Data storage management method and system based on solid state disk

Abstract

This invention relates to the field of data management technology and discloses a data storage management method and system based on solid-state drives (SSDs). The system includes a data acquisition and feature analysis module, an intelligent storage strategy decision-making module, and a physical resource scheduling and execution module. By constructing a data value heat value calculation model based on multi-dimensional data features, a continuously quantified data value heat value parameter is generated. Based on this parameter, the optimal target data storage strategy identifier is adaptively matched through a fuzzy inference system. This solves the problem that traditional methods rely on simple rules and cannot accurately reflect the multi-dimensional value of data, thus improving the accuracy and utilization efficiency of the overall resource allocation of the storage system. By incorporating the real-time health status, load, and wear level of the SSD into the decision factors and performing collaborative intelligent matching with the data storage strategy, a multi-objective decision-making algorithm is used to generate a physical block allocation scheme, extending the lifespan of the SSD and maintaining the long-term stability of system performance.

Description

Technical Field

[0001] This invention relates to the field of data management technology, specifically to a data storage management method and system based on solid-state drives (SSDs). Background Technology

[0002] Solid-state drives (SSDs), as high-performance data storage devices, have been widely used in various fields. With the rapid development of information technology, digital and network technologies are frequently used in daily life, scientific research, and corporate business, leading to an explosive growth in data storage demand. SSDs, with their flash memory-based storage technology, have advantages such as fast read and write speeds, high random access performance, low power consumption, and small size, making them one of the current research hotspots and mainstream application directions in the storage field.

[0003] Currently, in the process of data storage management based on solid-state drives, due to the complexity and dynamic changes in data access patterns, the data storage strategy mainly relies on simple access frequency or fixed time windows to distinguish between hot and cold data. It is impossible to quantify and dynamically evaluate the multi-dimensional value characteristics of data in real time. When the true value of data does not match the storage strategy, high-value data will not be able to obtain priority performance resources, while low-value data will occupy high-quality storage media for a long time, resulting in a decline in the overall cost-effectiveness of the storage system.

[0004] Therefore, a data storage management method and system based on solid-state drives (SSDs) are proposed to solve the above problems. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a data storage management method and system based on solid-state drives (SSDs), which solves the problem mentioned in the background that the mismatch between the true value of data and storage strategies leads to a decline in the overall cost-effectiveness of the storage system.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a data storage management method and system based on solid-state drives, wherein the method includes the following steps: S1. Collect the data feature parameters of the data to be stored and the storage resource status parameters of the solid-state drive, and generate the data feature set and the storage resource status set; S2. Based on the data feature set, perform data value heat value calculation and generate data value heat value parameters; S3. Based on the data value heat value parameters and the pre-built storage strategy knowledge base, perform adaptive storage strategy matching processing to generate a target data storage strategy identifier; S4. Based on the storage resource status set, the target data storage strategy identifier, and the pre-built physical block resource scheduling rule base, perform adaptive scheduling processing of solid-state drive physical block resources to generate a target physical block allocation scheme; S5. Based on the target data storage strategy identifier, perform pre-writing optimization processing on the data to be stored to generate an optimized data packet to be stored. S6. Write the optimized data packet to be stored into the solid-state drive according to the target physical block allocation scheme, and simultaneously perform wear leveling scheduling and secure erase triggering based on the data value heat value parameter. S7. Construct a data value heat value decay and shelf life prediction model, and based on the data value heat value parameter and the target data storage strategy identifier, initiate storage strategy adjustment suggestions and migration preprocessing for the data according to the shelf life predicted by the model.

[0007] Preferably, generating the data feature set and storage resource status set in step S1 includes the following steps: S11. Capture input / output requests sent by the host through the storage interface layer, extract logical block address, data size, operation type, access timestamp and source application identifier from the input / output requests, and generate a basic data feature parameter set; S12. Monitor the internal status of the solid-state drive controller, collect in real time the physical block wear count, number of available physical blocks, remaining capacity of each physical block type, historical read / write latency statistics and current operating temperature of the solid-state drive, and generate a raw parameter set of storage resource status. S13. Normalize and vectorize the basic data feature parameter set and the original parameter set of storage resource status to construct the data feature set and the storage resource status set respectively.

[0008] Preferably, the data value heat value calculation process in S2 includes the following steps: S21. Extract the data access frequency factor, data expected retention time factor, data association application priority factor, and data size weight factor from the data feature set; S22. Construct a data value heat value calculation model. The model is a composite function based on multi-factor weighted fusion and nonlinear activation. Its inputs are the data access frequency factor, the data expected retention time factor, the data association application priority factor, and the data size weight factor. S23. Input the analyzed factors into the data value heat value calculation model to calculate a quantified data value heat value parameter, wherein the level of the parameter value directly corresponds to the degree to which the data is judged as hot data, warm data, and cold data.

[0009] Preferably, the adaptive storage strategy matching process in S3 includes the following steps: S31. Pre-build the storage strategy knowledge base, which stores the mapping relationship between multiple standard data value hot value ranges and storage strategy identifiers. The storage strategy identifiers include high-performance caching strategy, balanced storage strategy, high-compression archiving strategy, and encrypted secure storage strategy. S32. Calculate the matching degree between the data value heat value parameter and all standard data value heat value ranges in the storage strategy knowledge base; S33. Based on the fuzzy inference system, select the storage strategy identifier mapped to the standard data value heat value range with the highest matching degree calculation value, and determine it as the target data storage strategy identifier.

[0010] Preferably, the adaptive scheduling of solid-state drive physical block resources in step S4 includes the following steps: S41. Pre-build the physical block resource scheduling rule base. The rule base defines the preferred physical block type by different target data storage strategy identifiers and the priority rules for physical block allocation under different sets of storage resource states. The physical block type includes single-level unit block, multi-level unit block and three-level unit block. S42. Based on the target data storage strategy identifier, retrieve the candidate physical block type set and its allocation constraints from the physical block resource scheduling rule base; S43. Based on the storage resource status set, evaluate the current availability, average wear level, and expected write performance of each type of physical block in the candidate physical block type set; S44. Based on the allocation constraints, availability assessment results, wear assessment results, and performance assessment results, an optimal physical block allocation sequence is generated through a multi-objective decision algorithm to form the target physical block allocation scheme.

[0011] Preferably, the pre-write optimization process in S5 includes the following steps: S51. Based on the target data storage strategy identifier, determine whether it is necessary to perform a compression operation on the data to be stored. If necessary, call the compression algorithm associated with the target data storage strategy identifier to compress the data to be stored in real time and generate a compressed data segment. S52. Based on the target data storage strategy identifier, determine whether it is necessary to perform duplicate data deletion operation on the data to be stored and the compressed data segment. If it is necessary, calculate its data fingerprint and compare it with the existing data fingerprint database. If there is a duplicate, only the reference pointer is stored; otherwise, the complete data is stored and the fingerprint database is updated. S53. Based on the target data storage strategy identifier, determine whether it is necessary to add metadata tags. If so, add metadata tags containing the data value heat value parameter, the target data storage strategy identifier, and the expected management strategy to the data to be stored, the compressed data segment, and the deduplication data. S54. The data after the above judgment and processing is encapsulated to form the optimized data packet to be stored with complete header information.

[0012] Preferably, the synchronous wear leveling scheduling process in S6 includes the following steps: S61. When writing the optimized data packet to be stored into the physical block specified by the target physical block allocation scheme, the wear count of the physical block is updated in real time. S62. Periodically scan the wear count of all physical blocks in the solid-state drive and calculate the global average wear level and wear difference. S63. When the wear difference is detected to exceed a preset threshold, start the background wear leveling task. S64. The wear leveling task selects valid data pages in high-wear physical blocks based on the current wear count, remaining lifetime prediction, and current load status of physical blocks, migrates them to low-wear physical blocks, and updates the logical-to-physical address mapping table.

[0013] Preferably, the synchronous execution of the secure erase triggering process in S6 includes the following steps: S65. Monitor the changes in the data value heat value parameter. When the heat value parameter of a specific data item is continuously lower than the cold data threshold for a preset time, trigger the security erasure assessment process. S66. The security erasure assessment process determines whether to perform security erasure on the specific data item based on the expected management strategy recorded in the metadata tag. S67. When it is determined that a secure erase is required, a secure erase instruction is sent to the physical block where the specific data item is located to perform multiple overwrite operations, and the corresponding physical storage space is released after completion.

[0014] Preferably, step S7, which involves initiating storage strategy adjustment suggestions and migration preprocessing for data, includes the following steps: S71. Construct a data value heat value decay and shelf life prediction model. The model is constructed based on the historical access pattern of the data, the life cycle status signal of the associated application, and the management strategy parameters input from the outside. S72. Using the data value calorific value decay and shelf life prediction model, process the data value calorific value parameters and the target data storage strategy identifier, and dynamically output the predicted shelf life of the data. S73. Compare the predicted shelf life with the preset shelf life threshold to identify data items whose shelf life is about to expire; S74. Generate a preprocessing instruction for the identified data item, which includes the adjusted storage policy identifier and the target migration physical block address.

[0015] Preferably, the system includes a data acquisition and feature analysis module, an intelligent storage strategy decision-making module, and a physical resource scheduling and execution module; The data acquisition and feature analysis module is used to receive host input / output requests and solid-state drive physical block status signals. It extracts logical block addresses, data sizes, operation types, access timestamps, and source application identifiers through the data feature acquisition unit. It collects physical block wear counts, available physical block counts, remaining capacity of each physical block type, historical read / write latency statistics, and current operating temperature through the hard drive status monitoring unit. Based on the collected data, the heat value calculation unit performs multi-factor fusion calculations and outputs data value heat value parameters and storage resource status sets. The intelligent storage strategy decision module receives the data value heat value parameter, provides a predefined mapping relationship between heat value range and storage strategy through the storage strategy knowledge base unit, and the strategy matching decision unit makes a matching decision based on the data value heat value parameter and the mapping relationship, outputs the target data storage strategy identifier, and the metadata management unit generates the corresponding metadata tag based on the identifier. The physical resource scheduling and execution module receives the storage resource status set, the target data storage policy identifier, and the metadata tag. The physical block resource scheduling unit makes decisions and outputs a target physical block allocation scheme based on a pre-built physical block resource scheduling rule base. The data pre-writing optimization unit compresses, deduplicates, and encapsulates the data based on the target data storage policy identifier. The data writing and wear leveling unit executes the data writing operation according to the target physical block allocation scheme and simultaneously performs wear leveling scheduling. The secure erasure management unit manages the triggering and execution of the secure erasure process based on the data value heat value parameter and the metadata tag.

[0016] Compared with existing technologies, the present invention provides a data storage management method and system based on solid-state drives, which has the following beneficial effects: 1. In this invention, when managing solid-state drive data storage, a data value heat value calculation model based on multi-dimensional data characteristics is constructed to generate continuously quantified data value heat value parameters. Based on these parameters, the optimal target data storage strategy identifier is adaptively matched through a fuzzy inference system. This solves the problem that traditional methods rely on simple rules and cannot accurately reflect the multi-dimensional value of data, ensuring that high-value data can obtain matching high-performance storage resources, thereby improving the accuracy and utilization efficiency of the overall resource allocation of the storage system.

[0017] 2. In this invention, when allocating physical storage resources, the real-time health status, load, and wear level of the solid-state drive are incorporated into the decision factors and intelligently matched with the data storage strategy. A multi-objective decision algorithm is used to generate a physical block allocation scheme, which solves the problem of the disconnect between resource allocation and physical media status. This allows the system to automatically avoid high-wear blocks and balance the write load while meeting performance requirements, thereby extending the lifespan of the solid-state drive and maintaining the long-term stability of system performance.

[0018] 3. In this invention, during the data lifecycle and security management process, a predictive model for data value decay and shelf life is constructed to achieve forward-looking judgment of data shelf life and trigger proactive storage strategy adjustments and data migration preprocessing. At the same time, a security erasure process is established based on the expected management strategy in the metadata tag and the security level configuration of the data source application for dynamic evaluation. This solves the problems of passive and lagging management and coarse security strategies in traditional systems, enabling the system to reclaim resources in advance, smooth performance fluctuations, and implement differentiated security erasure based on data sensitivity. This achieves a balance between efficient use of storage space, system performance optimization, and data security. Attached Figure Description

[0019] Figure 1 This is a flowchart of the solid-state drive-based data storage management method of the present invention; Figure 2 This is an architecture diagram of the solid-state drive-based data storage management system of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] For specific implementation examples, please refer to: Figure 1-2 A data storage management method and system based on solid-state drives (SSDs), the method including the following steps: S1. Collect the data feature parameters of the data to be stored and the storage resource status parameters of the solid-state drive, and generate the data feature set and the storage resource status set; S2. Calculate and process the data value heat value based on the data feature set to generate data value heat value parameters; S3. Based on the data value heat value parameters and the pre-built storage strategy knowledge base, perform adaptive storage strategy matching processing to generate the target data storage strategy identifier; S4. Based on the storage resource status set, target data storage policy identifier and pre-built physical block resource scheduling rule base, perform adaptive scheduling of solid-state drive physical block resources to generate target physical block allocation scheme; S5. Based on the target data storage strategy identifier, perform pre-write optimization processing on the data to be stored, and generate an optimized data packet to be stored. S6. Write the optimized data packets to be stored to the solid-state drive according to the target physical block allocation scheme, and synchronously perform wear leveling scheduling and secure erase triggering based on the data value heat value parameter. S7. Construct a data value heat value decay and shelf life prediction model, and based on the data value heat value parameters and target data storage strategy identifier, initiate storage strategy adjustment suggestions and migration preprocessing for the data according to the shelf life predicted by the model.

[0022] The steps involved in generating the data feature set and storage resource status set in S1 are as follows: S11. Capture input / output requests sent by the host through the storage interface layer, extract logical block address, data size, operation type, access timestamp and source application identifier from the input / output requests, and generate a basic data feature parameter set; S12. Monitor the internal status of the solid-state drive controller, collect in real time the physical block wear count, number of available physical blocks, remaining capacity of each physical block type, historical read / write latency statistics and current operating temperature of the solid-state drive, and generate a raw parameter set of storage resource status. S13. Normalize and vectorize the basic data feature parameter set and the original parameter set of storage resource status to construct the data feature set and the storage resource status set respectively.

[0023] The data value heat value calculation and processing in S2 includes the following steps: S21. Extract the data access frequency factor, data expected retention time factor, data association application priority factor, and data size weight factor from the data feature set; S22. Construct a data value heat value calculation model. The model is a composite function based on multi-factor weighted fusion and nonlinear activation. Its inputs are data access frequency factor, data expected retention time factor, data association application priority factor, and data size weight factor, specifically including: S221 is the data access frequency factor. Expected data retention time factor Data association application priority factor and data size weighting factor Assign preset weight coefficients respectively The initial fusion value is obtained by summing the products of each factor and its corresponding weight coefficient. The specific calculation is as follows: ; in, The weighting coefficients are preset and satisfy the following conditions: ; S222, Initial fusion value The input is fed into a pre-defined non-linear activation function, which is the Sigmoid function, and its mathematical expression is: ; in, For the output of the activation function, It is a natural constant. To adjust the slope parameter of the function curve, The offset parameter determines the center point of the function; S223. Output the value after mapping by the non-linear activation function. As a parameter of data value heat value ,Right now ; S23. Input the analyzed factors into the data value heat value calculation model to calculate a quantitative data value heat value parameter, where the level of this parameter directly corresponds to the degree to which the data is judged as hot data, warm data, and cold data.

[0024] The adaptive storage strategy matching process in S3 includes the following steps: S31. Pre-build a storage strategy knowledge base. The knowledge base stores the mapping relationship between multiple standard data value hot value ranges and storage strategy identifiers. The storage strategy identifiers include high-performance caching strategies, balanced storage strategies, high-compression archiving strategies, and encrypted secure storage strategies, including the following steps: S331 is the data value heat value parameter. Define fuzzy linguistic variables, whose values ​​include "low", "medium", and "high"; S332. Calculate the data value heat value parameter using the triangular membership function The membership degrees with respect to the three linguistic variables of "low, medium, high" , , , where the triangular membership function for "medium" is defined as: ; Among them, is the data value heat value parameter The membership degree with respect to the fuzzy linguistic variable of "medium", , , are preset parameters that satisfy , and respectively define the left boundary, vertex, and right boundary of the triangle of the membership function of this linguistic variable; S333. Match the calculated membership degree with the predefined fuzzy rule base, and the fuzzy rule base defines the mapping relationship between different combinations of linguistic variables and storage strategy preferences; S334. Use the centroid method to defuzzify the result of fuzzy reasoning to obtain an accurate matching degree calculation value , which identifies the standard data value heat value interval that best matches the data value heat value parameter, and its calculation formula is: ; Among them, is the matching degree calculation value, is the number of activated fuzzy rules, is the activation strength of the th rule, is the central value of the standard data value heat value interval corresponding to the conclusion of the th rule, is the data value heat value parameter; S32. Calculate the matching degree between the data value heat value parameter and all standard data value heat value intervals in the storage strategy knowledge base, and the calculation formula is: ; Among them, is the data value heat value parameter The matching degree calculation value with the rd standard data value heat value interval in the storage strategy knowledge base, is the data value heat value parameter, is the midpoint value of the th standard data value heat value interval in the storage strategy knowledge base; S33. Based on the fuzzy inference system, select the storage strategy identifier mapped to the standard data value heat value range with the highest matching degree calculation value, and determine it as the target data storage strategy identifier.

[0025] The adaptive scheduling of solid-state drive physical block resources in S4 includes the following steps: S41. A pre-built physical block resource scheduling rule base is constructed. The rule base defines the physical block type identified by different target data storage strategies, as well as the priority rules for physical block allocation under different storage resource state sets. The physical block types include single-level unit blocks, multi-level unit blocks, and three-level unit blocks. S42. Based on the target data storage strategy identifier, retrieve the candidate physical block type set and its allocation constraints from the physical block resource scheduling rule base; S43. Based on the storage resource state set, evaluate the current availability, average wear level, and expected write performance of each type of physical block in the candidate physical block type set, including the following steps: S431. Obtain the preset weight coefficients for the current availability assessment metric, average wear assessment metric, and expected write performance assessment metric from the pre-built physical block resource scheduling rule base. ,in ,and ; S432. Normalize the results of the three different evaluation metrics: current availability, average wear level, and expected write performance. For current availability assessment metrics Compared with expected write performance metrics Its normalized value , The calculation formula is: ; ; in, and The first The original evaluation value of the physical block. and These represent the maximum and minimum current availability values ​​for all candidate physical block types, respectively. and These represent the maximum and minimum expected write performance values ​​for all candidate physical block types, respectively. For the average wear assessment index Its normalized value The calculation formula is: ; in, For the first The original average wear assessment value of the physical block. and These represent the maximum and minimum average wear levels among all candidate physical block types; S433. Based on the weighting coefficients and normalized index values, calculate the comprehensive evaluation score for each type of candidate physical block. The calculation formula is as follows: ; in, For the first The overall evaluation score of candidate physical block types, For the first Current availability assessment metrics for physical blocks The value after normalization For the first Physical block expected write performance evaluation metrics The value after normalization For the first Average wear assessment value of physical blocks The value after normalization The weighting coefficients preset for the current availability assessment metrics. The pre-defined weighting coefficients for the expected write performance evaluation metrics. The pre-set weighting coefficients for the average wear assessment index; Overall assessment score This is used to quantify the overall suitability of this type of physical block under the weight allocation logic, providing input for subsequent multi-objective decision-making; S44. Based on the allocation constraints, availability assessment results, wear assessment results, and performance assessment results, an optimal physical block allocation sequence is generated using a multi-objective decision algorithm to form the target physical block allocation scheme. This specifically includes the following steps: S441. Construct a decision space with the physical block allocation scheme satisfying all allocation constraints as a hard constraint; S442. Within the decision space, maximize the "current availability" of the physical block. Compared to "expected write performance" The weighted sum is used to minimize the "average wear" of the physical blocks. To achieve the objective, construct a multi-objective optimization function. Its vector form is: ; in, For the vector representation of a multi-objective optimization function, This represents a candidate physical block allocation scheme. Indicates transpose. , , These represent the "current availability" of the physical blocks in the optimization objective. "Expected write performance" And "average wear and tear" The negative value, that is , , ; S443. Using a linear weighting method, assign preset weights to each sub-objective in the multi-objective optimization function. This is transformed into a single-objective optimization problem, whose scalarization function... for: ; in, These are preset positive weighting coefficients. These are the normalized values ​​of the corresponding indicators; S444. Within the decision space, find the single-objective function value after transformation. Optimal physical block allocation scheme ,Right now It then outputs the target physical block allocation scheme.

[0026] Pre-write optimization in S5 includes the following steps: S51. Based on the target data storage strategy identifier, determine whether compression is required for the data to be stored. If so, call the compression algorithm associated with the target data storage strategy identifier to compress the data to be stored in real time, generating compressed data segments, specifically including: S511. Establish a mapping table between storage strategy identifiers and compression algorithms, where high-performance caching strategies are associated with lossless fast compression algorithms, balanced storage strategies are associated with standard compression algorithms, and high-compression archiving strategies are associated with high-ratio compression algorithms. S512. Based on the target data storage strategy identifier, query the mapping table to determine the corresponding target compression algorithm; S513. Call the target compression algorithm to compress the data to be stored, and add a decompression identifier to the header of the compressed data segment to identify the target compression algorithm; S52. Based on the target data storage strategy identifier, determine whether it is necessary to perform deduplication on the data to be stored and the compressed data segment. If so, calculate its data fingerprint and compare it with the existing data fingerprint database. If duplicates are found, only the reference pointer is stored; otherwise, the complete data is stored and the fingerprint database is updated. Specifically, this includes: S521, Data to be stored and compressed data segments A secure hash algorithm is applied to calculate a fixed-length hash value, which serves as the data fingerprint. The calculation process is expressed as follows: ; in, For the calculated data fingerprint, For input data, This represents the 256-bit version of the secure hash algorithm, which outputs a fixed-length hash value of 256 bits. S522. Compare the calculated data fingerprint with all fingerprints stored in the existing data fingerprint database one by one; S523. If identical fingerprints are found during comparison, it is determined that there is duplicate data; if no identical fingerprints are found, it is determined that the data is unique. S53. Based on the target data storage strategy identifier, determine whether it is necessary to add metadata tags. If so, add metadata tags containing data value heat value parameters, target data storage strategy identifier and expected management strategy to the data to be stored, the compressed data segment and the deduplication data. S54. The data after the above judgment and processing is encapsulated to form an optimized data packet to be stored with complete header information.

[0027] Synchronous wear leveling scheduling in S6 includes the following steps: S61. When writing the optimized data packet to be stored into the physical block specified by the target physical block allocation scheme, update the wear count of the physical block in real time. S62. Periodically scan the wear count of all physical blocks within the solid-state drive and calculate the global average wear level and wear variation, including the following steps: S621, Get the total number of physical blocks in the solid-state drive. and read each physical block Current wear count ; S622, Calculate the global average wear level Its calculation formula is the arithmetic mean of the wear counts of all physical blocks: ; in, This represents the total number of physical blocks within the solid-state drive. For the first Wear count of each physical block This represents the summation operation; S623, Calculate wear difference The calculation formula is as follows: ; in, For wear difference, This represents the total number of physical blocks within the solid-state drive. This represents the global average wear level. This value is used to quantify the dispersion of wear on each physical block; the larger the value, the more uneven the wear. S63. When the wear difference is detected to exceed the preset threshold, start the background wear leveling task; S64. The wear leveling task, based on the current wear count, remaining lifetime prediction, and current load status of physical blocks, selects valid data pages from high-wear physical blocks, migrates them to low-wear physical blocks, and updates the logical-to-physical address mapping table. This is achieved through the following steps: S641. Obtain the maximum number of programming / erase cycles that the physical block can withstand. ; S642, Read the current wear count of the physical block. This refers to the number of programming / erase cycles already used. S643. Calculate the ratio of the current wear count to the maximum tolerable programming / erase cycles as the current wear rate. : ; S644, Based on current wear rate and wear count in recent The historical growth trend within a given period is used to predict, through linear extrapolation, whether the physical block will reach a preset scrap threshold. Required remaining programming / erase cycles As a predicted remaining useful life value, its calculation formula is: ; in, This is the average rate of change of wear rate per unit time, calculated based on historical data.

[0028] The synchronous execution of secure erase triggering in S6 includes the following steps: S65. Monitor changes in the heat value parameter of the data. When the heat value parameter of a specific data item is continuously lower than the cold data threshold for a preset time, trigger the security erasure assessment process. S66. The security erasure assessment process, based on the expected management policy recorded in the metadata tag, determines whether to perform security erasure on a specific data item, including the following steps: S661. Read the "Expected Management Strategy" field recorded in the metadata tag; S662. If the "Expected Management Policy" field contains the "Secure Deletion" flag, it is determined that secure erasure needs to be performed. S663. If the "Expected Management Policy" field contains the identifiers "Regular Deletion" and "Overwrite", then further examine the security level configuration of the application from which the data originates. ; S664, when the security level is configured Higher than the preset threshold If the threshold is not met, a secure erase operation is required. This is determined by reading the security level configurations of all applications in the system. maximum value And calculate: ; in, The preset scaling factor, and ; S67. When it is determined that a secure erase is required, a secure erase instruction is sent to the physical block containing the specific data item to perform multiple overwrite operations, and the corresponding physical storage space is released after completion.

[0029] Step S7, which involves proposing storage strategy adjustments and migration preprocessing for the data, includes the following steps: S71. Construct a data value heat value decay and shelf life prediction model. The model is constructed based on the historical access patterns of the data, the life cycle status signals of related applications, and externally input management strategy parameters. Specific steps include: S711. Historical access patterns of collected data, including access time interval sequences. Compared with the recent decline in access frequency ,in Represents the first in sequence I Each access time interval This represents the total number of intervals contained in the access time interval sequence I; S712, Listen for lifecycle status signals of associated applications. Signals include application activity, application suspension, and application termination; S713, Receive management policy parameters from external input. The parameters include the mandatory data retention period; S714. Establish a system that takes historical access patterns, lifecycle status signals, and management policy parameters as inputs, and data freshness period as the data freshness period. The output time series prediction model is trained using a multiple linear regression analysis method based on historical data. Its core relationship can be expressed as: ; in, For the intercept term, These are the regression coefficients of the model. and These are the access time interval sequences respectively. and lifecycle status signals Quantization feature extraction function, For random error term, These are management strategy parameters input from external sources. S72. Using a data value heat value decay and shelf life prediction model, process the data value heat value parameters and target data storage strategy identifier, and dynamically output the predicted shelf life of the data. S73. Compare the predicted shelf life with the preset shelf life threshold to identify data items whose shelf life is about to expire; S74. Generate a preprocessing instruction for the identified data item, which includes the adjusted storage policy identifier and the target migration physical block address.

[0030] The system includes a data acquisition and feature analysis module, an intelligent storage strategy decision-making module, and a physical resource scheduling and execution module; The data acquisition and feature analysis module is used to receive host input / output requests and solid-state drive physical block status signals. It extracts logical block addresses, data sizes, operation types, access timestamps, and source application identifiers through the data feature acquisition unit. It collects physical block wear counts, available physical blocks, remaining capacity of each physical block type, historical read / write latency statistics, and current operating temperature through the hard drive status monitoring unit. The heat value calculation unit performs multi-factor fusion calculations based on the collected data and outputs data value heat value parameters and storage resource status set. The intelligent storage strategy decision module receives data value heat value parameters, provides a predefined mapping relationship between heat value ranges and storage strategies through the storage strategy knowledge base unit, and the strategy matching decision unit makes matching decisions based on the data value heat value parameters and mapping relationship, outputs the target data storage strategy identifier, and the metadata management unit generates the corresponding metadata tag based on the identifier; The physical resource scheduling and execution module receives a set of storage resource statuses, target data storage policy identifiers, and metadata tags. The physical block resource scheduling unit makes decisions based on a pre-built physical block resource scheduling rule base and outputs a target physical block allocation scheme. The data pre-writing optimization unit compresses, deduplicates, and encapsulates the data based on the target data storage policy identifiers. The data writing and wear leveling unit executes data writing operations according to the target physical block allocation scheme and simultaneously performs wear leveling scheduling. The secure erasure management unit manages the triggering and execution of the secure erasure process based on data value heat value parameters and metadata tags.

[0031] The data storage management method and system operation steps based on solid-state drives are as follows: Step 1: Data Acquisition and Status Monitoring This method first collects the data feature parameters of the data to be stored and the storage resource status parameters of the solid-state drive (SSD), generating corresponding data feature sets and storage resource status sets. Specifically, it involves capturing input / output requests from the host through the storage interface layer, extracting logical block addresses, data sizes, operation types, access timestamps, and source application identifiers to form a basic data feature parameter set. Simultaneously, it monitors the internal status of the SSD controller, collecting real-time physical block wear counts, the number of available physical blocks, the remaining capacity of each physical block type, historical read / write latency statistics, and the current operating temperature to generate a raw storage resource status parameter set. Finally, these two parameter sets are normalized and vectorized to construct standardized data feature sets and storage resource status sets, respectively.

[0032] Step 2: Calculation of Data Value Heat Value: Based on the generated data feature set, a data value heat value calculation is performed to quantify the activity and importance of the data. First, data access frequency factors, expected data retention time factors, data association application priority factors, and data size weighting factors are extracted from the data feature set. Then, these factors are input into a pre-defined data value heat value calculation model. This model is a composite function based on multi-factor weighted fusion and nonlinear activation. An initial fusion value is obtained by calculating the sum of the products of each factor and its weight coefficient. This is then mapped by a nonlinear function to finally output a quantitative data value heat value parameter within a fixed range. The level of this parameter directly corresponds to the degree to which the data is classified as hot, warm, or cold data.

[0033] Step 3: Adaptive storage strategy matching: Based on the calculated data value heat value parameters, a pre-built storage strategy knowledge base is matched to determine the most suitable management strategy. The storage strategy knowledge base stores mappings between multiple standard data value heat value intervals and different storage strategy identifiers. The system calculates the matching degree between the data value heat value parameters and all standard intervals in the knowledge base, and based on a fuzzy inference system, selects the storage strategy identifier mapped to the standard interval with the highest matching degree, and determines it as the target data storage strategy identifier, serving as the guiding principle for all subsequent operations.

[0034] Step 4: Intelligent scheduling of physical block resources: Based on the current set of storage resource states, the identified target data storage strategy identifiers, and a pre-built physical block resource scheduling rule base, adaptive scheduling of SSD physical block resources is performed. First, based on the target data storage strategy identifiers, a set of candidate physical block types and their allocation constraints are retrieved from the rule base. Next, combined with the storage resource state set, the current availability, average wear level, and expected write performance of the candidate physical block types are evaluated. Finally, based on the allocation constraints and the evaluation results, a multi-objective decision algorithm is used for optimization calculations to generate a physical block allocation sequence with optimal overall performance, forming the final target physical block allocation scheme.

[0035] Step 5: Data optimization before writing: Before actually writing data to the solid-state drive, a series of optimization processes are performed based on the target data storage policy identifier. These processes include selective compression, deduplication, and adding metadata tags. The system determines and invokes the associated compression algorithm to compress the data in real time; it calculates the data fingerprint and compares it with an existing fingerprint database to achieve deduplication; and it adds metadata tags containing data value parameters and target storage policy identifier information to the data. After completing the above determination and processing, all data is encapsulated to form an optimized data package to be stored.

[0036] Step Six: Data Writing and Backend Management Based on the generated target physical block allocation scheme, the optimized data packets to be stored are written to the designated physical blocks of the solid-state drive. During this process, the system simultaneously executes two key background management tasks based on the data value heat value parameter. The first is wear leveling scheduling: real-time updates of block wear counts, periodic calculation of the global average wear level and wear difference, and when the difference is too large, initiating a background task to migrate valid data from high-wear blocks to low-wear blocks. The second is secure erase triggering: monitoring the data value heat value, and for data that consistently falls below the threshold, triggering a secure erase process to safely release storage space based on the expected management strategy and security level assessment in its metadata tag.

[0037] Step 7: Value Decline Prediction and Forward-Looking Adjustments: To enable longer-term optimization, the system constructs a data value decay and shelf-life prediction model. This model takes historical data access patterns, lifecycle status signals of associated applications, and external management strategy parameters as inputs, and is trained using multiple linear regression analysis to dynamically predict data shelf-life. Using this model, combined with the data's current value parameters and storage policy identifiers, the system identifies data items whose shelf-life is about to expire and generates preprocessing instructions for these data items, including adjusted storage policy identifiers and target migration physical block addresses. This allows for proactive and forward-looking adjustments to storage policies and physical resources.

[0038] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0039] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A data storage management method based on solid-state drives, characterized in that: The method includes the following steps: S1. Collect the data feature parameters of the data to be stored and the storage resource status parameters of the solid-state drive, and generate the data feature set and the storage resource status set; S2. Based on the data feature set, perform data value heat value calculation and generate data value heat value parameters; S3. Based on the data value heat value parameters and the pre-built storage strategy knowledge base, perform adaptive storage strategy matching processing to generate a target data storage strategy identifier; S4. Based on the storage resource status set, the target data storage strategy identifier, and the pre-built physical block resource scheduling rule base, perform adaptive scheduling processing of solid-state drive physical block resources to generate a target physical block allocation scheme; S5. Based on the target data storage strategy identifier, perform pre-writing optimization processing on the data to be stored to generate an optimized data packet to be stored. S6. Write the optimized data packet to be stored into the solid-state drive according to the target physical block allocation scheme, and simultaneously perform wear leveling scheduling and secure erase triggering based on the data value heat value parameter. S7. Construct a data value heat value decay and shelf life prediction model, and based on the data value heat value parameter and the target data storage strategy identifier, initiate storage strategy adjustment suggestions and migration preprocessing for the data according to the shelf life predicted by the model.

2. The solid state drive based data storage management method of claim 1, wherein: The steps involved in generating the data feature set and storage resource status set in step S1 are as follows: S11. Capture input / output requests sent by the host through the storage interface layer, extract logical block address, data size, operation type, access timestamp and source application identifier from the input / output requests, and generate a basic data feature parameter set; S12. Monitor the internal status of the solid-state drive controller, collect in real time the physical block wear count, number of available physical blocks, remaining capacity of each physical block type, historical read / write latency statistics and current operating temperature of the solid-state drive, and generate a raw parameter set of storage resource status. S13. Normalize and vectorize the basic data feature parameter set and the original parameter set of storage resource status to construct the data feature set and the storage resource status set respectively.

3. The solid state drive based data storage management method of claim 2, wherein: The data value heat value calculation process in S2 includes the following steps: S21. Extract the data access frequency factor, data expected retention time factor, data association application priority factor, and data size weight factor from the data feature set; S22. Construct a data value heat value calculation model. The model is a composite function based on multi-factor weighted fusion and nonlinear activation. Its inputs are the data access frequency factor, the data expected retention time factor, the data association application priority factor, and the data size weight factor. S23. Input the analyzed factors into the data value heat value calculation model to calculate a quantified data value heat value parameter, wherein the level of the parameter value directly corresponds to the degree to which the data is judged as hot data, warm data, and cold data.

4. The solid state drive based data storage management method of claim 3, wherein: The adaptive storage strategy matching process in S3 includes the following steps: S31. Pre-build the storage strategy knowledge base, which stores the mapping relationship between multiple standard data value hot value ranges and storage strategy identifiers. The storage strategy identifiers include high-performance caching strategy, balanced storage strategy, high-compression archiving strategy, and encrypted secure storage strategy. S32. Calculate the matching degree between the data value heat value parameter and all standard data value heat value ranges in the storage strategy knowledge base; S33. Based on the fuzzy inference system, select the storage strategy identifier mapped to the standard data value heat value range with the highest matching degree calculation value, and determine it as the target data storage strategy identifier.

5. The solid state drive based data storage management method of claim 4, wherein: The adaptive scheduling of solid-state drive physical block resources in step S4 includes the following steps: S41. Pre-build the physical block resource scheduling rule base. The rule base defines the preferred physical block type by different target data storage strategy identifiers and the priority rules for physical block allocation under different sets of storage resource states. The physical block type includes single-level unit block, multi-level unit block and three-level unit block. S42. Based on the target data storage strategy identifier, retrieve the candidate physical block type set and its allocation constraints from the physical block resource scheduling rule base; S43. Based on the storage resource status set, evaluate the current availability, average wear level, and expected write performance of each type of physical block in the candidate physical block type set; S44. Based on the allocation constraints, availability assessment results, wear assessment results, and performance assessment results, an optimal physical block allocation sequence is generated through a multi-objective decision algorithm to form the target physical block allocation scheme.

6. The solid state drive based data storage management method of claim 5, wherein: The pre-write optimization process in S5 includes the following steps: S51. Based on the target data storage strategy identifier, determine whether it is necessary to perform a compression operation on the data to be stored. If necessary, call the compression algorithm associated with the target data storage strategy identifier to compress the data to be stored in real time and generate a compressed data segment. S52. Based on the target data storage strategy identifier, determine whether it is necessary to perform duplicate data deletion operation on the data to be stored and the compressed data segment. If it is necessary, calculate its data fingerprint and compare it with the existing data fingerprint database. If there is a duplicate, only the reference pointer is stored; otherwise, the complete data is stored and the fingerprint database is updated. S53. Based on the target data storage strategy identifier, determine whether it is necessary to add metadata tags. If so, add metadata tags containing the data value heat value parameter, the target data storage strategy identifier, and the expected management strategy to the data to be stored, the compressed data segment, and the deduplication data. S54. The data after the above judgment and processing is encapsulated to form the optimized data packet to be stored with complete header information.

7. The data storage management method based on solid-state drives according to claim 6, characterized in that: The synchronous wear leveling scheduling process in S6 includes the following steps: S61. When writing the optimized data packet to be stored into the physical block specified by the target physical block allocation scheme, the wear count of the physical block is updated in real time. S62. Periodically scan the wear count of all physical blocks in the solid-state drive and calculate the global average wear level and wear difference. S63. When the wear difference is detected to exceed a preset threshold, start the background wear leveling task. S64. The wear leveling task selects valid data pages in high-wear physical blocks based on the current wear count, remaining lifetime prediction, and current load status of physical blocks, migrates them to low-wear physical blocks, and updates the logical-to-physical address mapping table.

8. The solid state drive based data storage management method of claim 7, wherein: The synchronous execution of the secure erase triggering process in S6 includes the following steps: S65. Monitor the changes in the data value heat value parameter. When the heat value parameter of a specific data item is continuously lower than the cold data threshold for a preset time, trigger the security erasure assessment process. S66. The security erasure assessment process determines whether to perform security erasure on the specific data item based on the expected management strategy recorded in the metadata tag. S67. When it is determined that a secure erase is required, a secure erase instruction is sent to the physical block where the specific data item is located to perform multiple overwrite operations, and the corresponding physical storage space is released after completion.

9. The solid state drive based data storage management method of claim 8, wherein: Step S7, which involves initiating storage strategy adjustment suggestions and migration preprocessing for data, includes the following steps: S71. Construct a data value heat value decay and shelf life prediction model. The model is constructed based on the historical access pattern of the data, the life cycle status signal of the associated application, and the management strategy parameters input from the outside. S72. Using the data value calorific value decay and shelf life prediction model, process the data value calorific value parameters and the target data storage strategy identifier, and dynamically output the predicted shelf life of the data. S73. Compare the predicted shelf life with the preset shelf life threshold to identify data items whose shelf life is about to expire; S74. Generate a preprocessing instruction for the identified data item, which includes the adjusted storage policy identifier and the target migration physical block address.

10. A data storage management system based on a solid state disk for implementing the data storage management method based on a solid state disk according to any one of claims 1 to 9, characterized in that, The system includes a data acquisition and feature analysis module, an intelligent storage strategy decision-making module, and a physical resource scheduling and execution module. The data acquisition and feature analysis module is used to receive host input / output requests and solid-state drive physical block status signals. It extracts logical block addresses, data sizes, operation types, access timestamps, and source application identifiers through the data feature acquisition unit. It collects physical block wear counts, available physical block counts, remaining capacity of each physical block type, historical read / write latency statistics, and current operating temperature through the hard drive status monitoring unit. Based on the collected data, the heat value calculation unit performs multi-factor fusion calculations and outputs data value heat value parameters and storage resource status sets. The intelligent storage strategy decision module receives the data value heat value parameter, provides a predefined mapping relationship between heat value range and storage strategy through the storage strategy knowledge base unit, and the strategy matching decision unit makes a matching decision based on the data value heat value parameter and the mapping relationship, outputs the target data storage strategy identifier, and the metadata management unit generates the corresponding metadata tag based on the identifier. The physical resource scheduling and execution module receives the storage resource status set, the target data storage policy identifier, and the metadata tag. The physical block resource scheduling unit makes decisions and outputs a target physical block allocation scheme based on a pre-built physical block resource scheduling rule base. The data pre-writing optimization unit compresses, deduplicates, and encapsulates the data based on the target data storage policy identifier. The data writing and wear leveling unit executes the data writing operation according to the target physical block allocation scheme and simultaneously performs wear leveling scheduling. The secure erasure management unit manages the triggering and execution of the secure erasure process based on the data value heat value parameter and the metadata tag.

Images