KR20200098971A - Method and apparatus for storing data based on single-level - Google Patents

Abstract

The present invention relates to a single-level based data storage device and a method thereof for providing higher performance and efficiency for reading and writing data. According to an embodiment of the present invention, the method of the single-level based data storage device comprises the following steps of: storing data in a first memory table included in a nonvolatile memory; storing at least a part of the data stored in the first memory table in a second memory table included in the nonvolatile memory, when the capacity of the data stored in the first memory table is greater than or equal to a predetermined value; identifying the data stored in the second memory table and performing a flush operation; and storing the data stored in the second memory table in a single-level on a disk separate from the nonvolatile memory based on the flush operation.

Description

Single-level based data storage device and method {METHOD AND APPARATUS FOR STORING DATA BASED ON SINGLE-LEVEL}

The present invention relates to a data storage device and method for storing data in a single-level using a non-volatile memory.

In recent years, according to the demand for big data and data analysis, a system having an index function that is arranged based on a key value indicating the location of a file in which data is stored is required to find desired information in a database. The index function has an efficient structure such as a non-tree and an LSM tree so that included data can be easily searched.

The index function using a B-tree keeps the data sorted according to the key, and when additional data is inserted or deleted, the sorted data is updated and rewritten. In this method, when additional data is inserted or deleted, the contents of the target data area are modified and then re-written. Such data rewriting occurs in the form of a random write. In this case, it makes it difficult to exert the maximum performance of the SSD and increases the garbage collection load.

The index function using the LSM tree (log structured merge tree) sorts and maintains data according to the key, but when additional data is inserted or deleted, it is independently recorded as newly sorted data. In this method, when additional data is inserted or deleted, data for additional operations are sorted and stored separately. At this time, since input/output (I/O) is generated by sequential writing, maximum write performance is exhibited. However, as the number of independently sorted data increases, there is a problem in that performance is degraded because all separated data must be accessed when searching data.

Korean Patent Publication No. 10-2014-0070834 (published on June 11, 2014)

The problem to be solved by the present invention is to provide a data storage method and apparatus that provides higher performance and efficiency for reading and writing data.

However, the problems to be solved by the present invention are not limited as mentioned above, and are not mentioned, but include objects that can be clearly understood by those of ordinary skill in the art from the following description. can do.

A single-level based data storage method according to an embodiment of the present invention includes storing data in a first memory table included in a nonvolatile memory, and when the capacity of data stored in the first memory table is greater than or equal to a predetermined value. , Storing at least some of the data stored in the first memory table in a second memory table included in the nonvolatile memory, and performing a flush operation by identifying data stored in the second memory table; and And storing the data stored in the second memory table in a single-level on a disk separate from the nonvolatile memory based on the flush operation.

In addition, the storing of the single-level may include storing the data stored in the second memory table as a single-level based on the flush operation so that data is retrieved based on a B-tree data structure. It may include the step of.

Further, it may include performing compaction on at least a part of the data stored in the single-level.

In addition, the performing of the compaction may include deleting the data before the update from among the data stored in the single-level, and storing the remaining data stored in the single-level based on a key value included in the remaining data. It may include the step of aligning.

In addition, the performing of the compaction may include: identifying a ratio of data in which at least a portion of the data stored in the single-level is stored before the update time and the data after the update time, and the identified ratio is predetermined. If it is lower than the value, it may include selecting at least a portion of the data prior to the update point as a compaction candidate, and deleting at least a portion of the selected compaction candidate.

In addition, the performing of the compaction may include scanning a predetermined number of nodes among each node of the non-tree, and a data table having a predetermined number or more of data related to the scanned node among the data stored in the single-level. In the case of being distributed and stored in a file, the step of selecting the data table file as the compaction candidate may be included.

In addition, the performing of the compaction may include calculating the number of data table files required to read the data stored in the single-level, and the calculated number of data table files related to the largest data. It may include the step of selecting a data table file as the compaction candidate.

A single-level based data storage device according to an embodiment of the present invention includes a first memory table for storing input data, and when a capacity of data stored in the first memory table is greater than a predetermined value, the first memory A second memory table that stores at least some data stored in the table, a flush execution unit that performs a flush operation by identifying a state of the data stored in the second memory table, and the second memory table based on the flush operation. A single-level storage unit for storing at least some of the data stored in the memory table as a single-level on a disk separate from the nonvolatile memory, wherein the first memory table and the second memory table are nonvolatile Stored in memory.

Further, the single-level storage unit may store data stored in the second memory table in a single-level based on the flush operation so that data is retrieved based on a B-tree data structure.

In addition, it may include a compaction performing unit that performs compaction on at least a part of the data stored in the single-level.

In addition, the compaction performing unit deletes the data before the update from among the data stored in the single-level among the data stored in the single-level, and stores the remaining data in the single-level as a key value included in the remaining data. You can sort based on value).

In addition, the compaction performing unit may further include identifying a ratio of data that is stored in at least some of the single-level data stored before the update time and data after the update time, and wherein the identified ratio is lower than a predetermined value. In this case, it may include selecting at least a part of data before the predetermined time point as a compaction candidate, and deleting at least some of the selected compaction candidates.

In addition, the compaction performing unit scans a predetermined number of nodes among each node of the non-tree, and among the data stored in the single-level, data related to the scanned node is distributed and stored in a data table file of a predetermined number or more. In this case, the data table file may be selected as the compaction candidate.

In addition, the compaction execution unit calculates the number of data table files required to read the data stored in the single-level, and the data table file associated with the data with the largest number of the calculated data table files. Can be selected as a compaction candidate.

The single-level-based data storage device and method according to an embodiment of the present invention may provide higher performance and efficiency for reading and writing data.

However, the effects obtainable in the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those of ordinary skill in the art from the following description. I will be able to.

1 is a conceptual diagram illustrating a single-level data storage method according to an embodiment of the present invention.
2 shows an example of a functional configuration of a single-level data storage device according to an embodiment of the present invention.
3 shows the flow of each step of the single-level data storage method according to an embodiment of the present invention.
4 conceptually illustrates a flow of a read step in a method for storing single-level data according to an embodiment of the present invention.
5 conceptually illustrates a flow of a write step in a method for storing single-level data according to an embodiment of the present invention.

Advantages and features of the present invention, and a method of achieving them will become apparent with reference to the embodiments described below in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various forms, and only these embodiments make the disclosure of the present invention complete, and those skilled in the art to which the present invention pertains. It is provided to fully inform the person of the scope of the invention, and the scope of the invention is only defined by the claims.

In describing the embodiments of the present invention, detailed descriptions of known functions or configurations will be omitted except when actually necessary in describing the embodiments of the present invention. In addition, terms to be described later are terms defined in consideration of functions in an embodiment of the present invention, which may vary according to the intention or custom of users or operators. Therefore, the definition should be made based on the contents throughout this specification.

Since the present invention can make various changes and include various embodiments, specific embodiments will be illustrated in the drawings and described in the detailed description. However, this is not intended to limit the present invention to a specific embodiment, and should be understood as including all changes, equivalents, and substitutes included in the spirit and scope of the present invention.

Terms including an ordinal number such as first and second may be used to describe various elements, but the corresponding elements are not limited by these terms. These terms are only used for the purpose of distinguishing one component from another.

When a component is referred to as being'connected' or'connected' to another component, it is understood that it may be directly connected or connected to the other component, but other components may exist in the middle. Should be.

1 is a conceptual diagram illustrating a single-level data storage method according to an embodiment of the present invention. This single-level data storage method may be performed by the single-level data storage device 100.

Referring to FIG. 1, a single-level data storage device 100 may include a nonvolatile memory 1 and a disk 2. The nonvolatile memory 1 may store a first memory table 11, a second memory table 13, a compaction log 15, and a non-tree index 17. The nonvolatile memory 1 may be a non-volatile memory (NVRAM), and the disk 2 may be a type of memory that is distinct from the nonvolatile memory 1. For example, the disk 20 may be a hard disk or an SSD.

When data is input, it may be primarily stored in the first memory table 11. The first memory table 11 is a component that serves as an uptade buffer, and may be, for example, a MemTable. The first memory table 11 may be a kind of skip list data structure. The skip list data structure may be a type of data storage method in which a sequence of gradually sparse data is connected and arranged in a hierarchical structure. The first memory table 11 is permanent, and logging can be avoided.

All input data may be updated in the database through the first memory table 11.

When the capacity of the first memory table 11 is greater than or equal to a predetermined value, at least some data stored in the first memory table 11 may be stored in the second memory table 13. For example, the capacity of the first memory table 11 may be predetermined, and when such capacity is full, information stored in the first memory table 11 is stored in the second memory table 13 and 2 A flush plan may be established in the memory table 13. Here, flushing may mean storing information stored in the second memory table 13 in an SSTable, which is a memtable of a disk version, and a detailed description thereof will be omitted since it is easy for a person skilled in the art.

The second memory table 13 may perform a function similar to that of the first memory table 11. However, since the second memory table 13 is immutable, only a read operation is possible and an update or write operation may not be possible. The second memory table 13 may generate a new memtable when full data is deleted based on the flush being performed by the flush execution unit 20.

More specifically, when the first memory table 11 is full of information, the information of the first memory table 11 is stored in the second memory table 13 so that flushing is performed, and the first memory table 11 New information can be entered in 11). In some cases, when the first memory table 11 is filled with information, the first memory table 11 is converted into the second memory table 13 so that flushing is performed, and the new first memory table 11 Can be created.

If the second memory table 13 is not flushed, new information is input until the flush is performed, that is, a new first memory table 11 is created or new information on the first memory table 11 is performed. Can wait for input of.

When the flush is performed by the flush execution unit 20, data may be transferred from the second memory table 13 to the disk 2 for at least some of the data stored in the second memory table 13.

The disk 2 may include an SSTable as a file, and when flushing is performed, transmitted data (eg, a key value pair (hereinafter, a KV pair)) may be stored in the SSTable of the disk 2. SSTable More specifically, the flushed data may be stored in a single-level SSTable in the disk 2. Single-level is not a hierarchical multi-level multi-level structure. As illustrated, it may mean a single level structure in which an SSTable is formed as a single layer and stored.

Since the disk 2 has such a single-level structure, the non-tree index 17 can be used to search data stored in a single-level.

More specifically, in the case of a conventional LSM tree, data is stored in a multi-level structure, that is, a hierarchical structure, and data is searched by sequentially moving each step. On the other hand, in the case of a non-tree, since it is set to search data stored in a single-level structure, the disk 2 can perform data search using the non-tree index 17, which is a non-tree type index. I can. In this case, since it is not necessary to maintain the sorted order of the data stored in the disk 2, write amplification is reduced, and the efficiency of data retrieval may increase.

Here, the non-tree index 17 may be a data structure having an index for all data stored on the disk 2 to be described later. The non-tree index 17 may have a shape similar to that of a general non-tree. However, it may be stored in the nonvolatile memory 1 and permanently preserved.

In FIG. 1, there is shown an example in which six SSTables exist in the disk 2 and one KV pair is stored in each. Specifically, when six KV pairs are stored on the disk 2, the first KV pair 21-1, the second KV pair 21-2, the third KV pair 21-3, and the fourth KV The pair 21-4, the fifth KV pair 21-5, and the sixth KV pair 21-6 are each stored in the SSTable to form one layer. However, it is not limited thereto, and two or more KV pairs may be stored in one SSTable. That is, although not shown, the first KV pair 21-1 and the second KV pair 21-2 are stored in one SSTable, and the third KV pair 21-3 and the fourth KV pair 21- 4), the fifth KV pair 21-5, and the sixth KV pair 21-6 may be stored in another SSTable.

Meanwhile, in order to provide reasonable performance for a scan operation for the entire range of a given key value, the sequentiality of the KV pair stored in the disk 2 needs to be maintained above a predetermined value. Here, the sequentiality may mean a degree of how well the KV pairs are stored in a sorted order when they are stored on the disk. This sequence may be maintained above a certain level by performing compaction to be described later.

On the other hand, the KV pair that has become obsolete in the disk 2 needs to be deleted through garbage collection. That is, when a new KV pair is inserted into the disk 2 and the update is performed, the use value of the previously stored KV pair is lost. Accordingly, the previously stored KV pair needs to be deleted in consideration of the capacity aspect of the disk 2.

This operation may be referred to as compaction, and the compaction performing unit 23 may perform compaction. The compaction operation may be performed for each file, that is, for each SSTable. Based on the compaction operation, useless KV pairs are deleted among the KV pairs of the file, and the remaining KV pairs may be sorted based on the key value.

More specifically, the KV pair may appear in the form of a key and a value for the key. For example, if the key is 5 and the value for the key is 100, it may be expressed as <5, 100>. In this case, when a key of '5' is requested, a value of 100 may be obtained. In some cases, the value for the key may be updated multiple times. For example, for a key of '5', a value of '100' is currently stored, but you can update the value to '200' afterwards. In this case, when a KV pair of <5,100> is initially written to the file X, the non-tree can point X to the case where the key is '5'. Thereafter, if the KV pair <5,200> is written to the new file Y, the non-tree can point to Y for the case where the tree has a key of '5'. In the compaction process, when the key is 5, it is possible to check which file the non-tree points to. In this case, when compaction is performed on the X file, the previous KV pair may be deleted.

For another example, F1= (<1, 10>, <5, 20>, <10, 30>) and F2=(<3, 30>, <4, 40>, <10, 100>) And assuming that <10, 100> of F2 is the updated new value for key=10, and compaction is performed for F1 and F2, the new files F3 and F4 are saved as F3=(<1, 10>, <3 , 30>, <4, 40>), F4 = (<5, 20>, <10, 100>) can be created. In this process, <10, 30> can be deleted, and if F3 and F4 are created and stored on the disk, the existing F1 and F2 can be deleted.

The compaction log 15 stored in the nonvolatile memory 1 may record various types of information related to the execution of compaction.

When an input for reading data is applied, data search may be performed based on the non-tree index 17.

2 shows an example of a functional configuration of a single-level data storage device according to an embodiment of the present invention. Used below'… A term such as'negative' means a unit that processes at least one function or operation, which may be implemented by hardware or software, or a combination of hardware and software. Hereinafter, in the description of FIG. 2, content overlapping with FIG. 1 may be omitted.

Referring to FIG. 2, the single-level data storage device 100 includes a first memory table 110, a second memory table 120, a flush execution unit 130, a single-level storage unit 140, and a merger alignment. It may include an execution unit 150.

The first memory table 110 may be implemented by a computing device including a microprocessor, which is a second memory table 120, a flush execution unit 130, and a single-level storage unit 140 to be described later. ), the same for the merger and alignment performing unit 150.

The first memory table 110 is included in the nonvolatile memory 1 and may store data input to form a database. The first memory table 110 may have a configuration corresponding to the first memory table 11 of FIG. 1.

The second memory table 120 may store data stored in the first memory table 110. The second memory table 120 is an immutable memtable, and may store data of the first memory table 110 immutable.

The flush execution unit 130 may perform a flush operation on data stored in the second memory table 120. The flush execution unit 130 may flush data stored in the second memory table 120 to the disk 2. When the flush is performed, data corresponding thereto may be deleted from the second memory table 120.

The single-level storage unit 140 may store data received based on the flush operation as a single-level. Specifically, the single-level storage unit 140 may store at least some of the data stored in the second memory table 120 in a single-level on a disk 2 that is separate from the nonvolatile memory 1.

The single-level storage unit 140 may store at least some of the data stored in the second memory table as a single-level so that data is retrieved based on a B-tree data structure.

The compaction execution unit 150 may perform compaction on useless data among data stored in a single-level. Compaction may refer to a process of creating a new file while performing merge sort of files on key values.

Prior to performing the merger sort, a list of compaction candidates may be generated, and merge sort may be performed on data included in the list.

The compaction execution unit 150 may delete unnecessary data from data stored in a single-level and sort the remaining data based on a key value. Deleted data may be extracted as a compaction candidate and generated as a list, and merger and sorting may be performed on some of the data included in the list. In some cases, when an input for reading data is applied, a search may be performed based on a key value in merged and sorted data using the non-tree index 17.

For another example, the compaction performing unit 150 may identify a ratio of data stored in single-level data before the update time and data after the update time. Here, the data before the update point includes, for example, a new value written or updated, a previous value, that is, an invalid (or obsolete) KV, and data after the update point May contain a new value, that is, a valid KV.

When the identified ratio is lower than a predetermined value, the compaction performing unit 150 may select at least some of the data prior to the update time as a compaction candidate. Accordingly, the compaction performing unit 150 may maintain a ratio of data before the update time and data after the update time to a predetermined value or more.

As another example, the compaction performing unit 150 may scan a predetermined number of nodes among each node of the non-tree. At this time, each node of the non-tree may be associated with at least a part of data stored in a single-level, and the compaction execution unit 150 determines in which data table file the data related to the scanned node is distributed and stored. Can be identified. When the degree of distribution of data to the data table file is greater than or equal to a predetermined value, the compaction execution unit 150 may select the data table file itself or data stored in the data table file as a candidate for compaction.

Here, the data table file may be an SStable file, and may mean a file in which a KV pair is stored, and the non-tree scan method of the compaction execution unit 150 is based on a round-robin method. In this regard, it will be omitted as it is easy for a person skilled in the art.

As another example, the compaction execution unit 150 calculates the number of data table files required to read data stored in a single-level, and data related to the data with the largest number of data table files Table files can be selected as candidates for compaction.

The compaction performing unit 150 may delete useless KV pairs through compaction, and may select valid KV pairs in files including the KV pairs to be deleted in the process and generate a new file. For example, if you compact two files F1, F2 (in this case, F1= (<1, 10>, <5, 20>, <10, 30>) and F2=(<3, 30> , <4, 40>, <10, 100>), and assuming that <10, 100> of F2 is an updated new value for key=10), the compaction execution unit 150 is a new file called F3, F4 Can be generated as F3=(<1, 10>, <3, 30>, <4, 40>), F4=(<5, 20>, <10, 100>). In this case, the compaction execution unit 150 may update to point F3 and F4 for keys 1, 3, 4, 5, and 10 in the non-tree. After that, the compaction performing unit 150 may delete the existing files F1 and F2.

As such compaction is performed, the single-level-based data storage device 100 according to an embodiment of the present invention selectively performs compaction to provide appropriate scan operation performance, and at the same time, it is stored in the disk 2 space. It can prevent unnecessary data from accumulating. That is, the single-level based data storage device 100 performs data read/write operations more efficiently, and improves utilization of the disk 2 space.

3 shows the flow of each step of the single-level data storage method according to an embodiment of the present invention. It goes without saying that each step of the method illustrated in FIG. 3 may be performed in a different order as illustrated in the drawings depending on the case.

Referring to FIG. 3, data may be stored in the first memory table 110 included in the nonvolatile memory 1 (S110 ). Data of the first memory table 110 may be stored in the second memory table 120. The second memory table 120 is also included in the nonvolatile memory 1 and may invariably store a part of data stored in the first memory table 110.

A flush operation may be performed on the data stored in the second memory table 120 (S130). For example, when data is full in the second memory table 120, a flush operation may be performed (S130).

Data is transferred from the second memory table 120 to the disk 2 based on the flush operation, and the transferred data may be stored in a single-level (S140).

Data is stored in a single-level, and when useless data is generated by performing an update among the stored data, compaction may be performed (S150). When compaction is performed, merger sorting can be performed. The data to be merged and sorted can be determined in various ways.

For example, useless data among data stored in a single-level are deleted, and remaining data may be sorted based on a key value. Useless data could be data before it was updated.

For another example, among data stored in a single-level, the ratio of the data before the update point and the data after the update (live data) is identified, and if the identified ratio is lower than a predetermined value, data that is before a predetermined point in time After deleting old data so that the ratio of and valid data exceeds a predetermined value, the remaining data can be sorted.

For another example, by scanning a predetermined number of nodes among each node of the non-tree, it is possible to identify how many data table files exist in the section corresponding to the scanned node. When the degree of distribution for such a data table file is more than a predetermined value, that is, when the data existing in the section corresponding to the scanned node is distributed in a certain number of data table files or more, the data table file itself in which the data is distributed is compressed. Sean can be selected as a candidate.

As another example, the number of data table files required to read data stored in a single-level may be calculated, and a unit of data having the largest number of data table files may be selected as a compaction candidate. On the other hand, reading here is not reading the value of one key (point query), but reading all the values of the keys in the range given by range query or scan, for example, keys in the range 200 to 300. have. When such a read is performed, the scan range is divided into sub-ranges, and the number of files accessed to read one sub-range is recorded, and the number of accessed files is the largest sub-range. A file corresponding to can be selected as a compaction candidate.

4 conceptually illustrates a flow of a read step in a method for storing single-level data according to an embodiment of the present invention.

Referring to FIG. 4, the single-level data storage device 100 may search for data in the first memory table 110 as a first step to read data. If data is not found in the first memory table 110, data may be searched in the second memory table 120.

When data is not found in the second memory table 120 as well, data may be searched from the disk 2 based on a non-tree structure. On the other hand, a file containing data may be composed of blocks, and the index of the non-tree may include file information stored on the disk for the key, for example, offset information of the block of the file in which the key is stored. have. Accordingly, when data is searched using a non-tree structure, that is, a non-tree index, it is possible to more easily perform a KV search.

As described above, since data is stored on the disk 2 in a single-level form, a search can be performed using a non-tree structure.

5 conceptually illustrates a flow of a write step in a method for storing single-level data according to an embodiment of the present invention.

Referring to FIG. 5, all input data may be applied to the first memory table 110. When the first memory table 110 is full, at least some data stored in the first memory table 110 may be applied to the second memory table 120 and stored in an immutable form. Thereafter, flushing may be performed according to the state of the second memory table 120, and in some cases, compaction may be performed on data that is flushed and stored as a single-level.

When the flush is performed, information on data stored in the disk 2 in a single-level, for example, information on a file and an offset may be stored or updated in an index form in a non-tree. Based on this, when compaction is performed, a non-tree search may be performed to check whether a KV pair in a file targeted for compaction is valid.

A single-level data storage device and method according to an embodiment of the present invention enables efficient search by storing data in a single-level so that data search is performed quickly, and furthermore, a compaction candidate is determined and Merging and sorting is performed on some of the compaction candidates to prevent unnecessary data from being stored, thereby efficiently storing data.

A single-level data storage device and method according to an embodiment of the present invention enables high-performance data writing by storing data in the first memory table 110 and the second memory table 120, and uses a non-tree. By using it, high-performance data reading is possible.

Combinations of each block in the block diagram attached to the present specification and each step in the flowchart may be performed by computer program instructions. Since these computer program instructions can be mounted on the processor of a general purpose computer, special purpose computer, or other programmable data processing equipment, the instructions executed by the processor of the computer or other programmable data processing equipment are shown in each block or flowchart of the block diagram. Each step creates a means to perform the functions described. These computer program instructions can also be stored in computer-usable or computer-readable memory that can be directed to a computer or other programmable data processing equipment to implement a function in a particular way, so that the computer-usable or computer-readable memory It is also possible to produce an article of manufacture in which the instructions stored in the block diagram contain instruction means for performing the functions described in each block or flow chart. Computer program instructions can also be mounted on a computer or other programmable data processing equipment, so a series of operating steps are performed on a computer or other programmable data processing equipment to create a computer-executable process to create a computer or other programmable data processing equipment. It is also possible for the instructions to perform the processing equipment to provide steps for performing the functions described in each block of the block diagram and each step of the flowchart.

In addition, each block or each step may represent a module, segment, or part of code comprising one or more executable instructions for executing the specified logical function(s). In addition, it should be noted that in some alternative embodiments, functions mentioned in blocks or steps may occur out of order. For example, two blocks or steps shown in succession may in fact be performed substantially simultaneously, or the blocks or steps may sometimes be performed in the reverse order depending on the corresponding function.

The above description is merely illustrative of the technical idea of the present invention, and those of ordinary skill in the art to which the present invention pertains will be able to make various modifications and variations without departing from the essential quality of the present invention. Accordingly, the embodiments disclosed in the present specification are not intended to limit the technical idea of the present disclosure, but to explain the technical idea, and the scope of the technical idea of the present disclosure is not limited by these embodiments. The scope of protection of the present invention should be interpreted by the following claims, and all technical ideas within the scope equivalent thereto should be construed as being included in the scope of the present invention.

1: non-volatile memory
2: disk
100: single-level data storage device
110: first memory table
120: second memory table
130: flush execution unit
140: single-level storage
150: compaction execution unit

Claims (14)

Storing data in a first memory table included in a nonvolatile memory; and
Storing at least some of the data stored in the first memory table in a second memory table included in the nonvolatile memory when the capacity of the data stored in the first memory table is greater than or equal to a predetermined value; and
Identifying data stored in the second memory table and performing a flush operation; and
And storing data stored in the second memory table in a single-level on a disk separate from the nonvolatile memory based on the flush operation.
Single-level based data storage method. The method of claim 1,
The step of storing the single-level,
Storing data stored in the second memory table in a single-level based on the flush operation so that data is retrieved based on a non-tree data structure.
Single-level based data storage method. The method of claim 2,
Comprising the step of performing compaction on at least a portion of the data stored in the single-level
Single-level based data storage method. The method of claim 3,
The step of performing the compaction,
Deleting the data before the update among the data stored in the single-level, and sorting the remaining data stored in the single-level based on a key value included in the remaining data.
Single-level based data storage method. The method of claim 3,
The step of performing the compaction,
Identifying a ratio of data in which at least a part of the data stored in the single-level is stored before the update time and data after the update time,
If the identified ratio is lower than a predetermined value, including the step of selecting at least a part of the data before the update time as a compaction candidate,
Including the step of deleting at least some of the selected compaction candidates
Single-level based data storage method. The method of claim 3,
The step of performing the compaction,
Scanning a predetermined number of nodes among each node of the non-tree;
When the data related to the scanned node among the data stored in the single-level are distributed and stored in a predetermined number or more data table files, selecting the data table file as the compaction candidate
Single-level based data storage method. The method of claim 5,
The step of performing the compaction,
Calculating the number of data table files required to read the data stored in the single-level;
And selecting a data table file related to the data with the largest number of the calculated data table files as the compaction candidate.
Single-level based data storage method. A first memory table for storing input data,
A second memory table for storing at least some data stored in the first memory table when the capacity of the data stored in the first memory table is greater than or equal to a predetermined value;
A flush execution unit that identifies data stored in the second memory table and performs a flush operation;
A single-level storage unit for storing data stored in the second memory table in a single-level on a disk separate from the nonvolatile memory based on the flush operation,
The first memory table and the second memory table are stored in a nonvolatile memory.
Single-level based data storage device. The method of claim 8,
The single-level storage unit,
Storing data stored in the second memory table in a single-level based on the flush operation so that data is retrieved based on a B-tree data structure
Single-level based data storage device. The method of claim 9,
Comprising a compaction performing unit that performs compaction on at least a part of the data stored in the single-level
Single-level based data storage device. The method of claim 10,
The compaction performing unit,
Deleting the data before the update among the data stored in the single-level, and sorting the remaining data stored in the single-level based on a key value included in the remaining data
Single-level based data storage device. The method of claim 10,
The compaction performing unit,
Identifying a ratio of data in which at least a part of the data stored in the single-level is stored before the update time and data after the update time,
If the identified ratio is lower than a predetermined value, including the step of selecting at least a portion of the data before the predetermined time point as a compaction candidate,
Deleting at least some of the selected compaction candidates
Single-level based data storage device. The method of claim 10,
The compaction performing unit,
Scanning a predetermined number of nodes among each node of the non-tree,
Selecting the data table file as the compaction candidate when data related to the scanned node among the data stored in the single-level are distributed and stored in a predetermined number or more data table files.
Single-level based data storage device. The method of claim 10,
The compaction performing unit,
Calculate the number of data table files required to read the data stored in the single-level,
Selecting a data table file related to the data with the largest number of the calculated data table files as the compaction candidate
Single-level based data storage device.

Images