 111 log buffer information storage Part 112 Log buffer 151 Control information setting device 152 Reachable information setting part 153 Log buffer 20 Calculation processing device 21 CPU
22 Memory 23 Disk 24 Nonvolatile Storage Medium 25 Input Device 26 Output Device 27 Communication IF
28 Display

Claims (9)

Whether the log information and the reachable state representing the state where it is possible to refer to the log information that follows the log information in a predetermined order, or the unreachable state representing the state where it is impossible to refer A log buffer in which storage areas that can be stored are logically arranged in the above order in a state in which they are associated with reachable information representing H, and control information including size information representing the size of the log information;
In the log buffer, the first control information is recorded from a first control information recorded in a storage area representing a boundary between a storage area to which the recording process for recording the log information has been allocated and a storage area to which the recording process is not allocated. The delivery information is read, the unreachable state is set to the reachable information included in the first control information, and a size is set to the size information included in the first control information, When the reachable information is in the reachable state, the second control information specifying that the size information satisfies the predetermined condition is specified according to the order in the log buffer, and the specified second control information is specified. And a deliverable information setting unit configured to set the included deliverable information to the deliverable state. Further, the control information is in the processing state representing a state in which the process of storing the log information is being executed, or in the non-processing state representing a state in which the process of storing the log information is not being performed. Contains recording operation status information indicating
The processing in progress is set in the recording operation state information included in the first control information, and the process of recording the log information related to the first control information in the log buffer is completed. The control information setting device according to claim 1, further comprising: log information storage means for setting the non-processing state to the recording operation state information included in the first control information. The control information further indicates whether the process of storing the log information is in an uncompleted state representing a state before completion, or in a completed state representing a state in which the process of storing the log information is completed. Includes recording operation complete information,
The log information storage means, the log information of the certain size, after recording the first storage area in which the boundary representing read by the reachability information setting means, before Symbol before type recording operation completion information from the second control information Is read, and when the read recording operation completion information is in the uncompleted state, the size information satisfies the predetermined condition in the log buffer according to the order in the log buffer, or the recording operation state information The third control information specifying that the current processing state is being specified is specified, and the uncompleted state is set in the recording operation completion information included in the specified third control information. Control information setting device. The log buffer further includes reachable position information updating means for specifying a position where the second control information is recorded.
The control information setting device according to claim 3, wherein the log information storage means executes a queuing process until the position becomes equal to or more than a value obtained by adding the boundary and the certain size. 5. The control information setting device according to claim 3, further comprising: a completion position update unit that specifies a position at which the third control information is recorded in the log buffer. It has a memory that can be input to and output from the plurality of log information storage means.
The control information setting device according to claim 4, wherein the reachable position information updating means stores the specified position in the memory. The control information setting device according to any one of claims 1 to 6, wherein the log buffer is a storage area in a non-volatile recording medium. The information processing apparatus is in a reachable state representing the log information and the state in which the log information following the log information can be referred to in a predetermined order, or the state in which the information can not be referred to With regard to the log buffer in which storage areas storable in a state in which they are associated with the reachable information and the control information including the size information indicating the size of the log information are logically arranged in the order ,
In the log buffer, the first control information is recorded from a first control information recorded in a storage area representing a boundary between a storage area to which the recording process for recording the log information has been allocated and a storage area to which the recording process is not allocated. The delivery information is read, the unreachable state is set to the reachable information included in the first control information, and a size is set to the size information included in the first control information, When the reachable information is in the reachable state, the second control information specifying that the size information satisfies the predetermined condition is specified according to the order in the log buffer, and the specified second control information is specified. A control information setting method for setting the included reachable information to the reachable state. Whether the log information and the reachable state representing the state where it is possible to refer to the log information that follows the log information in a predetermined order, or the unreachable state representing the state where it is impossible to refer With regard to a log buffer in which storage areas that can be stored are logically arranged in the above-described order in association with reachable information representing the control information and control information including size information representing the size of the log information.
In the log buffer, the first control information is recorded from a first control information recorded in a storage area representing a boundary between a storage area to which the recording process for recording the log information has been allocated and a storage area to which the recording process is not allocated. The delivery information is read, the unreachable state is set to the reachable information included in the first control information, and a size is set to the size information included in the first control information, When the reachable information is in the reachable state, the second control information specifying that the size information satisfies the predetermined condition is specified according to the order in the log buffer, and the specified second control information is specified. A database management program that causes a computer to realize a deliverable information setting function that sets the included deliverable information to the reachable state.

Images