Database management method and system based on heterogeneous computing

Abstract

The invention discloses a database management method and system based on heterogeneous computing, and relates to the technical field of database management.The method comprises the steps that file data of a plurality of data volumes applying for storage is obtained, data fragmentation is conducted on the file data, then sub-file data of a plurality of fragments is generated, and the sub-file data is stored in heterogeneous computing equipment; the heterogeneous computing device carries out parallel computing on the sub-file data to generate a corresponding parallel computing task, the computing load of the heterogeneous computing device is judged, and whether task scheduling is carried out or not is determined according to the judgment result; according to the method, a plurality of sub-file data subjected to parallel computing are subjected to data merging, then final storage data are generated, the final storage data are stored in a database, and regular data synchronization and regular data check are performed, so that the problems of data security, data consistency, load balance of data processing and the like are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of database management, and in particular to a database management method and system based on heterogeneous computing. Background Art

[0002] With the rapid development of Internet technology, the amount of data has exploded. The scale of data that enterprises and institutions need to process and manage is getting larger and larger. Traditional database management technology can no longer meet the needs. In order to improve the efficiency and performance of database processing, some new technologies have emerged in recent years, such as distributed databases and cloud computing. However, these technologies still have certain limitations, such as data security, data consistency, and load balancing of data processing. How to solve these problems is what we need to consider now. Summary of the invention

[0003] In order to solve the above problems, the object of the present invention is to provide a database management method and system based on heterogeneous computing.

[0004] The object of the present invention can be achieved by the following technical solution: A database management method based on heterogeneous computing, comprising the following steps: Step S1: obtaining a certain amount of file data applied for storage, slicing the file data, and then generating a certain number of sub-file data segments, and storing the sub-file data in a heterogeneous computing device; Step S2: The heterogeneous computing devices perform parallel computing on the sub-file data, thereby generating corresponding parallel computing tasks, judging the computing load of the heterogeneous computing devices, and deciding whether to perform task scheduling based on the judgment result; Step S3: Merge the data of several sub-files that have been calculated in parallel to generate final storage data, store the final storage data into the database, and perform regular data synchronization and regular data checking.

[0005] Furthermore, the process of obtaining a certain amount of file data applied for storage, performing data slicing, and then generating a certain number of sub-file data segments includes: Obtain several copies of file data and number them, denoted as i, then i=1, 2, 3, ..., n, where n is a natural number greater than 0, bind an initial storage identifier to each file data, the storage identifier includes identifier A and identifier B, different storage identifiers have corresponding storage status, the storage status includes applying for storage and not applying for storage, identifier A—applying for storage, identifier B—not applying for storage; obtain part of the file data corresponding to identifier A as the data set to be sharded, and obtain the data volume of the data set to be sharded, denoted as D1, set the shard capacity, denoted as D2, perform data sharding to generate several fragments of sub-file data, the number of fragments of the sub-file data is denoted as S, and S=D1 / D2.

[0006] Furthermore, the process of storing the sub-file data in the heterogeneous computing device includes: Obtain the sub-file data of S fragments and the corresponding initial file parameters, the initial file parameters include file integrity, file generation environment variables and external attack frequency, number the sub-file data of S fragments in sequence, record them as j, j = 1, 2, 3, ..., S, where S is a natural number greater than 0, and then mark the file integrity, file generation environment variables and external attack frequency corresponding to the sub-file data numbered j as Qt[j], Wm[j] and Er[j] respectively; A number of heterogeneous computing devices are set up, each of which is associated with a corresponding parameter screening program. The parameter screening program records reference data items of Qt[j], Wm[j] and Er[j], which are marked as α, β and γ respectively. If one of the following is true: Qt[j] is inconsistent with α, Wm[j] is inconsistent with β, and Er[j] is inconsistent with γ, the parameter screening program generates information 1; otherwise, information 2 is generated. When information 1 is obtained, no storage operation is performed. When information 2 is obtained, the sub-file data is stored in the corresponding heterogeneous computing device.

[0007] Furthermore, the process of performing parallel computing on sub-file data by heterogeneous computing devices to generate corresponding parallel computing tasks includes: When several heterogeneous computing devices receive sub-file data, parallel computing of the several sub-file data is started, and the pre-preparation time and computing time of the parallel computing are set. During the pre-preparation time, the computing environment and computing power of the current heterogeneous computing device are configured through a preset computing configuration file. During the computing time, parallel computing of the several sub-file data is performed, and corresponding parallel computing tasks are generated. The parallel computing tasks are associated with corresponding computing task data information.

[0008] Furthermore, the process of judging the computing load of the heterogeneous computing device and deciding whether to perform task scheduling according to the judgment result includes: The computing task data information includes the task difficulty coefficient, the task load requirement coefficient and the task processing deadline. The judgment of the computing load includes different judgment items. The judgment items include judgment item 1 and judgment item 2. The judgment item 1 is used to judge whether the task difficulty coefficient and the task load requirement coefficient meet their respective preset thresholds. The judgment item 2 is used to judge whether the parallel computing task can be completed within the task processing deadline. The judgment result includes result 1 and result 2. The judgment item 1 corresponds to result 1, and the judgment item 2 corresponds to result 2. The contents of result 1 and result 2 are: generating a field "0" or "1"; If either result 1 or result 2 generates the field "0", it means that the current heterogeneous computing device has a problem of exceeding the computing load, and task scheduling is performed to send the current parallel computing task to the preset cloud platform for processing. Otherwise, it means that the current heterogeneous computing device does not have a problem of exceeding the computing load, and task scheduling is not performed.

[0009] Furthermore, the process of merging the data of several sub-files to generate final storage data and storing the data in the database includes: The parallel computing task is set with a task progress bar, and the task progress bar is set with a left endpoint value and a right endpoint value, and the corresponding values ​​are 0% and 100% respectively. As the parallel computing task is processed, the value of the task progress bar gradually increases. When the value is 100%, it means that the parallel computing task is completed. The sub-file data of several parallel computing tasks that have been completed are obtained, and a discrimination identifier is bound to it, recorded as Sign, and a data fusion space is established. All sub-file data are imported into the data fusion space for data merging. Before the data is merged, the data fusion space determines whether the Sign corresponding to the sub-file data has changed. If so, it means that an external IP has tampered with the sub-file data, and the sub-file data is regenerated. If not, the final storage data is generated through data merging, and the final storage data is stored in the database.

[0010] Furthermore, the process of performing regular data synchronization and regular data checking includes: The database is provided with a plurality of storage partitions, each storage partition is used to store sub-file data of a heterogeneous computing device, the database is provided with a storage instruction area, periodic data synchronization is associated with a corresponding synchronization instruction, periodic data inspection is associated with a corresponding inspection instruction, which are respectively denoted as P1 and P2, and the synchronization instruction P1 is associated with an instruction text 1, and the instruction text 1 is used to record the synchronization frequency and the information generated by reading the data synchronization, and the inspection instruction P2 is associated with an instruction text 2, and the instruction text 2 is used to record the required inspection items and the inspection frequency; When the storage instruction area receives P1, it regularly synchronizes the data of several storage partitions according to the synchronization frequency, and exports the information generated by the data synchronization to generate a synchronization log. When the storage instruction area receives P2, it performs data checks on several storage partitions according to the inspection frequency and required inspection items. If error information is detected, an inspection log is generated. Otherwise, no operation is performed, and the synchronization log and inspection log are transmitted to the administrator, who arranges the corresponding database maintenance personnel to handle them.

[0011] Furthermore, a database management system based on heterogeneous computing includes: The data slicing module obtains the file data of a certain amount of data applied for storage, slices the file data, generates a certain number of sub-file data segments, and stores the sub-file data in a heterogeneous computing device; The parallel computing module uses heterogeneous computing devices to perform parallel computing on sub-file data, thereby generating corresponding parallel computing tasks, judging the computing load of the heterogeneous computing devices, and deciding whether to schedule tasks based on the judgment results; The data merging module merges the data of several sub-files that have undergone parallel calculations to generate final storage data, stores the final storage data into the database, and performs regular data synchronization and regular data checks.

[0012] Compared with the prior art, the beneficial effects of the present invention are as follows: by slicing file data and generating several fragments of sub-file data and storing them in heterogeneous computing devices, the heterogeneous computing devices perform parallel calculations on the sub-file data, and judge the computing load during parallel calculations, and then decide whether to perform task scheduling based on the judgment results, thereby reducing the computing load of the current heterogeneous computing devices to a certain extent and achieving the purpose of load balancing; merging several sub-file data to generate final storage data, and performing regular data synchronization and regular data checking on the final storage data, thereby ensuring data consistency and improving data security. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0014] like Figure 1 As shown, a database management method based on heterogeneous computing includes the following steps: Step S1: obtaining a certain amount of file data applied for storage, slicing the file data, and then generating a certain number of sub-file data segments, and storing the sub-file data in a heterogeneous computing device; Step S2: The heterogeneous computing devices perform parallel computing on the sub-file data, thereby generating corresponding parallel computing tasks, judging the computing load of the heterogeneous computing devices, and deciding whether to perform task scheduling based on the judgment result; Step S3: Merge the data of several sub-files that have been calculated in parallel to generate final storage data, store the final storage data into the database, and perform regular data synchronization and regular data checking.

[0015] It should be further explained that, in the specific implementation process, the process of obtaining a certain amount of file data applied for storage, performing data slicing, and then generating a certain number of sub-file data segments includes: Obtain several pieces of file data and number the file data, denoted by i, i=1, 2, 3, ..., n, where n is a natural number greater than 0, and bind an initial storage identifier to each file data; The storage identification includes identification A and identification B. Different storage identifications have corresponding storage statuses. The storage statuses include applying for storage and not applying for storage. The corresponding relationship between the storage identification and the storage status is as follows: identification A—applying for storage, identification B—not applying for storage; Get part of the file data corresponding to the identifier A as the data set to be sharded, and then get the data volume of the data set to be sharded, denoted as D1, set the shard capacity, denoted as D2, perform data sharding, and then generate several fragments of sub-file data. The number of fragments of the sub-file data is denoted as S, then S=D1 / D2.

[0016] It should be further explained that, in a specific implementation process, the process of storing sub-file data in a heterogeneous computing device includes: Acquire S fragments of sub-file data, each of which is provided with corresponding initial file parameters, wherein the initial file parameters include file integrity, file generation environment variables, and external attack frequency; The sub-file data of the S fragments are sequentially numbered in sequence, and the number is recorded as j, then j = 1, 2, 3, ..., S, where S is a natural number greater than 0, and S corresponds to the last sub-file data, and then the file integrity, file generation environment variables and external attack frequency corresponding to the sub-file data numbered j are marked as Qt[j], Wm[j] and Er[j] respectively; Set a number of heterogeneous computing devices, and obtain the numbers of the heterogeneous computing devices, which are recorded as IDs, and are strings of letters and numbers. Each heterogeneous computing device is associated with a corresponding parameter screening program, and the parameter screening program records reference data items of Qt[j], Wm[j], and Er[j], which are marked as α, β, and γ respectively; If Qt[j] is inconsistent with α, Wm[j] is inconsistent with β, and Er[j] is inconsistent with γ, then the parameter screening program generates information 1, which reads: "The current sub-file data screening fails, storage is prohibited!" Otherwise, the parameter screening program generates information 2, the content of which is: "The current sub-file data screening is qualified, please store it in the heterogeneous computing device"; When information 1 is obtained, no storage operation is performed. When information 2 is obtained, the sub-file data is stored in the corresponding heterogeneous computing device, and an associated sequence is generated, denoted as R, with R=<j,ID> .

[0017] It should be noted that the reference data item α corresponds to the minimum limit value of the file integrity. If it is lower than the minimum limit value, it indicates that the comparison is inconsistent. The reference data item β corresponds to the normal value range of all parameters of the file generation environment variables. If any parameter is not within the normal value range, it indicates that the comparison is inconsistent. The parameters include the environment network speed, environment bandwidth and network packet loss rate. The reference data item γ corresponds to the normal value range of the external attack frequency. If the value of Er[j] is within the γ value range, the comparison is consistent.

[0018] It should be further explained that, in the specific implementation process, the process of performing parallel computing on sub-file data by heterogeneous computing devices and then generating corresponding parallel computing tasks includes: When several heterogeneous computing devices receive the sub-file data, they start parallel computing of the corresponding sub-file data, set the pre-preparation time and computing time of the parallel computing, which are recorded as T and T' respectively, T = [t1, t2], T' = (t2, t3], t1 < t2 < t3, where t1 is the initial time of starting the parallel computing, and t3 is the end time of the parallel computing; During the time corresponding to the lead preparation time T, the computing environment and computing power of the current heterogeneous computing device are configured through the preset computing configuration file, and during the time corresponding to the computing time T', parallel computing of several sub-file data is performed, and corresponding parallel computing tasks are generated; The parallel computing tasks are associated with corresponding computing task data information; The computing configuration file records the configuration codes of several environmental parameters required by the computing environment. The initial values ​​of the environmental parameters of the computing environment are adjusted through the configuration codes. The initial values ​​are non-standard values. The computing environment is configured successfully until the initial values ​​reach the preset standard values. The initial value of the computing power is 0. The initial value is increased by a fixed value until the required computing power value of the current heterogeneous computing device is met, and then the increase is stopped. The fixed value can be manually input and changed.

[0019] It should be further explained that, in the specific implementation process, the process of judging the computing load of the heterogeneous computing device and deciding whether to perform task scheduling according to the judgment result includes: Obtain the computing task data information corresponding to each heterogeneous computing device, wherein the computing task data information includes the task difficulty coefficient, the task load requirement coefficient, and the task processing deadline, and record the task difficulty coefficient, the task load requirement coefficient, and the task processing deadline as Task1, Task2, and T respectively. 截 ; The judgment of the computing load includes different judgment items, the judgment items include judgment item 1 and judgment item 2, the judgment item 1 is used to judge whether the task difficulty coefficient and the task load requirement coefficient meet their respective preset thresholds, the judgment item 2 is used to judge whether the parallel computing task can be completed within the task processing deadline, and the judgment results include result 1 and result 2; The corresponding relationship is as follows: Judgment item one corresponds to result one, and judgment item two corresponds to result two; The thresholds of the task difficulty coefficient and the task load requirement coefficient are set to ψ1 and ψ2 respectively. If any one of Task1>ψ1 or Task2>ψ2 holds, the content of Result 1 is: Generate field "0", otherwise, the content of Result 1 is: Generate field "1"; Intercept a unit time and obtain the amount of data of the parallel computing task processed in the unit time, recorded as C 单 , preset the expected amount of data to be processed per unit time, denoted as C 期 , if C 单 ≤C 期 , it means that at the task processing deadline T 截 Within, heterogeneous computing devices can complete the corresponding parallel computing tasks, and the corresponding result 2 is: Generate field "1", if C 单 >C 期 , then the content of result 2 is: generate field "0"; If either result 1 or result 2 generates the field "0", it means that the current heterogeneous computing device has a problem of exceeding the computing load, and task scheduling is performed to send the current parallel computing task to the preset cloud platform for processing. Otherwise, it means that the current heterogeneous computing device does not have a problem of exceeding the computing load, and task scheduling is not performed.

[0020] It should be further explained that, in the specific implementation process, the data of several sub-files that have been calculated in parallel are merged to generate the final storage data, and the process of storing the final storage data into the database includes: Each parallel computing task that has been parallelized is set with a corresponding task progress bar. The task progress bar is set with a left endpoint value and a right endpoint value. The value of the left endpoint value is 0%, and the value of the right endpoint value is 100%. As the parallel computing task is processed, the value of the task progress bar gradually increases. When the value of the task progress bar is 100%, it indicates that the parallel computing task has been completed. Obtain the sub-file data of several parallel computing tasks that have been executed, bind a discriminant identifier to each sub-file data, record it as Sign, establish a data fusion space, and import all sub-file data into the data fusion space for data merging; Before data merging, the data fusion space determines whether the Sign corresponding to the sub-file data has changed. If so, it means that an external IP has tampered with the sub-file data, and the sub-file data is regenerated. If not, the final storage data is generated through data merging and stored in the database.

[0021] It should be further explained that, in the specific implementation process, the process of regular data synchronization and regular data checking includes: The database is provided with a plurality of storage partitions, each storage partition is used to store sub-file data of a heterogeneous computing device, and the database is also provided with a storage instruction area, the periodic data synchronization is associated with a corresponding synchronization instruction, and the periodic data check is associated with a corresponding check instruction; The synchronization instruction and the inspection instruction are respectively recorded as P1 and P2. The synchronization instruction P1 is associated with an instruction text 1, and the instruction text 1 is used to record the synchronization frequency and the information generated by reading data synchronization. The inspection instruction P2 is associated with an instruction text 2, and the instruction text 2 is used to record the required inspection items and the inspection frequency. When the storage instruction area receives P1, it periodically synchronizes the data of several storage partitions according to the synchronization frequency, and exports the information generated by the data synchronization to generate a synchronization log. When the storage instruction area receives P2, it performs data checks on several storage partitions according to the check frequency and required check items. If error information is found, a check log is generated. Otherwise, no operation is performed. The synchronization log and the inspection log are transmitted to the administrator, who arranges the corresponding database maintenance personnel to process them.

[0022] The present invention also provides a database management system based on heterogeneous computing, which includes the following modules: The data slicing module obtains the file data of a certain amount of data applied for storage, slices the file data, generates a certain number of sub-file data segments, and stores the sub-file data in a heterogeneous computing device; The parallel computing module uses heterogeneous computing devices to perform parallel computing on sub-file data, thereby generating corresponding parallel computing tasks, judging the computing load of the heterogeneous computing devices, and deciding whether to schedule tasks based on the judgment results; The data merging module merges the data of several sub-files that have undergone parallel calculations to generate final storage data, stores the final storage data into the database, and performs regular data synchronization and regular data checks.

[0023] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A database management method based on heterogeneous computing, characterized in that: The following steps are involved: Step S1: obtaining a certain amount of file data applied for storage, slicing the file data, and then generating a certain number of sub-file data segments, and storing the sub-file data in a heterogeneous computing device; Step S2: The heterogeneous computing devices perform parallel computing on the sub-file data, thereby generating corresponding parallel computing tasks, judging the computing load of the heterogeneous computing devices, and deciding whether to perform task scheduling based on the judgment result; Step S3: Merge the data of several sub-files that have been calculated in parallel to generate final storage data, store the final storage data into the database, and perform regular data synchronization and regular data checking.

2. The database management method based on heterogeneous computing according to claim 1, characterized in that: The process of obtaining a certain amount of file data applied for storage, performing data slicing, and then generating a certain number of sub-file data segments includes: Obtain several copies of file data and number them, denoted as i, then i=1, 2, 3, ..., n, where n is a natural number greater than 0, bind an initial storage identifier to each file data, the storage identifier includes identifier A and identifier B, different storage identifiers have corresponding storage status, the storage status includes applying for storage and not applying for storage, identifier A—applying for storage, identifier B—not applying for storage; obtain part of the file data corresponding to identifier A as the data set to be sharded, and obtain the data volume of the data set to be sharded, denoted as D1, set the shard capacity, denoted as D2, perform data sharding to generate several fragments of sub-file data, the number of fragments of the sub-file data is denoted as S, and S=D1 / D2.

3. The database management method based on heterogeneous computing according to claim 2 is characterized in that: The process of storing sub-file data in a heterogeneous computing device includes: Obtain the sub-file data of S fragments and the corresponding initial file parameters, the initial file parameters include file integrity, file generation environment variables and external attack frequency, number the sub-file data of S fragments in sequence, record them as j, j = 1, 2, 3, ..., S, where S is a natural number greater than 0, and then mark the file integrity, file generation environment variables and external attack frequency corresponding to the sub-file data numbered j as Qt[j], Wm[j] and Er[j] respectively; A number of heterogeneous computing devices are set up, each of which is associated with a corresponding parameter screening program. The parameter screening program records reference data items of Qt[j], Wm[j] and Er[j], which are marked as α, β and γ respectively. If one of the following is true: Qt[j] is inconsistent with α, Wm[j] is inconsistent with β, and Er[j] is inconsistent with γ, the parameter screening program generates information 1; otherwise, information 2 is generated. When information 1 is obtained, no storage operation is performed. When information 2 is obtained, the sub-file data is stored in the corresponding heterogeneous computing device.

4. The database management method based on heterogeneous computing according to claim 3 is characterized in that: The process of performing parallel computing on sub-file data by heterogeneous computing devices to generate corresponding parallel computing tasks includes: When several heterogeneous computing devices receive sub-file data, parallel computing of the several sub-file data is started, and the pre-preparation time and computing time of the parallel computing are set. During the pre-preparation time, the computing environment and computing power of the current heterogeneous computing device are configured through a preset computing configuration file. During the computing time, parallel computing of the several sub-file data is performed, and corresponding parallel computing tasks are generated. The parallel computing tasks are associated with corresponding computing task data information.

5. The database management method based on heterogeneous computing according to claim 4 is characterized in that: The process of judging the computing load of heterogeneous computing devices and deciding whether to schedule tasks based on the judgment results includes: The computing task data information includes the task difficulty coefficient, the task load requirement coefficient and the task processing deadline. The judgment of the computing load includes different judgment items. The judgment items include judgment item 1 and judgment item 2. Judgment item 1 is used to judge whether the task difficulty coefficient and the task load requirement coefficient meet their respective preset thresholds. Judgment item 2 is used to judge whether the parallel computing task can be completed within the task processing deadline. The judgment result includes result 1 and result 2. Judgment item 1 corresponds to result 1, and judgment item 2 corresponds to result 2. The contents of result 1 and result 2 are: generating field "0" or "1"; If either result 1 or result 2 generates the field "0", it means that the current heterogeneous computing device has a problem of exceeding the computing load, and task scheduling is performed to send the current parallel computing task to the preset cloud platform for processing. Otherwise, it means that the current heterogeneous computing device does not have a problem of exceeding the computing load, and task scheduling is not performed.

6. The database management method based on heterogeneous computing according to claim 5, characterized in that: The process of merging several sub-file data to generate final storage data and storing them in the database includes: The parallel computing task is set with a task progress bar, and the task progress bar is set with a left endpoint value and a right endpoint value, and the corresponding values ​​are 0% and 100% respectively. As the parallel computing task is processed, the value of the task progress bar gradually increases. When the value is 100%, it means that the parallel computing task is completed. The sub-file data of several parallel computing tasks that have been completed are obtained, and a discrimination identifier is bound to it, recorded as Sign, and a data fusion space is established. All sub-file data are imported into the data fusion space for data merging. Before the data is merged, the data fusion space determines whether the Sign corresponding to the sub-file data has changed. If so, it means that an external IP has tampered with the sub-file data, and the sub-file data is regenerated. If not, the final storage data is generated through data merging, and the final storage data is stored in the database.

7. A database management method based on heterogeneous computing according to claim 6, characterized in that: The process of performing regular data synchronization and regular data checking includes: The database is provided with a plurality of storage partitions, each storage partition is used to store sub-file data of a heterogeneous computing device, the database is provided with a storage instruction area, periodic data synchronization is associated with a corresponding synchronization instruction, periodic data inspection is associated with a corresponding inspection instruction, which are respectively denoted as P1 and P2, and the synchronization instruction P1 is associated with an instruction text 1, and the instruction text 1 is used to record the synchronization frequency and the information generated by reading the data synchronization, and the inspection instruction P2 is associated with an instruction text 2, and the instruction text 2 is used to record the required inspection items and the inspection frequency; When the storage instruction area receives P1, it regularly synchronizes the data of several storage partitions according to the synchronization frequency, and exports the information generated by the data synchronization to generate a synchronization log. When the storage instruction area receives P2, it performs data checks on several storage partitions according to the inspection frequency and required inspection items. If error information is detected, an inspection log is generated. Otherwise, no operation is performed, and the synchronization log and inspection log are transmitted to the administrator, who arranges the corresponding database maintenance personnel to handle them.

8. A database management system based on heterogeneous computing, used to implement the database management method according to any one of claims 1 to 7, the system comprising: The data slicing module obtains the file data of a certain amount of data applied for storage, slices the file data, generates a certain number of sub-file data segments, and stores the sub-file data in a heterogeneous computing device; The parallel computing module uses heterogeneous computing devices to perform parallel computing on sub-file data, thereby generating corresponding parallel computing tasks, judging the computing load of the heterogeneous computing devices, and deciding whether to schedule tasks based on the judgment results; The data merging module merges the data of several sub-files that have undergone parallel calculations to generate final storage data, stores the final storage data into the database, and performs regular data synchronization and regular data checks.

Images