Distributed database management method and system

Abstract

The invention discloses a distributed database management method and system, and the method comprises the steps: determining an incidence relation between data in a database, and carrying out the classification of the data based on the incidence relation, and obtaining a plurality of data fragments; determining security features of each piece of data in the data fragments, and obtaining security levels based on security analysis and calculation of the data fragments; determining a storage node of each data fragment based on the security level, and respectively storing the data fragments; determining load prediction data of each storage node, and evaluating the load state of the storage node based on the load prediction data to obtain a load state evaluation value; and determining a dynamic storage adjustment strategy of the data fragments based on the load state evaluation value, and dynamically adjusting the storage distribution of the data fragments on the storage nodes based on the dynamic storage adjustment strategy. According to the invention, by dynamically adjusting the storage distribution of the data fragments on the storage nodes, the storage resources are pooled, and the perception load and intelligent scheduling are given to the storage resources, so that the data distribution can dynamically and smoothly adapt to the instantaneous change of the service load.

Description

Technical Field

[0001] This invention relates to the field of data management technology, and in particular to a distributed database management method and system. Background Technology

[0002] With the full-scale launch of the digital economy era, data has become a core production factor driving enterprise development. The total amount of global data is growing exponentially, and traditional single-machine centralized databases are increasingly inadequate in handling massive data storage, high-concurrency access, business continuity, and agile elastic scaling. Their inherent performance bottlenecks, limited scalability, and single-point-of-failure risks make it difficult to support the stringent requirements of globalized business and uninterrupted service in modern internet, IoT, and fintech scenarios.

[0003] While existing static distributed database management methods have solved the initial problem of massive storage, their pre-planned, fixed-number sharding model is inadequate when faced with such extreme elasticity requirements. The lag in manual scaling, the long downtime window caused by data migration, and the hotspot bottleneck caused by uneven sharding all hinder agile business innovation. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a distributed database management method and system, comprising: Determine the relationships between the data in the database, and classify the data based on these relationships to obtain multiple data shards; Determine the security characteristics of each data in the data shard, and analyze and calculate the security of the data shard based on the security characteristics of each data to obtain the security level of each data shard; The storage nodes for each data shard are determined based on the security level, and the data shards are stored in their respective storage nodes. Determine the load prediction data for each storage node, and evaluate the load status of the storage nodes based on the load prediction data to obtain the load status evaluation value for each storage node. The dynamic storage adjustment strategy for data shards is determined based on the load status assessment value, and the storage distribution of data shards on storage nodes is dynamically adjusted based on the dynamic storage adjustment strategy.

[0005] Furthermore, the step of obtaining the relationships between various data in the database and classifying the data based on these relationships to obtain multiple data fragments includes: Obtain data information from each data point in the database and construct a relationship graph based on the data information, where nodes represent data and edges represent the relationships between data points; Determine the connection query frequency and the number of common modifications between the two ends of each edge from the association relationship, and evaluate the connection query frequency and the number of common modifications respectively; The evaluation results are summed to obtain the correlation between the data connected at both ends of each edge. Data with a correlation greater than a preset threshold are grouped into the same data shard, resulting in multiple data shards.

[0006] Furthermore, the process of determining the security characteristics of each data segment within the data shard, and analyzing and calculating the security of the data shards based on these characteristics to obtain the security level of each data shard includes: The security characteristics of each data segment are determined from the data information, including the content of sensitive data and the content of confidential data. The content of sensitive data and confidential data are evaluated and values ​​are obtained respectively to obtain sensitivity evaluation value and confidentiality evaluation value, and a pre-set security level conversion coefficient is determined; The security level of each data fragment is calculated based on its sensitivity and confidentiality assessment values ​​and security level conversion coefficient.

[0007] Furthermore, the formula for calculating the security level of the data fragmentation is as follows: , Where S is the security level of the data fragment, k is the security level conversion coefficient, X is the sensitivity assessment value of the data fragment, and n is the confidentiality assessment value of the data fragment.

[0008] Furthermore, the step of determining the storage node for each data shard based on the security level and storing the data shards in their respective storage nodes includes: A preset security level-security level range correspondence is set in advance. For each security level range, a corresponding preset security level is associated with it. Obtain the security level of each data shard, and based on the mapping relationship between the security level range to which the security level belongs and the preset security level-security level range correspondence, select the preset security level corresponding to the security level range and determine it as the security level of the storage node corresponding to each data shard. Each data shard is stored in a storage node with the corresponding security level, and backups of each data shard are stored in other storage nodes with the same security level.

[0009] Furthermore, determining the load prediction data for each storage node includes: Determine the pre-defined load metrics and the corresponding prediction models for each load metric. The load metrics include CPU utilization, disk utilization, and memory utilization. Based on the prediction model corresponding to each load indicator, the prediction output is obtained to obtain the prediction value corresponding to each load indicator, and the prediction value corresponding to each load indicator is determined as the load prediction data of each storage node.

[0010] Furthermore, the assessment of the load status of storage nodes based on load prediction data to obtain a load status assessment value for each storage node includes: Determine the pre-set baseline value for each load indicator, and calculate the difference between the predicted value and the baseline value for each load indicator. These differences are evaluated and values ​​are obtained to obtain the estimated difference of each load index, and the preset weight of each load index is determined. The load status assessment value of the storage node is obtained by calculating based on the preset weights and differential estimates of each load indicator.

[0011] Furthermore, the formula for calculating the load status assessment value of the storage node is as follows: , Where L is the load status assessment value of the storage node, bi is the preset weight of the i-th load indicator, Pi is the differential estimate of the i-th load indicator, and n is the number of load indicators.

[0012] Furthermore, the step of determining the dynamic storage adjustment strategy for data shards based on load status assessment values, and dynamically adjusting the storage distribution of data shards on storage nodes based on the dynamic storage adjustment strategy, includes: Identify storage nodes whose load status assessment value is greater than a preset threshold, and determine the security level corresponding to the storage node; Data shards that cannot meet the load requirements in this storage node are migrated to other storage nodes with the same security level to dynamically adjust the storage distribution of data shards on the storage nodes.

[0013] The present invention also provides a distributed database management system, comprising: The classification module is used to determine the relationships between data in the database and classify the data based on these relationships to obtain multiple data fragments. The calculation module is used to determine the security characteristics of each data in the data shard, and to analyze and calculate the security of the data shard based on the security characteristics of each data, so as to obtain the security level of each data shard. The storage module is used to determine the storage node for each data shard based on the security level, and to store the data shards to the corresponding storage nodes respectively; The evaluation module is used to determine the load prediction data for each storage node and evaluate the load status of the storage nodes based on the load prediction data, thereby obtaining the load status evaluation value for each storage node. The adjustment module is used to determine the dynamic storage adjustment strategy for data shards based on the load status assessment value, and to dynamically adjust the storage distribution of data shards on storage nodes based on the dynamic storage adjustment strategy.

[0014] Compared with existing technologies, the distributed database management method and system of this invention have the following advantages: This invention pools storage resources by dynamically adjusting the storage distribution of data shards on storage nodes and endows them with load awareness and intelligent scheduling, enabling data distribution to dynamically and smoothly adapt to the ever-changing business load. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the process structure of the distributed database management method in an embodiment of the present invention; Figure 2 This is a schematic diagram of the composition of the distributed database management system in an embodiment of the present invention. Detailed Implementation

[0016] The specific embodiments of this application will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0017] In the description of this application, it should be understood that the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the platform or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0018] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0019] like Figure 1As shown in the embodiments of this application, a distributed database management method is provided, including: S100: determining the association relationship between various data in the database, and classifying the data based on the association relationship to obtain multiple data shards; S200: determining the security characteristics of each data in the data shards, and analyzing and calculating the security of the data shards based on the security characteristics of each data to obtain the security level of each data shard; S300: determining the storage nodes of each data shard based on the security level, and storing the data shards in the corresponding storage nodes respectively; S400: determining the load prediction data of each storage node, and evaluating the load status of the storage nodes based on the load prediction data to obtain the load status evaluation value of each storage node; S500: determining the dynamic storage adjustment strategy of the data shards based on the load status evaluation value, and dynamically adjusting the storage distribution of the data shards on the storage nodes based on the dynamic storage adjustment strategy.

[0020] Furthermore, this invention pools storage resources by dynamically adjusting the storage distribution of data shards on storage nodes and endows them with load awareness and intelligent scheduling, enabling the data distribution to dynamically and smoothly adapt to the ever-changing business load.

[0021] In embodiments of this application, a distributed database management method is provided. The method involves obtaining the relationships between data in the database and classifying the data based on these relationships to obtain multiple data shards. The method includes: obtaining data information for each data item in the database and constructing a relationship graph based on the data information, where nodes represent data and edges represent relationships between data; determining the connection query frequency and the number of common modifications between the data connected at both ends of each edge from the relationship, and evaluating the connection query frequency and the number of common modifications respectively; summing the evaluation results to obtain the degree of association between the data connected at both ends of each edge, and grouping data with a degree of association greater than a preset threshold into the same data shard to obtain multiple data shards.

[0022] Specifically, metadata information is retrieved from the database. Based on this, a graph structure is constructed with data entities as nodes and relationships between entities as edges. The number of JOIN queries involving these relationships is counted within a specific time window; the higher the frequency, the higher the value, indicating a deeper coupling between the two types of data at the query level. The number of times data at both ends of the relationship is modified simultaneously during a transaction is also counted; the more times, the higher the value, indicating a need for strong consistency guarantees at the transaction level. The above two evaluation values ​​are added together to calculate the comprehensive relationship score between the two data entities connected by each edge. A preset threshold is set, and all strongly related data entities with relationship scores higher than this threshold are forcibly assigned or migrated to the same data shard. Conversely, data with weak relationships may be distributed to different shards. By traversing all strongly related edges in the graph and performing clustering, the entire dataset is finally divided into multiple data shards. Each shard contains multiple data entities, and these entities have a high degree of access and modification locality. This step places strongly correlated data within the same shard, transforming complex JOIN queries and distributed transactions that previously required cross-network and cross-node collaboration into local operations completed within a single database node. This significantly reduces query latency, increases throughput, and avoids the performance overhead and complexity of distributed transactions (such as two-phase commit), thereby greatly improving the overall performance of the database. The sharding strategy is no longer based on guesswork or simple hash / range rules, but is driven and optimized by the actual runtime workload characteristics. The sharding scheme is more scientific and reasonable, adapting to the actual business situation and automatically identifying and eliminating "hotspot shards," effectively solving the performance bottleneck problem caused by data access skew. By aggregating closely correlated data and separating loosely correlated data, the new load can be more evenly distributed across new shard nodes as the business grows, achieving a more near-linear horizontal scaling capability. At the same time, it avoids the situation where some nodes are idle while others are overloaded due to uneven data distribution, improving the overall resource utilization of the cluster.

[0023] In embodiments of this application, a distributed database management method is provided. The method involves determining the security characteristics of each data segment in a data shard, and analyzing and calculating the security of each data shard based on these security characteristics to obtain the security level of each data shard. This includes: determining the security characteristics of each data segment from the data information, wherein the security characteristics include sensitive data content and confidential data content; evaluating and assigning values ​​to the sensitive data content and confidential data content respectively to obtain sensitivity evaluation values ​​and confidentiality evaluation values, and determining a pre-set security level conversion coefficient; and calculating the security level of each data shard based on the sensitivity evaluation values, confidentiality evaluation values, and security level conversion coefficient.

[0024] Specifically, through data discovery and classification tools, the content of data shards is scanned to identify two key security characteristics: sensitive data content (the proportion of data protected by general laws and regulations in a shard) and confidential data content (the proportion of data in a shard that, if leaked, would cause significant business losses to the enterprise). These two contents are then standardized and assessed, calculating sensitivity and confidentiality assessment values ​​respectively. Pre-defined key coefficients are used, and a calculation model combines these assessment values ​​with the coefficients to obtain a quantified security level for each shard. This step, by quantifying security attributes and integrating them into the sharding logic, brings significant technical advantages and improved management efficiency. It changes the traditional "one-size-fits-all" security strategy (such as applying the highest level of encryption to the entire database), allowing differentiated security control measures based on the security level of each shard. This ensures that the cost of security investment is precisely matched with the value of data assets, avoiding resource waste. This step significantly reduces the risk of data leakage and compliance costs by centrally isolating high-security-level data in a few shards, greatly reducing the attack surface that requires focused protection. Limited auditing, monitoring, and protection resources can be focused on core high-risk shards, rather than being distributed across the entire database cluster, thus more effectively preventing data leakage risks.

[0025] In an embodiment of this application, a distributed database management method is provided, wherein the formula for calculating the security level of the data sharding is: , Where S is the security level of the data fragment, k is the security level conversion coefficient, X is the sensitivity assessment value of the data fragment, and n is the confidentiality assessment value of the data fragment.

[0026] In embodiments of this application, a distributed database management method is provided. The step of determining the storage node for each data shard based on the security level and storing the data shards in the corresponding storage nodes includes: pre-setting a preset security level-security level interval correspondence, wherein each security level interval is associated with a corresponding preset security level; obtaining the security level of each data shard, and based on the mapping relationship between the security level interval to which the security level belongs and the preset security level interval correspondence, selecting the preset security level corresponding to the security level interval as the security level of the storage node corresponding to each data shard; storing each data shard in the storage node corresponding to the security level, and storing backups of each data shard in other storage nodes of the same security level.

[0027] Specifically, a mapping table is predefined to establish a correspondence between "preset security level - security level range". This mapping table associates a preset security level with each security level range. The security level of each data shard calculated earlier is obtained. Then, based on the mapping relationship established in the first step, the security level range to which the value belongs is determined. The preset security level associated with the range is automatically selected, and this level is determined as the security admission standard for the target storage node of the data shard. Based on the preset security level determined for the data shard, all storage nodes that meet the security level are searched in the resource pool, and the data shard is stored on one of the nodes following strategies such as load balancing. To ensure that backup data enjoys the same level of protection as primary data, the backup strategy of this method explicitly stipulates that copies of data shards must be stored on other nodes with the same preset security level. This ensures consistent security protection throughout the data lifecycle and avoids the risk of data leakage caused by storing backup data on low-security nodes. This step transforms manually defined security policies (mapping tables) into rules that the system can automatically execute, completely eliminating the possibility of high-sensitivity data being mistakenly stored in a low-security environment due to human error or negligence. This ensures the rigid execution of security compliance requirements and provides irrefutable automated evidence for passing rigorous IT audits. This step establishes a defense-in-depth link between "data security level" and "infrastructure security capabilities." Highly sensitive data is "locked" in the highest-security "vault" (node). Even if attackers breach some peripheral defenses, they cannot access core assets by accessing low-security nodes. This significantly reduces the exposure surface of critical data, provides the strongest protection for core assets, and greatly increases the difficulty and cost for attackers to steal all valuable data. This step avoids the over-protection strategy of applying the "highest security level" to all data. Low-sensitivity data (such as public content) can be reasonably stored on lower-cost infrastructure, while expensive high-security resources (such as encryption hardware and physically isolated data centers) are precisely invested in protecting the data that truly needs them. This achieves refined operation of security investment and effectively reduces the total cost of ownership while meeting compliance requirements.

[0028] In an embodiment of this application, a distributed database management method is provided. The method for determining the load prediction data of each storage node includes: determining pre-set load indicators and prediction models corresponding to each load indicator, wherein the load indicators include CPU utilization, disk utilization, and memory utilization; performing prediction output based on the prediction models corresponding to each load indicator to obtain the prediction values ​​corresponding to each load indicator, and determining the prediction values ​​corresponding to each load indicator as the load prediction data of each storage node.

[0029] Specifically, a set of core load metrics that comprehensively reflect the health status and performance bottlenecks of storage nodes are pre-defined. These metrics typically include CPU utilization, reflecting the saturation of the node's computing power; memory utilization, reflecting the pressure on node caching and data exchange; and disk utilization, reflecting the capacity of the node's persistent storage and read / write throughput pressure. A predictive model is configured for each load metric. The model is trained based on the historical, time-series recorded metric data of each node to learn its load change patterns. Each model outputs a predicted value for its corresponding metric in a future time window. Combining these predicted values ​​constitutes a panoramic view of the future state of the storage node—the load prediction data.

[0030] In an embodiment of this application, a distributed database management method is provided. The method for evaluating the load status of storage nodes based on load prediction data to obtain a load status evaluation value for each storage node includes: determining a pre-set benchmark value corresponding to each load indicator, and calculating the difference between the predicted value and the benchmark value corresponding to each load indicator; evaluating these differences to obtain a difference estimate for each load indicator, and determining a preset weight for each load indicator; and calculating the load status evaluation value of the storage node based on the preset weight and difference estimate of each load indicator.

[0031] Specifically, a reasonable baseline value is set for each key load metric (CPU utilization, memory utilization, disk utilization), representing the expected value of that metric under ideal steady-state conditions. The predicted values ​​of each metric obtained from the aforementioned prediction model are subtracted from their respective baseline values ​​to obtain a series of differences. These differences intuitively reflect the "deviation" of future load from the safe baseline. The calculated raw differences are evaluated to ensure comparability between metrics of different dimensions. A preset weight is assigned to each load metric based on its importance to the overall system performance. A weighted summation model is used to calculate the final load status assessment value of the node. This step condenses multi-dimensional load information (CPU, memory, disk) into a single meaningful comprehensive score, greatly simplifying the decision-making logic. It eliminates the need to simultaneously examine multiple potentially contradictory metrics; a single score is sufficient to quickly and accurately determine the overall state of the node, improving decision-making efficiency and accuracy.

[0032] In an embodiment of this application, a distributed database management method is provided, wherein the formula for calculating the load status evaluation value of the storage node is: , Where L is the load status assessment value of the storage node, bi is the preset weight of the i-th load indicator, Pi is the differential estimate of the i-th load indicator, and n is the number of load indicators.

[0033] In an embodiment of this application, a distributed database management method is provided. The method involves determining a dynamic storage adjustment strategy for data shards based on a load status assessment value, and dynamically adjusting the storage distribution of data shards on storage nodes based on the dynamic storage adjustment strategy. The method includes: determining storage nodes whose load status assessment values ​​are greater than a preset threshold, and determining the security level corresponding to the storage node; migrating data shards in the storage node that cannot meet the load requirements to other storage nodes with the same security level, so as to dynamically adjust the storage distribution of data shards on storage nodes.

[0034] Specifically, the system continuously monitors the load status assessment values ​​of all storage nodes. When the assessment value of a node exceeds a preset threshold, the system marks it as a high-load node "to be mitigated." The system then confirms the security level of the data shards carried by the high-load node. The migration target is not to find the least idle node globally, but rather to select nodes with lower load status assessment values ​​and sufficient resources from all storage node clusters with the same security level as migration targets. From the high-load nodes, the system intelligently filters out the subset of data shards that contribute the main load and identifies them as migration targets to ensure migration efficiency. Finally, the system initiates the data migration process, safely migrating the selected data shards from the high-load source node to the target node. This step perfectly resolves the potential conflict between security strategies and performance optimization, ensuring that data compliance requirements are never compromised during any performance tuning operations. It provides technical assurance for enterprises to confidently use automated operation and maintenance tools in stringent compliance environments, avoiding security incidents and compliance risks caused by automatic scaling and migration. It narrows the load balancing decision-making scope from the "entire cluster" to a "node pool of the same security level," satisfying security requirements while still achieving effective load distribution within the security pool, avoiding ineffective scheduling. By dynamically distributing the load of high-security data across multiple equally secure nodes, it effectively prevents performance degradation of these critical businesses due to resource bottlenecks, providing continuous and stable high-performance assurance for core businesses, directly improving the end-user experience and the reliability of business systems.

[0035] like Figure 2As shown in the embodiments of this application, a distributed database management system is provided, including: a classification module, used to determine the association relationship between various data in the database, and classify the data based on the association relationship to obtain multiple data shards; a calculation module, used to determine the security characteristics of each data in the data shards, and analyze and calculate the security of the data shards based on the security characteristics of each data to obtain the security level of each data shard; a storage module, used to determine the storage nodes of each data shard based on the security level, and store the data shards in the corresponding storage nodes respectively; an evaluation module, used to determine the load prediction data of each storage node, and evaluate the load status of the storage nodes based on the load prediction data to obtain the load status evaluation value of each storage node; and an adjustment module, used to determine the dynamic storage adjustment strategy of the data shards based on the load status evaluation value, and dynamically adjust the storage distribution of the data shards on the storage nodes based on the dynamic storage adjustment strategy.

[0036] In summary, this invention provides a distributed database management method and system, comprising: determining the relationships between data in the database and classifying the data accordingly to obtain multiple data shards; determining the security characteristics of each data in the data shards and performing security analysis and calculation on the data shards to obtain a security level; determining the storage nodes for each data shard based on the security level and storing the data shards separately; determining the load prediction data for each storage node and evaluating the load status of the storage nodes based on the prediction data to obtain a load status evaluation value; determining a dynamic storage adjustment strategy for the data shards based on the load status evaluation value and dynamically adjusting the storage distribution of the data shards on the storage nodes based on the strategy. This invention pools storage resources by dynamically adjusting the storage distribution of data shards on storage nodes and endows them with load awareness and intelligent scheduling capabilities, enabling data distribution to dynamically and smoothly adapt to the rapidly changing business load.

[0037] Finally, it should be noted that those skilled in the art can obviously make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims and their equivalents, this invention also intends to include these modifications and variations.

[0038] The above description is merely one embodiment of the present invention, and should not be construed as limiting the scope of the invention. Any structural changes made based on the present invention, as long as they do not depart from the essence of the invention, should be considered as falling within the protection scope of the present invention and subject to its restrictions. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process and related descriptions of the platform described above can be referred to the corresponding processes in the foregoing platform embodiments, and will not be repeated here.

[0039] The term "comprising" or any other similar term is intended to cover non-exclusive inclusion, such that a process, platform, article, or device / platform that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to those processes, platforms, articles, or devices / platforms.

[0040] The technical solutions of the present invention have been described in conjunction with the accompanying drawings and further embodiments. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to closely related technical features, and the technical solutions resulting from such changes or substitutions will all fall within the scope of protection of the present invention.

[0041] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention.

Claims

1. A distributed database management method, characterized in that, include: Determine the relationships between the data in the database, and classify the data based on these relationships to obtain multiple data shards; Determine the security characteristics of each data in the data shard, and analyze and calculate the security of the data shard based on the security characteristics of each data to obtain the security level of each data shard; The storage nodes for each data shard are determined based on the security level, and the data shards are stored in their respective storage nodes. Determine the load prediction data for each storage node, and evaluate the load status of the storage nodes based on the load prediction data to obtain the load status evaluation value for each storage node. The dynamic storage adjustment strategy for data shards is determined based on the load status assessment value, and the storage distribution of data shards on storage nodes is dynamically adjusted based on the dynamic storage adjustment strategy.

2. A distributed database management method according to claim 1, characterized in that, The process involves obtaining the relationships between data in the database and classifying the data based on these relationships to obtain multiple data shards, including: Obtain data information from each data point in the database and construct a relationship graph based on the data information, where nodes represent data and edges represent the relationships between data points; Determine the connection query frequency and the number of common modifications between the two ends of each edge from the association relationship, and evaluate the connection query frequency and the number of common modifications respectively; The evaluation results are summed to obtain the correlation between the data connected at both ends of each edge. Data with a correlation greater than a preset threshold are grouped into the same data shard, resulting in multiple data shards.

3. A distributed database management method according to claim 2, characterized in that, The process involves determining the security characteristics of each data segment within the data shard, and analyzing and calculating the security of each data segment based on these characteristics to obtain the security level of each data segment, including: The security characteristics of each data segment are determined from the data information, including the content of sensitive data and the content of confidential data. The content of sensitive data and confidential data are evaluated and values ​​are obtained respectively to obtain sensitivity evaluation value and confidentiality evaluation value, and a pre-set security level conversion coefficient is determined; The security level of each data fragment is calculated based on its sensitivity and confidentiality assessment values ​​and security level conversion coefficient.

4. A distributed database management method according to claim 3, characterized in that, The formula for calculating the security level of the data fragmentation is: ， Where S is the security level of the data fragment, k is the security level conversion coefficient, X is the sensitivity assessment value of the data fragment, and n is the confidentiality assessment value of the data fragment.

5. A distributed database management method according to claim 3, characterized in that, The process of determining the storage nodes for each data shard based on security levels and storing the data shards in their respective storage nodes includes: A preset security level-security level range correspondence is set in advance. For each security level range, a corresponding preset security level is associated with it. Obtain the security level of each data shard, and based on the mapping relationship between the security level range to which the security level belongs and the preset security level-security level range correspondence, select the preset security level corresponding to the security level range and determine it as the security level of the storage node corresponding to each data shard. Each data shard is stored in a storage node with the corresponding security level, and backups of each data shard are stored in other storage nodes with the same security level.

6. A distributed database management method according to claim 5, characterized in that, The determination of load prediction data for each storage node includes: Determine the pre-defined load metrics and the corresponding prediction models for each load metric. The load metrics include CPU utilization, disk utilization, and memory utilization. Based on the prediction model corresponding to each load indicator, the prediction output is obtained to obtain the prediction value corresponding to each load indicator, and the prediction value corresponding to each load indicator is determined as the load prediction data of each storage node.

7. A distributed database management method according to claim 6, characterized in that, The load status assessment of storage nodes based on load prediction data, to obtain load status assessment values ​​for each storage node, includes: Determine the pre-set baseline value for each load indicator, and calculate the difference between the predicted value and the baseline value for each load indicator. These differences are evaluated and values ​​are obtained to obtain the estimated difference of each load index, and the preset weight of each load index is determined. The load status assessment value of the storage node is obtained by calculating based on the preset weights and differential estimates of each load indicator.

8. A distributed database management method according to claim 7, characterized in that, The formula for calculating the load status assessment value of the storage node is as follows: ， Where L is the load status assessment value of the storage node, bi is the preset weight of the i-th load indicator, Pi is the differential estimate of the i-th load indicator, and n is the number of load indicators.

9. A distributed database management method according to claim 7, characterized in that, The dynamic storage adjustment strategy for determining data shards based on load status assessment values, and the dynamic adjustment strategy for dynamically distributing data shards across storage nodes, includes: Identify storage nodes whose load status assessment value is greater than a preset threshold, and determine the security level corresponding to the storage node; Data shards that cannot meet the load requirements in this storage node are migrated to other storage nodes with the same security level to dynamically adjust the storage distribution of data shards on the storage nodes.

10. A distributed database management system, characterized in that, include: The classification module is used to determine the relationships between data in the database and classify the data based on these relationships to obtain multiple data fragments. The calculation module is used to determine the security characteristics of each data in the data shard, and to analyze and calculate the security of the data shard based on the security characteristics of each data, so as to obtain the security level of each data shard. The storage module is used to determine the storage node for each data shard based on the security level, and to store the data shards to the corresponding storage nodes respectively; The evaluation module is used to determine the load prediction data for each storage node and evaluate the load status of the storage nodes based on the load prediction data, thereby obtaining the load status evaluation value for each storage node. The adjustment module is used to determine the dynamic storage adjustment strategy for data shards based on the load status assessment value, and to dynamically adjust the storage distribution of data shards on storage nodes based on the dynamic storage adjustment strategy.

Images