A cloud data processing and isolation method based on multi-tenant architecture

Abstract

The present invention provides a cloud data processing and isolation method based on a multi-tenant architecture, and relates to the technical field of cloud data processing and isolation. The method includes: S1, establishing a data classification system, adding classification labels to the data in the cloud database according to the first feature of the data; S2, establishing an importance evaluation system, judging the importance level of the data according to the second feature of the data, and using the judgment result as the importance label of the data; S3, setting a hierarchical encryption strategy; S4, creating a separate indexing mechanism; S5, formulating a differentiated access control strategy; S6, establishing a monitoring mechanism. The present invention makes full use of data characteristics and importance for refined management, while ensuring data security and isolation, it also improves data management efficiency, business responsiveness and multi-tenant management flexibility, and ultimately reduces the overall data management cost.

Description

Technical Field

[0001] The present invention relates to the field of cloud data processing and isolation technology, and in particular to a cloud data processing and isolation method based on a multi-tenant architecture. Background Art

[0002] With the rapid development and widespread application of cloud computing technology, enterprises have an increasing demand for data storage and management. When facing large-scale users and diversified application scenarios, the traditional single-tenant database architecture has problems such as low resource utilization, high cost, and complex maintenance. In this case, the multi-tenant architecture came into being. It aims to provide services to multiple tenants, users, or organizations by sharing database resources, while achieving data isolation and security. On the one hand, enterprises hope to reduce the cost and complexity of database management, and be able to flexibly expand or shrink resources according to business needs to improve resource utilization efficiency. The multi-tenant architecture allows multiple tenants to share hardware, software, and operation and maintenance costs, thereby reducing the cost of use for a single tenant. On the other hand, with the popularization of Internet applications, the number of users and the amount of data are growing rapidly, which puts higher requirements on the concurrent processing capability and response speed of the database. The multi-tenant architecture can better cope with high concurrent access and improve the performance and scalability of the database by optimizing resource allocation and data storage methods.

[0003] In cloud database management, it is impossible to adopt differentiated storage, computing and other optimization strategies for data of different importance. This leads to serious waste of computing, storage and other resources, and increases overall operating costs. The lack of differentiated management strategies for data of different types and importance also limits the efficiency and flexibility of data management. Summary of the invention

[0004] The present invention provides a cloud data processing and isolation method based on a multi-tenant architecture, which is used to solve the defect of the prior art that there is a lack of differentiated management strategies for data of different types and importance.

[0005] The present invention provides a cloud data processing and isolation method based on a multi-tenant architecture, which is used to ensure that data of different tenants can be securely isolated and effectively managed in a shared cloud database. The method includes:

[0006] S1. Establish a data classification system and add classification labels to the data in the cloud database based on the first characteristics of the data. The first characteristics include data format, data source, data field and data structure;

[0007] S2. Establish an importance assessment system to determine the importance level of data based on the second feature of the data, and use the determination result as the importance label of the data. The second feature includes data access frequency index, data update frequency index, data sensitivity index, data value assessment index, and data-business process correlation index.

[0008] S3. Set up a hierarchical encryption strategy and adopt corresponding encryption methods for data of different importance levels;

[0009] S4. Create a separate indexing mechanism based on the first feature;

[0010] S5. Develop differentiated access control strategies based on the type and importance of data;

[0011] S6. Establish a monitoring mechanism to realize global abnormal behavior monitoring and local abnormal behavior monitoring of cloud databases.

[0012] According to a cloud data processing and isolation method based on a multi-tenant architecture provided by the present invention, in step S1, the process of establishing a data classification system includes:

[0013] S11, standardize each data in the cloud database, extract the first feature of each data, and construct a data feature vector;

[0014] S12, construct a classification model based on support vector machine, input feature vector, and obtain the type score of each data;

[0015] S13. Set a type score threshold, and add the same classification label to data with type scores within the same threshold range.

[0016] According to a cloud data processing and isolation method based on a multi-tenant architecture provided by the present invention, the process of building a classification model based on a support vector machine includes:

[0017] Collect the first features of historical data in the cloud database, construct feature vectors, and label each data sample with the correct classification label, dividing it into training set and test set;

[0018] Select Kernel Support Vector Machine to build the base model;

[0019] Train the basic model on the training set and adjust the kernel function;

[0020] Use the test set to evaluate the trained basic model and calculate the accuracy;

[0021] The basic model parameters with the highest accuracy are retained to obtain the classification model.

[0022] According to a cloud data processing and isolation method based on a multi-tenant architecture provided by the present invention, in step S2, the process of establishing an importance evaluation system includes:

[0023] S21, collecting the second feature of each data;

[0024] S22, calculating the weight of each indicator of the second feature;

[0025] S23, obtaining the indicator score of each indicator;

[0026] S24. Calculate the importance score of each data according to the indicator score and the corresponding weight;

[0027] S25. Setting a threshold range of importance scores corresponding to each importance level, where the importance levels include low importance, medium importance, and high importance;

[0028] S26. Determine the importance level of each data according to the importance score of each data, and use the determination result as the importance label of the data.

[0029] According to a cloud data processing and isolation method based on a multi-tenant architecture provided by the present invention, in step S22, the weight calculation process includes:

[0030] S221. Calculate the objective weight of each indicator using the following formula: ;

[0031] In the formula, represents the objective weight of the ith indicator, represents the i-th index value, represents the sum of all data values ​​under the i-th indicator, It indicates the proportion of the i-th indicator to all the data of this indicator.

[0032] S222. Calculate the subjective weight of each indicator , the process includes:

[0033] Establish a pairwise comparison matrix A, where ;

[0034] Normalize each column of matrix A to get the matrix ,in, ;

[0035] Sum the normalized matrix B row by row to get the vector ,in, ;

[0036] Normalize the vector C to get the weight vector ,in, ;

[0037] In the formula, n represents the number of indicators, represents the subjective weight of the ith indicator, Indicates the importance of indicator i relative to indicator j, represents the i-th element in the vector C obtained by summing the normalized matrix B row by row. represents the elements in matrix B obtained by normalizing the columns of matrix A. is an element in the pairwise comparison matrix A, indicating the importance of indicator k relative to indicator j. is the kth element in vector C, which is obtained by summing the rows of the normalized matrix B.

[0038] S223. Calculate the comprehensive weight of each indicator using the following formula: ;

[0039] In the formula, represents the comprehensive weight of the i-th indicator, is the adjustment coefficient, and its value range is (0, 1).

[0040] According to a cloud data processing and isolation method based on a multi-tenant architecture provided by the present invention, in step S23, the method of calculating the data sensitivity index score includes:

[0041] Collect historical cloud data and annotate sensitivity scores based on data information;

[0042] Extract sensitive features from historical cloud data, including data types and keywords;

[0043] Construct a scoring model based on a convolutional neural network, take sensitivity features as input, output sensitivity scores, train the scoring model, and retain model parameters that meet test accuracy;

[0044] Extract the sensitivity features of the current data, input them into the trained scoring model, and output the sensitivity score of the current data;

[0045] The data sensitivity index score is obtained according to the sensitivity score. The formula is expressed as: ;

[0046] The methods for calculating the data access frequency index score, data update frequency index score, data value assessment index score, and data and business process correlation index score are as follows: ;

[0047] in, represents the data sensitivity score, represents the data access frequency score, Indicates the data update frequency score, Indicates the data value assessment score represents the correlation score between data and business process, G represents the actual number of visits, Indicates the minimum number of visits, represents the maximum number of visits, T represents the actual update frequency, Indicates the minimum update frequency, represents the maximum update frequency, Z represents the revenue generated by the data, represents the minimum benefit generated by the data, represents the maximum benefit generated by the data, Q represents the number of business processes in which the data is used, Indicates the minimum number of business processes, represents the maximum number of business processes, k represents the index coefficient, Denotes sensitivity score.

[0048] According to a cloud data processing and isolation method based on a multi-tenant architecture provided by the present invention, in step S24, the calculation formula of the importance score is:

[0049] ;

[0050] In the formula, P is the importance score, represents the weight of the second feature index, Represents the second feature index score, i=1, 2, 3, 4, 5.

[0051] According to a cloud data processing and isolation method based on a multi-tenant architecture provided by the present invention, in step S3, the layered encryption strategy includes:

[0052] Use homomorphic encryption algorithm to encrypt and protect data of high importance level;

[0053] Use role-based access control mechanism to encrypt data of medium importance level;

[0054] A symmetric encryption algorithm is used to encrypt data of low importance level.

[0055] According to a cloud data processing and isolation method based on a multi-tenant architecture provided by the present invention, in step S4, the process of creating a separate index mechanism is:

[0056] Creating separate indexes according to the first feature, including a data format index, a data source index, a data field index, and a data structure index;

[0057] A multi-layer B+ tree is used as the underlying data structure. Each internal node stores the interval range of the first feature of the data, and the leaf node stores the data ID and its corresponding storage location.

[0058] The index key is the one-hot coded value of the data format, data source, data field, and data structure classification dimension;

[0059] Define the index value as the unique data ID of all data under this classification label;

[0060] When data is stored in the database, the data feature vector is inserted into four independent B+ tree indexes in sequence. In each index, the data ID and its storage location are added to the corresponding leaf node.

[0061] According to a cloud data processing and isolation method based on a multi-tenant architecture provided by the present invention, in step S5, the differentiated access strategy includes:

[0062] For high-importance data, establish fine-grained role-based permissions;

[0063] For medium-importance data, establish department- or project-based access permissions;

[0064] For low-importance data, establish open access permissions and record access logs.

[0065] The present invention provides a cloud data processing and isolation method based on a multi-tenant architecture, which comprehensively labels and grades the data by establishing a data classification system and an importance assessment mechanism. Then, differentiated encryption strategies are adopted according to different importance levels to effectively protect sensitive data. Independent indexing mechanisms and access control strategies can effectively isolate data from different tenants and reduce the risks of data leakage and unauthorized access. By formulating differentiated access control strategies, the security and availability of key business data can be ensured, and the responsiveness of key businesses can be improved. At the same time, global and local abnormal behavior monitoring mechanisms help to promptly discover and respond to various security risks and ensure business continuity. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0067] Figure 1 It is a schematic diagram of the steps of a cloud data processing and isolation method based on a multi-tenant architecture provided by an embodiment of the present invention;

[0068] Figure 2 It is a schematic diagram of the steps of establishing an importance division system in a cloud data processing and isolation method based on a multi-tenant architecture provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0069] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0070] Combine the following Figure 1-Figure 2 The present invention describes a cloud data processing and isolation method based on a multi-tenant architecture.

[0071] Figure 1 It is a schematic diagram of the steps of a cloud data processing and isolation method based on a multi-tenant architecture provided by an embodiment of the present invention.

[0072] like Figure 1 As shown, an embodiment of the present invention provides a cloud data processing and isolation method based on a multi-tenant architecture, including:

[0073] S1. Establish a data classification system and add classification labels to the data in the cloud database based on the first characteristics of the data. The first characteristics include data format, data source, data field and data structure.

[0074] The process of establishing a data classification system includes:

[0075] S11. Standardize each data in the cloud database, extract the first feature of each data, and construct a data feature vector.

[0076] S12. Build a classification model based on support vector machine, input feature vector, and obtain the type score of each data.

[0077] The process of building a classification model based on support vector machines includes:

[0078] Collect the first features of historical data in the cloud database, construct feature vectors, and label each data sample with the correct classification label, dividing it into training set and test set;

[0079] Select Kernel Support Vector Machine to build the base model;

[0080] Train the basic model on the training set and adjust the kernel function;

[0081] Use the test set to evaluate the trained basic model and calculate the accuracy;

[0082] The basic model parameters with the highest accuracy are retained to obtain the classification model.

[0083] S13. Set a type score threshold, and add the same classification label to data with type scores within the same threshold range.

[0084] By standardizing the first feature of the data, constructing the feature vector, and then using the support vector machine model for automatic classification, the efficiency and accuracy of data classification are greatly improved. The traditional method of relying on manual labeling is avoided, and scalability is improved. The support vector machine has strong generalization ability and adaptability. By collecting historical data for model training and evaluation, it can adapt well to cloud data of different types and structures to meet the needs of various business scenarios. By setting up an automated classification scheme, not only the efficiency is improved, but also the uniformity and consistency of the classification standards can be ensured, thereby greatly reducing the overall data management cost.

[0085] S2. Establish an importance assessment system to determine the importance level of data based on the second characteristic of the data, and use the determination result as the importance label of the data. The second characteristic includes data access frequency index, data update frequency index, data sensitivity index, data value assessment index and data-business process correlation index.

[0086] Figure 2 It is a schematic diagram of the steps of establishing an importance division system in a cloud data processing and isolation method based on a multi-tenant architecture provided in an embodiment of the present invention.

[0087] like Figure 2 As shown, the process of establishing an importance evaluation system provided by the embodiment of the present invention includes:

[0088] S21. Collect the second feature of each data.

[0089] S22. Calculate the weight of each indicator of the second feature. The weight calculation process includes: S221. Calculate the objective weight of each indicator. The formula is:

[0090] ;

[0091] In the formula, represents the objective weight of the ith indicator, represents the i-th index value, represents the sum of all data values ​​under the i-th indicator, It indicates the proportion of the i-th indicator to all the data of this indicator.

[0092] S222. Calculate the subjective weight of each indicator , the process includes:

[0093] Establish a pairwise comparison matrix A, where ;

[0094] Normalize each column of matrix A to get the matrix ,in, ;

[0095] Sum the normalized matrix B row by row to get the vector ,in, ;

[0096] Normalize the vector C to get the weight vector ,in, ;

[0097] In the formula, n represents the number of indicators, represents the subjective weight of the ith indicator, Indicates the importance of indicator i relative to indicator j, represents the i-th element in the vector C obtained by summing the normalized matrix B row by row. represents the elements in matrix B obtained by normalizing the columns of matrix A. is an element in the pairwise comparison matrix A, indicating the importance of indicator k relative to indicator j. is the kth element in vector C, which is obtained by summing the rows of the normalized matrix B.

[0098] S223. Calculate the comprehensive weight of each indicator using the following formula: ;

[0099] In the formula, represents the comprehensive weight of the i-th indicator, is the adjustment coefficient, and its value range is (0, 1).

[0100] S23. Obtain an indicator score for each indicator, wherein the method for calculating the data sensitivity indicator score includes:

[0101] Collect historical cloud data and annotate sensitivity scores based on data information;

[0102] Extract sensitive features from historical cloud data, including data types and keywords;

[0103] Construct a scoring model based on a convolutional neural network, take sensitivity features as input, output sensitivity scores, train the scoring model, and retain model parameters that meet test accuracy;

[0104] Extract the sensitivity features of the current data, input them into the trained scoring model, and output the sensitivity score of the current data;

[0105] The data sensitivity index score is obtained according to the sensitivity score. The formula is expressed as: ;

[0106] The methods for calculating the data access frequency index score, data update frequency index score, data value assessment index score, and data and business process correlation index score are as follows: ;

[0107] in, represents the data sensitivity score, represents the data access frequency score, Indicates the data update frequency score, represents the data value assessment score, represents the correlation score between data and business process, G represents the actual number of visits, Indicates the minimum number of visits, represents the maximum number of visits, T represents the actual update frequency, Indicates the minimum update frequency, represents the maximum update frequency, Z represents the revenue generated by the data, represents the minimum benefit generated by the data, represents the maximum benefit generated by the data, Q represents the number of business processes in which the data is used, Indicates the minimum number of business processes, represents the maximum number of business processes, k represents the index coefficient, Denotes sensitivity score.

[0108] In this embodiment, by constructing a scoring model based on a convolutional neural network, the data sensitivity score is mapped in the range of 0 to 1, thereby mapping the data sensitivity index score in the range of 0 to k.

[0109] S24. Calculate the importance score of each data according to the indicator score and the corresponding weight. The calculation formula is: ;

[0110] In the formula, P is the importance score, represents the weight of the second feature index, Represents the second feature index score, i=1, 2, 3, 4, 5.

[0111] S25. Set an importance score threshold range corresponding to each importance level, where the importance levels include low importance, medium importance, and high importance.

[0112] S26. Determine the importance level of each data according to the importance score of each data, and use the determination result as the importance label of the data.

[0113] By collecting the second characteristics of data, including data access frequency index, data update frequency index, data sensitivity index, data value assessment index and data and business process correlation index, a highly comprehensive importance assessment index system is constructed from multiple dimensions. The multi-angle assessment method can more accurately reflect the actual importance of data. Not only can the indicator weights be flexibly adjusted according to different business scenarios, but also the relative contribution of each indicator to the final importance score can be better reflected. The pertinence of the assessment results is improved. The multi-dimensional data importance assessment mechanism provides strong support for subsequent refined management strategies such as differentiated encryption and access control. It can ensure the high protection of key data while reducing the management cost of non-key data. The automated importance assessment system can greatly improve the efficiency and accuracy of the assessment. At the same time, the assessment indicators and weights can be adjusted at any time according to business needs, which improves the flexibility of the entire data management.

[0114] S3. Set up a layered encryption strategy and adopt corresponding encryption methods for data of different importance levels.

[0115] The layered encryption strategy includes:

[0116] Use homomorphic encryption algorithm to encrypt and protect data of high importance level;

[0117] Use role-based access control mechanism to encrypt data of medium importance level;

[0118] A symmetric encryption algorithm is used to encrypt data of low importance level.

[0119] S4. Create a separate indexing mechanism based on the first feature.

[0120] The process of creating a separate indexing mechanism involves:

[0121] Creating separate indexes according to the first feature, including a data format index, a data source index, a data field index, and a data structure index;

[0122] A multi-layer B+ tree is used as the underlying data structure. Each internal node stores the interval range of the first feature of the data, and the leaf node stores the data ID and its corresponding storage location.

[0123] The index key is the one-hot coded value of the data format, data source, data field, and data structure classification dimension;

[0124] Define the index value as the unique data ID of all data under this classification label;

[0125] When data is stored in the database, the data feature vector is inserted into four independent B+ tree indexes in sequence. In each index, the data ID and its storage location are added to the corresponding leaf node.

[0126] S5. Develop differentiated access control strategies based on the type and importance level of the data.

[0127] Differentiated access strategies include:

[0128] For high-importance data, formulate role-based fine-grained permissions, including: granting each role only the minimum access rights, strictly reviewing and approving access rights applications, implementing two-factor authentication security measures, recording each access behavior and conducting regular audits;

[0129] For medium-importance data, formulate department or project-based access rights, including: department heads or project heads approving access rights, restricting cross-department or cross-project data access, logging and auditing key operations, and establishing data access management processes and approval mechanisms;

[0130] For low-importance data, open access rights are established and access logs are recorded. No approval is required, but identity authentication is required, real-time alerts are issued for abnormal access behavior, and access logs are analyzed and reviewed regularly.

[0131] S6. Establish a monitoring mechanism to realize global abnormal behavior monitoring and local abnormal behavior monitoring of cloud databases.

[0132] The steps to achieve global abnormal behavior monitoring include:

[0133] Set monitoring data for cloud databases, including the number of database connections, CPU and memory usage, disk I / O, transaction processing latency, and query execution time.

[0134] Based on historical data analysis, a reasonable abnormal threshold is set for the monitoring data, and if the threshold is exceeded, it is determined as global abnormal behavior.

[0135] Use monitoring tools to obtain monitoring data in real time and compare it with the preset abnormal threshold. If an indicator is found to be abnormal, the alarm mechanism will be triggered immediately and the relevant operation and maintenance personnel will be notified to handle it.

[0136] Further analyze the root cause of the problem and take corresponding optimization measures.

[0137] The steps to achieve local abnormal behavior monitoring include:

[0138] Collect tenants' access logs, including access time, access frequency, and access location information, to establish normal user behavior patterns.

[0139] Monitor the user's access behavior in real time and compare it with the pre-established normal pattern. If a tenant's access behavior is found to be significantly different from the normal pattern, it will be considered as local abnormal behavior.

[0140] Further classification is carried out for different types of local abnormal behaviors, including: abnormal login: change of login location, abnormal login time, abnormal access: abnormal access frequency, abnormal access object, abnormal operation: batch deletion, data modification.

[0141] According to the severity of abnormal behavior, the risk to the system is assessed and the corresponding level of alarm is triggered. The alarm information includes the type of abnormal behavior, risk level, and impact range.

[0142] For low-risk local abnormal behaviors, the system automatically takes isolation and flow control measures. For high-risk behaviors, manual intervention is used for further investigation and processing.

[0143] In summary, this embodiment provides a cloud data processing and isolation method based on a multi-tenant architecture. By establishing a complete data classification system and importance assessment mechanism, cloud data is comprehensively classified and labeled according to importance, which can effectively identify and distinguish different tenants and data of different importance. Combined with the layered encryption strategy, data is encrypted and protected in a differentiated manner, which greatly improves the security of the data. At the same time, the independent indexing mechanism and differentiated access control strategy can effectively isolate the data of different tenants and avoid the risks of data leakage and cross-access. The refined data classification and importance labeling enable data managers to take targeted management measures for data of different types and importance. The differentiated access control strategy based on data importance can ensure the security and availability of key business data and improve the responsiveness of key businesses. The management method based on data characteristics and importance enables the data of different tenants to be effectively isolated and managed in a differentiated manner. Tenants can customize the data classification system and access control strategy according to their own data characteristics and business needs, which greatly improves the flexibility of data management and tenant experience.

[0144] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0145] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A cloud data processing and isolation method based on a multi-tenant architecture, characterized in that: The method comprises: S1. Establish a data classification system to add classification labels to the data in the cloud database according to the first characteristics of the data, wherein the first characteristics include data format, data source, data domain and data structure; S2. Establish an importance classification system, judge the importance level of data according to the second feature of the data, and use the judgment result as the importance label of the data. The second feature includes a data access frequency index, a data update frequency index, a data sensitivity index, a data value assessment index, and an index of the correlation between data and business processes; The process of establishing the importance classification system includes: S21, collecting the second feature of each data; S22. Calculate the weight of each indicator in the second feature, the process comprising: S221. Calculate the objective weight of each indicator using the following formula: In the formula, θ i represents the objective weight of the ith indicator, X i represents the i-th index value, ∑X i represents the sum of all data values ​​under the i-th indicator, It indicates the proportion of the i-th indicator to all the data of this indicator; S222. Calculate the subjective weight of each indicator, the process includes: Establish a pairwise comparison matrix A, the formula is expressed as: A=(a ij ) n*n Normalize each column of matrix A to obtain matrix B = (b ij ) n*n , the formula is: Sum the normalized matrix B row by row to get the vector C = (c1, c2, ..., c n ) T , the formula is: Normalize the vector C to get the weight vector W = (w1, w2, ..., w n ) T , the formula is: In the formula, n represents the number of indicators, w i represents the subjective weight of the ith indicator, a ij Indicates the importance of indicator i relative to indicator j, c i represents the i-th element in the vector C obtained by summing the rows of the normalized matrix B, b ij represents the elements in matrix B obtained after normalizing the columns of matrix A, a kj is an element in the pairwise comparison matrix A, indicating the importance of indicator k relative to indicator j, c k is the kth element in vector C, which is obtained by summing the rows of the normalized matrix B; S223. Calculate the comprehensive weight of each indicator. The formula is: H i =σ*θ i +(1-σ)*w i In the formula, H i represents the comprehensive weight of the i-th indicator, σ is the adjustment coefficient, and its value range is (0, 1); S23, obtaining the indicator score of each indicator; S24, calculating the importance score of each data according to the indicator score and the corresponding weight; S25, setting an importance score threshold range corresponding to each importance level, where the importance levels include low importance, medium importance and high importance; S26, judging the importance level of each data according to the importance score of each data, and using the judgment result as the importance label of the data; S3. Set up a hierarchical encryption strategy and adopt corresponding encryption methods for data of different importance levels; S4. Create a separate indexing mechanism according to the first feature; S5. Formulate differentiated access control strategies based on the type of data and the importance level; S6. Establish a monitoring mechanism to realize global abnormal behavior monitoring and local abnormal behavior monitoring of cloud databases.

2. A cloud data processing and isolation method based on a multi-tenant architecture according to claim 1, characterized in that: In step S1, the process of establishing the data classification system includes: S11, standardize each data in the cloud database, extract the first feature of each data, and construct a data feature vector; S12, constructing a classification model based on a support vector machine, inputting the feature vector, and obtaining a type score for each data; S13: Set a type score threshold, and add the same classification label to the data whose type scores are within the same threshold range.

3. A cloud data processing and isolation method based on a multi-tenant architecture according to claim 2, characterized in that: The process of constructing the support vector machine-based classification model includes: Collect the first features of historical data in the cloud database, construct feature vectors, and label each data sample with the correct classification label, dividing it into training set and test set; Select Kernel Support Vector Machine to build the base model; Training the basic model on the training set and adjusting the kernel function; Use the test set to evaluate the trained basic model and calculate the accuracy; The basic model parameters with the highest accuracy are retained to obtain the classification model.

4. The cloud data processing and isolation method based on a multi-tenant architecture according to claim 1, characterized in that: In step S23, the method of calculating the data sensitivity index score includes: Collect historical cloud data and annotate sensitivity scores based on data information; Extracting sensitivity features from the historical cloud data, wherein the sensitivity features include data types and keywords; Constructing a scoring model based on a convolutional neural network, taking the sensitivity feature as input, outputting the sensitivity score, training the scoring model, and retaining model parameters that meet the test accuracy; Extract the sensitivity features of the current data, input them into the trained scoring model, and output the sensitivity score of the current data; The data sensitivity index score is obtained according to the sensitivity score, and the formula is expressed as: M1=k*CNN(β) The methods for calculating the data access frequency index score, data update frequency index score, data value assessment index score, and data and business process correlation index score are as follows: Among them, M1 represents the data sensitivity score, M2 represents the data access frequency score, M3 represents the data update frequency score, M4 represents the data value assessment score, M5 represents the correlation score between data and business processes, G represents the actual number of accesses, and G represents the actual number of accesses. min represents the minimum number of visits, G max represents the maximum number of accesses, T represents the actual update frequency, and T min Indicates the minimum update frequency, T max represents the maximum update frequency, Z represents the revenue generated by the data, and Z min represents the minimum benefit generated by the data, Z max represents the maximum benefit generated by the data, Q represents the number of business processes in which the data is used, and Q min Indicates the minimum number of business processes, Q max represents the maximum number of business processes, k represents the indicator coefficient, and CNN(β) represents the sensitivity score.

5. The cloud data processing and isolation method based on a multi-tenant architecture according to claim 1, characterized in that: In step S24, the calculation formula of the importance score is: P=∑H i *M i In the formula, P is the importance score, H i Represents the weight of the second feature index, M i Represents the second feature index score, i=1, 2, 3, 4, 5.

6. A cloud data processing and isolation method based on a multi-tenant architecture according to claim 1, characterized in that: In step S3, the layered encryption strategy includes: Use homomorphic encryption algorithm to encrypt and protect data of high importance level; Use role-based access control mechanism to encrypt data of medium importance level; A symmetric encryption algorithm is used to encrypt data of low importance level.

7. The cloud data processing and isolation method based on a multi-tenant architecture according to claim 1, characterized in that: In step S4, the process of creating a separate index mechanism is: Creating separate indexes according to the first feature, including a data format index, a data source index, a data field index, and a data structure index; A multi-layer B+ tree is used as the underlying data structure. Each internal node stores the interval range of the first feature of the data, and the leaf node stores the data ID and its corresponding storage location. The index key is the one-hot coded value of the data format, data source, data field, and data structure classification dimension; Define the index value as the unique data ID of all data under this classification label; When data is stored in the database, the data feature vector is inserted into four independent B+ tree indexes in sequence. In each index, the data ID and its storage location are added to the corresponding leaf node.

8. The cloud data processing and isolation method based on a multi-tenant architecture according to claim 1, characterized in that: In step S5, the differentiated access strategy includes: For high-importance data, establish fine-grained role-based permissions; For medium-importance data, establish department- or project-based access permissions; for low-importance data, establish open access permissions and record access logs.

Images