KR20180082753A - Database management system and method thereof using a non-volatile memory - Google Patents

Abstract

Provided is a database management system capable of storing a first record in a free-space depending on a transaction during running and updating a slot header with atomicity when the transaction is committed. The database management system may include a processor for executing the transaction and a hardware transaction memory for storing the first record associated with the writing in a free space of a first slotted page when writing is executed with respect to the first slotted page based on the transaction. In addition, when the commit of the transaction is determined by the processor, the hardware transaction memory may update information on the first record in a slot header of the first slotted page.

Description

[0001] DATABASE MANAGEMENT SYSTEM AND METHOD THEREOF USING A NON-VOLATILE MEMORY [0002]

The present invention relates to a database management system and method, and more particularly to a database management system that utilizes non-volatile memory.

Nonvolatile memory is a non-volatile memory that can read and write data byte-wise like DRAM (Dynamic Random Access Memory), and has the speed of reading and writing as fast as DRAM, It is a device that is attracting much attention as a next generation memory because it has non-volatility. When non-volatile memory is used as the main memory, many applications, including databases requiring data recovery and integrity, use non-volatile memory with relatively high processing speed, ) And flash memory can be substituted, and the performance can be expected to be improved.

However, there is a need to permanently store certain data while the application is running, but there may be data that is not. Also, there is a feature that the writing order is not guaranteed in the operation of writing data to the main memory. Today, a central processing unit (CPU) sometimes adjusts the order in which data is written to improve performance. Therefore, even if data is written, there may be a case where data written in only the cache in the central processing unit remains and the data is not updated in the main memory.

Conventionally, in order to prevent rearrangement of a write operation when a non-volatile memory is used as a main memory, there is a need to additionally perform a memory barrier operation such as mfence and sfence, There is a need to additionally perform an operation such as a cache line flush to drop data from the cache of the nonvolatile main memory to the nonvolatile main memory. These operations are faster than the fsync () operation that writes data to the HDD or flash memory, but the overhead is not negligible, so there is a possibility that the overall system performance may be degraded if it is used heavily.

According to one aspect, a database management system is provided that stores a first record in a free space according to a transaction in execution, and atomically updates a slot header when the transaction is committed. Wherein the database management apparatus comprises: a processor for executing a transaction; and when a write is performed on a first slotted page according to the transaction, a first record associated with the write is written to the first slotted page In a free space of the hardware transaction memory. In addition, when a commit of the transaction is determined by the processor, the hardware transactional memory may update information about the first record in the slot header of the first slotted page.

According to one embodiment, the hardware transactional memory stores the start offset of the free space updated in accordance with the first record, the number of records stored in the first slotted page, the end offset of the free space, and the first record And update at least one of the record offset arrays containing the information with respect to the first record.

According to another embodiment, the hardware transaction memory is implemented with a Restricted Transactional Memory (RTM), and the size of the slot header is fixed to 64 bytes.

According to another embodiment, the processor makes the fragmented free space into one continuous space when it is determined that the space for writing of the first record is insufficient, and the hardware transaction memory stores the fragmentation free space The first record may be stored in a defragmentation free space according to a copy on write technique.

According to yet another embodiment, the hardware transactional memory deletes the offset of the second record associated with the deletion from the slot header and stores it in the log when the deletion on the first slotted page is executed according to the transaction And if the commit of the transaction is determined by the processor, the hardware transactional memory may update information on the second record in the slot header.

According to another embodiment, the hardware transactional memory connects the free space associated with the second record as a linked list, and updates the start offset of the free space, And update at least one of the number of records stored in the one slotted page, the end offset of the free space, and the record offset array as information on the second record.

According to yet another embodiment, the hardware transactional memory further comprises means for, when an update on the first slotted page is executed in accordance with the transaction, adding a third record associated with the update, If a commit is determined, the hardware transactional memory may update information on the third record in the slot header.

According to another aspect, when a write on a first slotted page is executed in accordance with a transaction being executed, storing a first record associated with the write in a free space of the first slotted page in the buffer cache, Determining a commit for a transaction and updating the information on the first record in a slot header of the first slotted page if a commit of the transaction is determined, / RTI >

According to one embodiment, updating the information about the first record in the slot header of the first slotted page further comprises: updating the start offset of the free space updated in accordance with the first record, Updating the at least one of the number of records stored in the first record, the end offset of the free space, and the record offset array including the first record as information on the first record.

According to another aspect of the present invention there is provided a method of processing a plurality of slotted pages in a buffer cache for storing a plurality of records associated with the writes in respective slotted pages, cache memory, and a non-volatile memory including a write ahead log area for storing offsets containing information about the plurality of records.

According to one embodiment, the non-volatile memory stores each slot header corresponding to each slotted page in the first write log area according to the stored plurality of records, Lt; RTI ID = 0.0 > of < / RTI >

According to another embodiment, when a commit of the transaction is determined by the processor, the non-volatile memory updates the respective slot headers stored in the first write log area to the corresponding respective slotted pages .

According to another aspect, there is provided a method comprising: updating a plurality of records associated with a write to each slotted page stored in a buffer cache, when a write on a plurality of slotted pages is executed according to a transaction being executed; And storing the offsets containing information about the records in a pre-designated first write log area as a slot header.

According to an embodiment of the present invention, the step of storing the offsets in a pre-designated first write log area as a slot header may further include storing each slot header corresponding to each slotted page according to the stored plurality of records, And wherein each of the slot headers may include an offset relative to each of the slotted pages.

1 is an exemplary diagram illustrating a structure of a slotted page used by a database management system according to an embodiment.
2 is a flowchart of a buffer caching method performed by the database management system according to an embodiment.
FIGS. 3A and 3B are diagrams illustrating a process of performing a write operation on a slotted page according to the buffer caching method illustrated in FIG. 2. FIG.
4 is a flowchart of a buffer caching method performed by the database management system according to another embodiment.
5 is a flowchart of a buffer caching method performed by the database management system according to another embodiment.
6A is a block diagram illustrating a database management system according to another embodiment.
FIG. 6B is an exemplary diagram showing a logging method performed by the database management system illustrated in FIG. 6A.
7 is an exemplary diagram illustrating a logging method performed by the database management system according to another embodiment.
FIG. 8 is an illustration of a defragmentation performed by the database management system according to another embodiment.

Specific structural or functional descriptions of embodiments are set forth for illustration purposes only and may be embodied with various changes and modifications. Accordingly, the embodiments are not intended to be limited to the specific forms disclosed, and the scope of the disclosure includes changes, equivalents, or alternatives included in the technical idea.

The terms first or second, etc. may be used to describe various elements, but such terms should be interpreted solely for the purpose of distinguishing one element from another. For example, the first component may be referred to as a second component, and similarly, the second component may also be referred to as a first component.

It is to be understood that when an element is referred to as being "connected" to another element, it may be directly connected or connected to the other element, although other elements may be present in between.

The singular expressions include plural expressions unless the context clearly dictates otherwise. In this specification, the terms "comprises ", or" having ", and the like, are used to specify one or more of the features, numbers, steps, operations, elements, But do not preclude the presence or addition of steps, operations, elements, parts, or combinations thereof.

Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art. Terms such as those defined in commonly used dictionaries are to be interpreted as having a meaning consistent with the meaning of the context in the relevant art and, unless explicitly defined herein, are to be interpreted as ideal or overly formal Do not.

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS In the following description of the present invention with reference to the accompanying drawings, the same components are denoted by the same reference numerals regardless of the reference numerals, and a duplicate description thereof will be omitted.

1 is an exemplary diagram illustrating a structure of a slotted page used by a database management system according to an embodiment. Referring to FIG. 1, there is shown the structure of a slotted page that a database management system uses to perform buffer caching or logging. The slotted page may include a slot header 110, a free space 120, and a record content area 130. The slot header 110 may include metadata associated with the slotted page and information regarding stored records. In addition, the record content area 130 may include a key for each record and for each record. In addition, the free space 120 may indicate an empty space disposed between the slot header 110 and the record content area 130. [ Illustratively, if a new record is inserting into the page, the database management system may write the record sequentially from the end of the slotted page, regardless of the size of the key value. Also, the order according to the key of each record may be stored in the slot header 110 as a record offset array 116. [ More specifically, the record offset array 116 may be maintained in an aligned state according to the magnitude of the key value. In this embodiment, an embodiment in which the size of one slotted page is 1 kilobyte (byte) is shown, but this should not be construed as limiting or limiting the scope of other embodiments.

The slotted page may include a flag 110 indicating whether the current page is an internal node or a leaf node in the slot header 110 as one byte. In addition, the slotted page may store the start offset 112 of the free space 120 indicating the page area that is not currently used in the slot header 110. 14 can be stored in the start offset 112 since 14 bytes of free space 120 start in the slotted page described in FIG. In addition, the slotted page may store the number 113 of records stored in the current slotted page at the next two bytes following the start offset 112. Since three records are stored in this embodiment, 3 can be stored. The slotted page may also store the end offset 114 of the free space 120 in the next two bytes following the number of records 113. [ In this embodiment, since the last area of the free space 120 is 972, 972 is stored in the end offset 114. Fragmentation information for connecting the free space 120 may be stored in the last 1 byte of the slot header 110 when the free space 120 is separated.

In the embodiment illustrated in FIG. 1, three records 132, 134, 136 are stored within the slotted page. The key values of the respective records 132, 134, and 136 are 20, 10, and 30, and the number of bytes occupied by each of the records 132, 134, and 136 is 18, 14, and 14 bytes. These records may be stored in the record content area 130 of the slotted page. As shown in FIG. 1, the data in the record content area 130 will not be sorted according to the key value. For example, a record 132 having a key value of 20 greater than 10 may be placed at the front.

Therefore, the database management system needs to sort the records according to the key value. In the slotted page, the start byte of each record is sorted and stored in the record offset array 116 of the slot header 110 according to the key value . More specifically, since 10 is the smallest key value, the start offset 992 of the record with ten keys can be stored as the first element of the record offset array 116. Similarly, a start offset 972 of a record having a key of 20 may be stored in a second element of the record offset array 116, and a start offset 1008 of a record having a key of 30 may be stored as a third element. The slotted page according to the present embodiment does not sort the entire record but arranges and manages only the offset having a small size as 2 bytes in the separate record offset array 116 to change the order of the entire record, It is possible to expect the effect of increasing the overall system performance by reducing the size.

2 is a flowchart of a buffer caching method performed by the database management system according to an embodiment. Referring to FIG. 2, a buffer caching method is a method in which, when a write operation is performed on a first slotted page according to a transaction to be executed, a first record associated with the write is stored in a free space of the first slotted page in the buffer cache (220) of determining a commit for the transaction, and if the commit of the transaction is determined, transmitting information about the first record to a slot header of the first slotted page (Step < RTI ID = 0.0 > 230). ≪ / RTI > The database management system described in this embodiment can be implemented through a processor executing a transaction and a hardware transactional memory for storing and updating data associated with the transaction.

In step 210, the processor of the database management system may execute a predefined transaction. In the following description, a transaction represents a set of operations of a database necessary to perform one operation, and may represent recovery, storage, update, deletion, etc. of data as a unit of logical operation in the database. Also, if a write on the first slotted page is executed in accordance with the transaction being executed in step 210, the database management system may store the first record associated with the write in the free space of the first slotted page have. More specifically, the hardware transaction memory of the database management system can find a free space by referring to the slot header when writing is performed according to the transaction. The hardware transactional memory may also store the first record sequentially from the last byte of the free space of the first slotted page. The stored record through the database management system of this embodiment is not visible by other transactions until the slot header is updated. Since processes running in the database seek valid records through the slot headers before reading the records, the first record written to the free space can provide the effect of not corrupting existing data.

In another embodiment, there may be a case where there is not enough space in the free space to store the first record in step 210. [ In this case, the processor of the database management system can make the fragmented free space into one continuous space when it is determined that the space of the first record is insufficient. In addition, the hardware transactional memory may store the first record in a defragmentation free space according to a copy on write technique.

In step 220, the processor may determine a commit for a transaction that was being executed. The commit can be performed when all the operations included in the transaction are executed and the update contents of the corresponding database are recorded in the work area to determine when the application of the transaction is determined to be completed. When a commit is executed, the update data is written to the database, the associated lock is released, and the updated contents can be accessed from another transaction.

If the commit of the transaction is determined in step 220, then in step 230, the hardware transaction memory of the database management system may update information about the first record in the slot header of the first slotted page . More specifically, the hardware transactional memory includes a start offset of the updated free space, a number of records stored in the first slotted page, an end offset of the free space, and the first record as the first record is newly stored And update at least one of the record offset arrays in the slot header as information regarding the first record.

The database management system of this embodiment can implement a hardware transactional memory using Restricted Transactional Memory (RTM). Hence, the hardware transactional memory can atomically update the slot header including the record offset array. If using RTM, the memory operation for the first record occurring in the transaction will not be referenced outside of the transaction. In addition, the modified cache line in the transaction can remain in the L1 cache without being flushed and can guarantee atomicity for the cache line unit. In one embodiment, in the buffer caching method according to the present embodiment, the length of the slot header can be fixed to 64 bytes, which is the size of one cache line, and the slot headers can be atomically updated through the RTM, It can be implemented to perform the same function. Therefore, in the buffer caching method of the present embodiment, even if a system crash occurs before the slot header is updated, the newly added record can not be seen in the header of the slot, so that the newly added record can be ignored, A guaranteed effect can be provided.

FIGS. 3A and 3B are diagrams illustrating a process of performing a write operation on a slotted page according to the buffer caching method illustrated in FIG. 2. FIG. Referring to FIG. 3A, a slotted page managed by the database management system is shown. The slotted page may include a first slot header 310, a first free space 320, and a first record content area 330. The database management system can execute a write operation on the slotted page in accordance with the transaction being executed. The database management system can refer to the first slot header 110 to confirm that the last byte of the current first free space 320 is 972 bytes. Illustratively, the database management system may write a fourth record with a key value of 40 and write the byte length 44 corresponding to the fourth record to 926 bytes of free space. Accordingly, the last byte 321 of the newly updated first free space 320 may be 926 bytes. The database management system of this embodiment may not perform the update for the first slot header 310 during the execution of the transaction. Accordingly, even if a system error occurs before the transaction is committed, the newly recorded record does not affect other transactions, and the stability of the entire system can be expected to be improved.

3B is an exemplary diagram showing a process of determining completion of a transaction to be performed and updating a slot header of a slotted page by the database management system. The updated slotted page may include a second slot header 340, a second free space 350, and a second record content area 360.

The database management system can store 16 at the start offset 341 since the second free space 350 starts from 16 bytes in the newly updated slotted page. Also, since a new fourth record has been added in the slotted page in accordance with the transaction, the database management system can store the number 342 of records currently stored in the slotted page as four. Also, since the last area of the second free space 350 is 926, the database management system may store the end offset 343 of the second free space 350 as 926. [ In addition, the database management system may update the record offset 926 corresponding to the fourth record in the last area 344 in the record offset array. A fourth record is added so that the second free space 250 can be updated to represent an area from 16 bytes to 926 bytes. In addition, the second record content area 360 may be updated to include new data for the fourth record and represent an area from 926 bytes to 1023 bytes.

4 is a flowchart of a buffer caching method performed by the database management system according to another embodiment. Referring to FIG. 4, the buffer caching method includes the steps of deleting, from a slot header, an offset of a second record associated with the deletion when the deletion of the first slotted page is executed according to a transaction to be executed, 410), determining (420) a commit for the transaction, and updating (430) information about the second record in a slot header of the first slotted page if commit of the transaction is determined, . ≪ / RTI >

In step 410, the database management system can check the position of the second record corresponding to the deletion request by referring to the slot header included in the first slotted page. The database management system may delete the second record in the record content area according to the transaction being executed. In this embodiment, however, the database management system will not update the slot header during the transaction. More specifically, the database management system may delete the offset of the second record associated with the deletion from the slot header and store it in the log.

At step 420, the database management system may determine a commit for the transaction. The detailed description of the process of determining the transaction commit may be applied to the description of the step 220 described in conjunction with FIG. 2 as it is, and redundant description will be omitted.

If it is determined in step 420 that the transaction is to be committed, the database management device may update the information on the second record in the slot header of the first slotted page. More specifically, the database management device may use the RTM to atomically delete the offset associated with the second record within the record offset array in the slot header. The area in which the deleted record was written may need to be associated with a free space. In this case, the database management apparatus can link the free space associated with the deleted second record as a linked list. Also, the database management apparatus may update at least one of the number of records stored in the first slotted page, the end offset of the free space, and the record offset array as information on the second record in the slot header.

5 is a flowchart of a buffer caching method performed by the database management system according to another embodiment. Referring to FIG. 5, a buffer caching method may include adding 510 a third record associated with the update when an update on a first slotted page is executed according to a transaction being executed, (530) updating information about the third record in the slot header of the first slotted page if the commit of the transaction is determined (520).

In step 510, the database management system can check the position of the third record corresponding to the update request by referring to the slot header included in the first slotted page. The database management system may add a new third record having a new value to the free space when updating the third record in accordance with the transaction being executed. Likewise, in this embodiment, the database management system will not update the slot header during the transaction.

At step 520, the database management system may determine a commit for the transaction. The detailed description of the process of determining the transaction commit may be applied to the description of the step 220 described in conjunction with FIG. 2 as it is, and redundant description will be omitted.

If the commit of the transaction is determined in step 520, the database management device may update the information on the updated third record in the slot header of the first slotted page in step 530. More specifically, the hardware transactional memory of the database management system may overwrite the new offset of the third record in the record offset array. In addition, the database management system can atomically update a slot header having a size of a 64-byte cache line using hardware transactional memory.

6A is a block diagram illustrating a database management system according to another embodiment. Referring to FIG. 6A, the database management system 600 may include a non-volatile memory 610 and a processor 620. Processor 620 may perform a predefined transaction. A transaction executed by the processor 620 may include a write operation on a plurality of slotted pages.

Non-volatile memory 610 may operate as main memory within the database management system. Non-volatile memory 610 may include a buffer cache that stores a plurality of records associated with a write in each slotted page. The non-volatile memory 610 may also include a write ahead log area for storing offsets containing information regarding a plurality of records.

FIG. 6B is an exemplary diagram showing a logging method performed by the database management system illustrated in FIG. 6A. Referring to FIG. 6B, the database management system may include a non-volatile memory 610 and a processor 620, including a buffer cache 611 and a write log area 612 first. The database management system according to the present exemplary embodiment may include a method of copying only the slot headers related to the plurality of slotted pages into the write log area 612 in the case where a modification to a plurality of slotted pages occurs according to a transaction to be executed Can be performed. Accordingly, the database management system can copy only the offset pointing to the updated record to the log area without copying the entire page to be updated, minimizing the memory write operation by logging only a small 2-byte offset, Effect can be expected. The nonvolatile memory 610 of the present embodiment may be implemented in various forms of nonvolatile memory as well as RTM, and the length of the slot header is also not limited to 64 bytes.

Processor 620 may perform write operations on a plurality of slotted pages in accordance with the transaction in progress. The processor 620 can directly write new data of the first slotted page and the second slotted page to be modified through the nonvolatile memory 610. [ More specifically, the updated first slotted page and the updated second slotted page may be included in the buffer area 611 in the nonvolatile memory 610. However, the processor 620 may not update the respective slot headers included in the first slotted page and the second slotted page.

In addition, the processor 620 may first write the offsets pointing to the location of newly written records using the write log area 612 in the non-volatile memory 610 first. More specifically, the processor 620 first uses the write log area 612 to update the first slot header corresponding to the updated first slotted page and the second slot header corresponding to the updated second slotted page, respectively It can be stored in the write log area 612 first.

Processor 620 may determine the commit of the transaction that was being executed. If the commit of the transaction is determined, the processor 620 first copies each of the first slot header and the second slot header recorded in the write log area 612 to the corresponding slotted page, To be visible to the user.

The database management system according to the present embodiment can perform logging only on the slot header including the offset of the records without performing writing on the entire slotted page during the transaction to ensure the integrity of the database while minimizing the write operation Effect can be expected.

The buffer caching method described in FIG. 2 is available in a processor that supports RTM, and will also be used when a write is performed on one slotted page in accordance with a transaction. On the other hand, the logging method illustrated in FIG. 6B can be used when a B-tree page is divided due to a transaction or an overflow that updates a plurality of slotted pages. The database management apparatus of the present embodiment can determine whether an overflow occurs in the B-tree during the transaction. Also, the database management apparatus may perform a method of performing buffer caching on one slotted page or a method of performing update on a plurality of slotted pages and logging a slot header, depending on whether the overflow occurs or not You can choose.

7 is an exemplary diagram illustrating a logging method performed by the database management system according to another embodiment. Referring to FIG. 7, an exemplary diagram illustrating a logging method performed by the database management system when a B-tree page overflows. There may be a case where an overflow occurs in the third page according to the insertion request of the data having the key value 14 in step 710. [ In step 720, the database management device may generate as a CPU cache a new record offset array comprising offsets of records having a key value greater than the key value 14. In this embodiment, the offsets of records 700, 500, 800 each having key values 15, 17 and 19 can be generated as a new record offset array. Also, in step 720, the database management system may record a plurality of slot headers including the generated new record offset array in a separate slot header log. In this case, there will be no change in the existing third page.

At step 730, the database management system may allocate a new sibling page and copy records having a key that is less than or equal to the specified key value 14 to a new neighboring page. Likewise, there will be no change in the existing third page.

In step 740, the database management system may store the key values and points for the neighboring pages as a new record. More specifically, the neighboring page may be divided from the parent page of the third page in which an overflow occurs to indicate a newly created page. The database management system may record a record relating to the neighboring page in a free space and record a new slot header in which the offset of the record is added in the slot header log.

If the transaction is completed in step 750, the database management device may record a commit mark in the slot header log. In addition, the database management apparatus may perform checkpointing to reflect the transactions recorded in the log on the slotted page. While the checkpointing is being performed, the DBMS may replace the slot headers of the slotted pages with the slot headers recorded in the log area.

FIG. 8 is an illustration of a defragmentation performed by the database management system according to another embodiment. In a slotted page structure, there may be cases where data should be lost from a slotted page due to splitting or erasing operations. In this case, the database management system may delete the offset corresponding to the record in the record offset array to clear the record to be deleted. However, in the case where only the offset is deleted, the deleted record in the record content area can be left intact. Accordingly, in order to use the area for storing another record, the database management system can connect the areas as a free list and generate the free space as a free space.

Referring to FIG. 8, the third page shows a state in which an overflow has occurred and records corresponding to key value 11 and key value 13 have moved to a new neighboring page. Therefore, in the third page of FIG. 8, an area corresponding to 600 to 700 bytes and an area corresponding to 800 to 900 bytes can be connected to the free list.

In step 810, if the third page contains more than 200 bytes of data larger than the empty space, the database management apparatus will need to perform a de-fragmentation operation for collecting the two divided spaces. The database management apparatus of the present embodiment copies data in a new page in a new copying method in a manner that there is no empty space and overwrites the address of the page of the parent entry of the parent page with the address of the newly created page, Update can be performed.

The embodiments described above may be implemented in hardware components, software components, and / or a combination of hardware components and software components. For example, the devices, methods, and components described in the embodiments may be implemented within a computer system, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, such as an array, a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. The processing device may also access, store, manipulate, process, and generate data in response to execution of the software. For ease of understanding, the processing apparatus may be described as being used singly, but those skilled in the art will recognize that the processing apparatus may have a plurality of processing elements and / As shown in FIG. For example, the processing unit may comprise a plurality of processors or one processor and one controller. Other processing configurations are also possible, such as a parallel processor.

The software may include a computer program, code, instructions, or a combination of one or more of the foregoing, and may be configured to configure the processing device to operate as desired or to process it collectively or collectively Device can be commanded. The software and / or data may be in the form of any type of machine, component, physical device, virtual equipment, computer storage media, or device , Or may be permanently or temporarily embodied in a transmitted signal wave. The software may be distributed over a networked computer system and stored or executed in a distributed manner. The software and data may be stored on one or more computer readable recording media.

The method according to an embodiment may be implemented in the form of a program command that can be executed through various computer means and recorded in a computer-readable medium. The computer readable medium may include program instructions, data files, data structures, and the like, alone or in combination. Program instructions to be recorded on a computer-readable medium may be those specially designed and constructed for an embodiment or may be available to those skilled in the art of computer software. Examples of computer-readable media include magnetic media such as hard disks, floppy disks and magnetic tape; optical media such as CD-ROMs and DVDs; magnetic media such as floppy disks; Magneto-optical media, and hardware devices specifically configured to store and execute program instructions such as ROM, RAM, flash memory, and the like. Examples of program instructions include machine language code such as those produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter or the like. The hardware devices described above may be configured to operate as one or more software modules to perform the operations of the embodiments, and vice versa.

Although the embodiments have been described with reference to the drawings, various technical modifications and variations may be applied to those skilled in the art. For example, it is to be understood that the techniques described may be performed in a different order than the described methods, and / or that components of the described systems, structures, devices, circuits, Lt; / RTI > or equivalents, even if it is replaced or replaced.

Claims (14)

A processor executing a transaction; And
A hardware transactional memory that stores a first record associated with the write in a free space of the first slotted page when a write on a first slotted page is performed in accordance with the transaction,
Lt; / RTI >
Wherein the hardware transactional memory updates the information about the first record in a slot header of the first slotted page when a commit of the transaction is determined by the processor.
The method according to claim 1,
Wherein the hardware transactional memory is operative to determine a record offset array comprising a start offset of the free space updated according to the first record, a number of records stored in the first slotted page, an end offset of the free space, And updates at least one as information on the first record.
The method according to claim 1,
Wherein the hardware transaction memory is implemented with a Restricted Transactional Memory (RTM), and the size of the slot header is fixed to 64 bytes.
The method according to claim 1,
Wherein the processor makes the fragmented free space into one continuous space when it is determined that the space for writing the first record is insufficient, and the hardware transaction memory stores the free space in the defragmentation free space A database management system for storing a first record according to a copy on write technique.
The method according to claim 1,
Wherein the hardware transactional memory deletes the offset of the second record associated with the deletion from the slot header and stores the offset in the slot header when deletion is performed on the first slotted page according to the transaction,
Wherein the hardware transactional memory updates the information regarding the second record in the slot header when a commit of the transaction is determined by the processor.
6. The method of claim 5,
Wherein the hardware transaction memory links the free space associated with the second record as a linked list and updates a start offset of the updated free space according to the second record, a start offset of the record stored in the first slotted page, The end offset of the free space, and the record offset array as information on the second record.
The method according to claim 1,
Wherein the hardware transactional memory adds a third record associated with the update when an update on the first slotted page is executed in accordance with the transaction,
Wherein the hardware transactional memory updates information on the third record in the slot header when a commit of the transaction is determined by the processor.
Storing a first record associated with the write in a free space of the first slotted page in the buffer cache when a write on the first slotted page is executed in accordance with a transaction being executed;
Determining a commit for the transaction; And
Updating the information on the first record in the slot header of the first slotted page if the commit of the transaction is determined;
/ RTI >
9. The method of claim 8,
The step of updating the information on the first record in the slot header of the first slotted page comprises:
At least one of a start offset of the free space updated in accordance with the first record, a number of records stored in the first slotted page, an end offset of the free space, and a record offset array including the first record, Lt; RTI ID = 0.0 > 1 < / RTI > record.
A processor executing a transaction; And
And a buffer cache that stores a plurality of records associated with the write in each slotted page when a write is performed on a plurality of slotted pages in accordance with the transaction, A non-volatile memory including a write ahead log area for storing offsets containing information,
And a database management system.
11. The method of claim 10,
Wherein the non-volatile memory stores each slot header corresponding to each slotted page in the first write log area according to the stored plurality of records, A database management system that includes offsets.
12. The method of claim 11,
Wherein the non-volatile memory updates the respective slot headers stored in the first write log area to the corresponding respective slotted pages when a commit of the transaction is determined by the processor.
Updating a plurality of records associated with the write to each slotted page stored in a buffer cache when a write on a plurality of slotted pages is executed according to a transaction to be executed; And
Storing the offsets including information on the plurality of records in a pre-designated first write log area as a slot header
≪ / RTI >
14. The method of claim 13,
Storing the offsets in a pre-designated first write log area as a slot header,
Storing each slot header corresponding to each slotted page in the first write log area according to the stored plurality of records
Wherein each of the slot headers comprises an offset for each of the slotted pages.

Images