Safety control method for realizing large model computing power center data gridding

Abstract

The invention discloses a security control method for realizing large model computing power center data gridding, and relates to the technical field of data security, and the method comprises the steps: firstly determining computing power center data, obtaining a sensitivity index, an access frequency and a data size of the computing power center data, generating an optimal total number of blocks, and carrying out the blocking processing of the data, thereby obtaining a data block sequence; analyzing pre-stored physical location hash values and logic partitions of the data blocks, constructing quantized grid coordinates and carrying out gridding storage; meanwhile, the cross-region access frequency is monitored in real time, and a fusing mechanism is triggered to interrupt access if the cross-region access frequency exceeds the limit; and after fusing, adding offset to the quantized grid coordinates of the related data block sequence, regenerating and storing, and clearing the data block at the original storage position to ensure the safe storage and access control of the large model computing power center data.

Description

Technical Field

[0001] This invention belongs to the field of data security technology, specifically, it relates to a security control method for realizing data gridding in large-scale computing power centers. Background Technology

[0002] With the advent of the big data era, data has become a key element driving economic development and social progress. In large data centers, efficient data management and security control are crucial. Existing data storage technologies face many challenges. The surge in data volume not only requires storage systems to have greater capacity, but also efficient data access and management mechanisms.

[0003] Existing technologies have certain shortcomings in data storage management. Traditional storage methods are mostly centralized, which is difficult to meet the needs of data growth and is prone to single points of failure, affecting data security and reliability. In terms of data access control, existing technologies are mostly based on simple permission management, which is difficult to effectively monitor and control complex situations such as cross-regional access, making it easy to lead to unauthorized access. Data storage locations are mostly fixed, and once they are cracked, data is easily leaked. To address the aforementioned issues, this invention proposes a security control method for implementing data gridding in large-scale model computing centers. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a security control method for realizing data gridding in large-scale computing centers, solving the problems existing in data storage management, access control, and storage security in existing technologies.

[0005] The objective of this invention can be achieved through the following technical solutions: A security control method for implementing data gridding in large-scale computing power centers, comprising the following: Step 1: Determine any one of the computing center data from the several computing center data received by the large model, obtain the sensitivity index, access frequency, and data size of the computing center data, generate the optimal total number of blocks associated with the computing center data, perform block processing on the computing center data, and obtain the computing center data block sequence. Step 2: Obtain the pre-stored physical location hash value and logical partition of each data block in the data block sequence of the computing center, analyze them, construct the quantized grid coordinates corresponding to each data block, and store them in a grid. Step 3: Monitor the frequency of cross-region access when accessing data blocks in real time. When the frequency of cross-region access exceeds the cross-region access frequency limit, it is considered an illegal access and the circuit breaker mechanism is automatically triggered to interrupt the access. Step 4: When the circuit breaker mechanism is triggered, extract the data block sequence of the computing center where the data block corresponding to the triggering circuit breaker mechanism is located, add offsets to the original quantization grid coordinates associated with each data block in the computing center data block sequence, and regenerate the quantization grid coordinates associated with each data block for grid storage. After the gridded storage is completed, each data block in the computing center data block sequence at the original storage location will be cleared.

[0006] As a further aspect of the present invention, the specific method for generating the optimal total number of blocks associated with the computing center data in step one is as follows: At the current moment, acquire data from several computing centers received by the large model, extract data from any one computing center, and denote it as... ; Acquiring large model reception The moment is recorded as the start moment. ; Use the current time as the end time. ,statistics to Internally, users The total number of visits is calculated by averaging them per unit time. The access frequency is marked as Wherein, the unit time is the time period preset by the operator; Determine based on the data category table pre-built by the operator Data categories, labeled as ; Then extract from the sensitivity index table pre-built by the operator The associated sensitivity index is denoted as ; Get the operator as the data category The preset target access frequency is marked as ; Remove , and The dimensions of the quantity are removed, and the dimensionless value is reassigned to the value. , and ; use ,get Total number of associated blocks ,in, This is the floor function.

[0007] As a further aspect of the present invention, in step one, the data category table pre-constructed by the operator is as follows: ,in, Represents the total number of all data categories and ; The sensitivity index table is as follows: , correspond Where i is the counting index, and its value ranges from 1 to... .

[0008] As a further aspect of the present invention, the specific method for dividing the computing center data into blocks to obtain the computing center data block sequence in step one is as follows: Based on the determined total number of blocks Data from the computing center Average score A data block, denoted as the data block sequence of the computing center. .

[0009] As a further aspect of the present invention, the specific method for constructing the quantization grid coordinates corresponding to each data block in step two is as follows: Extracting data block sequences from computing centers Determine any one of the data blocks ,in, This is a counting index, with values ​​ranging from 1 to... ; extract Pre-stored physical location, and generated using a hash algorithm The physical location hash value is used to obtain the data block. The first 6 bits of the physical location hash value are taken, converted to an integer, and then processed according to the base to obtain the data block. The associated physical dimension coordinates are denoted as ; Get data blocks Pre-stored database DB; Logically partition the database DB using a grid-based approach based on the data category table; Sure and the associated computing center data Data categories The logical partition in question is marked as ; merge Logical partitions and physical dimension coordinates Determine data blocks The associated quantized grid coordinates are denoted as: .

[0010] As a further aspect of the present invention, the specific method for performing gridded logical partitioning of the database DB in step two is as follows: Get data category table Extract the total number of data categories. ; Divide DB into equal parts One logical partition, namely , where any logical partition Indicates the type of stored data. The computing center data and its associated data blocks, and .

[0011] As a further aspect of the present invention, in step two, data blocks are obtained through radix processing. The specific method for determining the associated physical dimension coordinates is as follows: Obtain the data after the database has been logically partitioned into grids. Logical partitions ; Get Any logical partition among the logical partitions ; Construct a two-dimensional coordinate system with 8 scales from 0 to 7 on the horizontal axis and 8 scales from 0 to 7 on the vertical axis, resulting in a total of 64 integer coordinate points on both the horizontal and vertical axes; Each integer coordinate point represents One of the storage units, that is, a logical partition It is divided into 64 storage units; Get The first 6 digits of the associated physical location hash value are converted to an integer; Calculate the average of the first 6 digits. Starting from the first digit, compare it with the average. If the digit is greater than or equal to the average, assign the value 1 to the digit; otherwise, assign the value 0. The first three digits are used as the first group, converted to decimal, and denoted as: ; The next three digits are grouped as the second set, converted to decimal, and denoted as . ; collection , have to The associated physical dimension coordinates are represented as ,express Two-dimensional coordinates stored in a two-dimensional coordinate system In the corresponding storage unit; Each data block in the process is processed using the method described above.

[0012] As a further aspect of the present invention, in step three, the specific method for automatically triggering the circuit breaker mechanism and denying access is as follows: Real-time monitoring of any access request, and locking onto the computing center data accessed by that request, marking it as... ; Determine the data of the computing center The total number of associated data blocks is denoted as ; Get the total number of times the database storage units were accessed during the time period from the start to the end of this access request. ; This is considered a normal access and will not be processed. Conversely, if the access is deemed unauthorized, the circuit breaker mechanism will be automatically triggered, interrupting the access request.

[0013] As a further aspect of the present invention, in step four, the specific method for regenerating the quantized grid coordinates associated with each data block for gridded storage is as follows: If access to the computing center data is confirmed If the access request is deemed invalid and triggers the circuit breaker mechanism, then the data will be retrieved. The associated computing center data block sequence ; Get any data block and its associated quantized grid coordinates ; Separate Physical dimension coordinates Two sets of binary arrays of three-digit integers are generated sequentially using a random function; Convert the two binary arrays to decimal to obtain two integers; The two integers obtained are used in the order of their generation. The associated physical dimension coordinates are denoted as ; Combination as well as Get data block Associated quantized grid coordinates ; by right Implement grid-based storage; Each data block in the process is processed in the same way as described above.

[0014] As a further aspect of the present invention, in step four, after the gridded storage is completed, the specific method for clearing each data block in the computing center data block sequence at the original storage location is as follows: The central data block sequence any data block In its associated quantized grid coordinates Clear the data blocks stored in the middle; Similarly, for Repeat the above steps for each data block to complete the cleanup.

[0015] The beneficial effects of this invention are: This application, when processing data from large-scale computing centers, generates an optimal total number of blocks based on factors such as the data's sensitivity index, access frequency, and data size, and then performs block processing accordingly. This enables more refined and rational data partitioning. Secondly, by constructing quantized grid coordinates and storing the data blocks in a grid format, the data storage location is clearly quantified and managed in an orderly manner, which improves data retrieval and storage efficiency, as well as the optimized utilization of storage space. The distributed storage of data also enhances data security control capabilities to a certain extent. Furthermore, by monitoring cross-regional access frequency in real time and setting a limit-triggered circuit breaker mechanism, this application can effectively and promptly block unauthorized access, ensuring data security. This application achieves the dual advantages of intelligent dynamic block partitioning and precise coordinate positioning in data grid security control. Based on the dynamic coupling relationship between data sensitivity index, real-time access frequency, and preset security goals, the total number of blocks is intelligently adjusted: the number of blocks for data with higher sensitivity or more frequent access is automatically increased, significantly improving the difficulty of cracking unauthorized access. At the same time, by integrating physical location hash values ​​and logical partitions based on data categories to construct quantized grid coordinates, the precise location of each data block in the distributed environment is efficiently anchored, while avoiding the complexity and single-point failure risk of traditional metadata management. This design greatly enhances the anti-attack capability of highly sensitive data in computing centers and optimizes the positioning efficiency of authorized access through coordinate gridding, providing data grid support with both strong security and high manageability for large-scale model computing environments. This application divides the database into N dedicated logical partitions based on the total number of data categories, enabling logical isolation of data with different sensitivity levels before physical storage. This avoids the risk of cross-access of unauthorized data from the source. At the same time, this application innovatively converts the first 6 bits of the hash value into two-dimensional coordinates (7 scales on each of the horizontal and vertical axes to form 49 coordinate points). The physical dimension coordinates are generated by a dynamic threshold binarization algorithm, which ensures the randomness and unpredictability of the coordinates. Furthermore, the database is gridded and partitioned using a two-dimensional coordinate system, so that each data block can be accurately anchored to the corresponding storage unit, reducing the retrieval complexity of authorized access. This application combines dynamic circuit breaking with a coordinate randomization reconstruction mechanism to achieve intelligent interception of unauthorized access and real-time self-healing of data storage locations. A circuit breaking threshold is established based on the total number of data blocks, determining the number of access partitions. When the number of logical partitions scanned in a single request exceeds the total number of data blocks, the circuit breaker is triggered, accurately identifying fragmented theft behavior in a distributed environment. Once the circuit breaker is triggered, the physical coordinates of all data blocks are randomly reset, generating two sets of random binary numbers to reconstruct the physical dimension coordinates. These reconstructed coordinates are then bound to the original logical partitions to form new quantized grid coordinates. This operation instantly invalidates the coordinate information already obtained by the attacker, significantly increasing the cost of persistent attacks. Furthermore, the dynamic combination of random coordinates and logical partitions ensures that the new location of the data blocks cannot be predicted. Finally, the operation of clearing data from the original storage location completely eliminates residual risks, forming a closed-loop protection of "circuit breaking-migration-clearing," providing anti-attack capabilities that combine real-time threat blocking with automated location concealment. Attached Figure Description

[0016] The invention will now be further described with reference to the accompanying drawings.

[0017] Figure 1 This is a flowchart illustrating the method described in this invention; Figure 2 This is a flowchart illustrating the method described in Embodiment 2 of the present invention; Figure 3 This is a flowchart illustrating the method described in Embodiment 3 of the present invention. Detailed Implementation

[0018] 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.

[0019] Example 1 A security control method for implementing data gridding in large-scale computing power centers, such as... Figure 1 As shown, the specific steps include the following: Before this method is implemented, it is necessary to ensure that it can interact with the large model that is determined or needs to be determined in real time, and to obtain data from several computing centers received by the large model in real time, as well as to obtain access requests sent by users when using the large model in real time. During the preparation phase, operators also need to pre-build structured metadata for this method, including: a data category table, a sensitivity index table, and the target access frequency associated with the computing center data for each data category, with the data categories in the data category table corresponding one-to-one with the sensitivity indices in the sensitivity index table.

[0020] Specifically, the data category table is represented as follows: This data category table needs to be satisfied. The data category table should include data from various computing center types. If it doesn't include all data categories, it should be added and supplemented promptly when new data categories of computing center data are encountered. Total number of all data categories, data category table The data categories are sorted according to the order in which they were designed by the operators; The sensitivity index table is represented as follows: Similarly, the sensitivity index table contains sensitivity indices associated with data categories of various computing center data. If it cannot contain sensitivity indices associated with all data categories, then when encountering new data categories of computing center data, the sensitivity indices associated with that data category should be added to the sensitivity index table promptly. The data categories in the data category table correspond one-to-one with the sensitivity indices in the sensitivity index table. correspond Where i is the counting index, and its value ranges from 1 to... ; Next, it is necessary to record the relevant data involved in the process of the large model processing user access requests, including: the computing center data that needs to be obtained in the user access request, as well as the sensitivity index, access frequency, and data size associated with this computing center data. The sensitivity index is based on the data category of this computing center data, the access frequency is based on the number of times this computing center data is accessed per unit time, and the data size is obtained directly from this computing center data.

[0021] By analyzing the sensitivity index and access frequency corresponding to the data of this computing center, the optimal total number of blocks associated with this computing center data is generated. The computing center data is then processed by block division using the optimal total number of blocks, and the data blocks obtained by block division are recorded as the computing center data block sequence according to the order of block division. Obtain the physical location of the data block pre-stored associated with the data from this computing center (this is before the data is stored after being divided into blocks). Process the physical location of each data block using a hash algorithm (such as SHA3-128 iterative hash) to obtain the physical location hash value associated with each data block. (During this process, there may be cases where the pre-stored physical locations of data blocks from multiple computing centers are the same, resulting in the same physical location hash value obtained by the final hash algorithm. Here, a random function is added to generate a set of random numbers (such as a globally unique UUID), and this random number is used as the retrieval index for this data block. At the same time, this random number is concatenated with the physical location of the data block to obtain a unique string. Then, the hash algorithm is executed on this unique string to obtain the physical location hash value associated with each data block.) Next, the database containing the data blocks associated with the data of this computing center, which are determined by the operators in the above steps, is logically partitioned into grids to obtain the logical partitions in the database corresponding to the data category table (that is, the computing center data of each data category is stored in a corresponding logical partition in the database). Then, by combining the data categories of this computing center data, the logical partition in which the database of this computing center data will be stored is determined. Next, by analyzing the physical location hash value associated with the physical location of each data block, and combining the logical partition in which the database of this computing center data will be stored, the quantized grid coordinates of each data block are determined, and the gridded storage operation is completed for each data block according to the quantized grid coordinates.

[0022] The gridded storage operation of the data blocks is now complete. The next step is to monitor the secure access to the data frames. After the gridded storage operation is completed, it is necessary to monitor every access request (the request to access the data blocks, which is also the request to access the data in the computing center) in real time. Get the cross-region access frequency of each access request from the start time to the end time, and monitor in real time whether the cross-region access frequency exceeds the cross-region access frequency limit. If it exceeds the limit, it is considered an illegal access, indicating that the access request has obtained data that is not related to the current access request, and the circuit breaker mechanism is automatically triggered to interrupt the current access request. Once it is determined that the circuit breaker mechanism will be automatically triggered, the data block sequence of the computing center where the data block obtained in this access request is located (obtained after processing the data block of the computing center) will be extracted. Any data block obtained by this illegal access request will be processed in accordance with this step. Next, extract the determined data block sequence of the computing center and the quantized grid coordinates determined during the first grid storage operation associated with each data block in the data block sequence of the computing center. Then, add an offset to the determined quantized grid coordinates to obtain the new quantized grid coordinates associated with each data block. Then, re-execute the grid storage operation for each data block according to the new quantized grid coordinates. After re-executing the gridded storage operation, it indicates that all data blocks obtained from this illegal access have been migrated. Then, the original storage locations of all data blocks are retrieved, and the data blocks at the original storage locations are cleared to release storage space.

[0023] This embodiment aims to ensure the security, efficiency, and compliance of data storage, while preventing unauthorized access and data leakage. In summary, by monitoring data access requests in real time, when the frequency of cross-regional access exceeds the limit, a circuit breaker mechanism is triggered to interrupt unauthorized access. The data blocks involved in the unauthorized access are then migrated and their storage locations updated to ensure data storage security and controllable access. This approach effectively improves the flexibility and security of data management, optimizes the allocation and utilization of storage resources, and meets the data security management needs of large models in different scenarios.

[0024] Example 2 This embodiment, based on Embodiment 1, discloses a method for determining the data block sequence of the computing power center based on the optimal total number of blocks, such as... Figure 2 As shown, the specific steps include the following: As described in Example 1, several computing center data points received by the large model can be obtained. This example uses any one of these computing center data points for processing, and this computing center data point is labeled as: ; The large model was used to determine the data it received from the computing center. That moment is marked as the start time and denoted as... Then, the current moment (i.e., the data from the computing center in the large model) is obtained. The time at which the event ends is marked as the end time, and denoted as... ; Next, the start time of the statistics is calculated. Until the end time During this period, users accessed data from the computing center. The total number of access requests; Then obtain the unit time preset by the operator based on the actual situation, where the unit time is a time period; Obtain user data for the computing center within a unit of time. The average number of access requests, and this average number is used as the data for the computing center. The associated access frequency is tagged as: ; At this point, the computing center data... Access frequency Acquisition complete.

[0025] The computing center data is determined from the data category table pre-built by the operator as described in Example 1. The data categories are labeled as follows: ,and , This represents a single identifier, signifying a data category table. Any one of the data from the computing center Data categories that match the data category are marked as ; Next, the sensitivity index table pre-built by the operators is retrieved, and data from the computing center is obtained from it. Data categories The associated sensitivity index is denoted as: ,and , This represents a single identifier, signifying a sensitivity index table. Any one of the data from the computing center The sensitivity index that meets the sensitivity index is denoted as... ; Similarly, acquiring data from computing centers Data categories The preset target access frequency is marked as follows: ; Thus, we obtain: Access frequency. Sensitivity Index and target access frequency Remove the dimensions from these three data points, take only the numerical values, and reassign them to the access frequency. Sensitivity Index and target access frequency (At this time, access frequency) Sensitivity Index and target access frequency (This only represents numerical values ​​and has no dimensions.) Dimensionless access frequency Sensitivity Index and target access frequency Substitute: Finally, the data from the computing center was calculated. Total number of associated blocks ,in, () is the floor function, which rounds the final calculation result to the nearest whole number to ensure that the total number of blocks is not rounded down. Integer, total number of blocks With access frequency and sensitivity index They are all directly proportional, that is, the frequency of access. The larger the value, the greater the total number of blocks. The more, the higher the sensitivity index. The larger the value, the greater the total number of blocks. The more that happens, the more there will be.

[0026] Data from the computing center has now been obtained. Total number of associated blocks According to the total number of blocks Data from computing centers Perform block processing to divide the data from the computing center Average score A data block, denoted as the data block sequence of the computing center, is represented as follows: .

[0027] The purpose of this embodiment is to determine the total number of data blocks associated with the computing center data, so as to perform reasonable block processing on the data. First, from the start time of receiving computing center data from the large model to the current end time, the total number of times users request to access the computing center data is counted. Combined with the unit time preset by the operator, the average number of requests to access the computing center data per unit time is calculated, i.e., the access frequency. Then, the data category of the computing center data is determined from the data category table pre-built by the operator, and the sensitivity index corresponding to the data category is obtained from the sensitivity index table. At the same time, the preset target access frequency is also obtained. Next, after removing the dimensions of access frequency, sensitivity index and target access frequency, they are substituted into a specific formula to calculate the total number of blocks. The total number of blocks is proportional to the access frequency and sensitivity index and is an integer. Finally, the computing center data is evenly divided according to the total number of blocks to form a data block sequence. This realizes block management based on comprehensive consideration of access demand and sensitivity of computing center data, and improves the flexibility and targeting of data processing.

[0028] Example 3 This embodiment, based on Embodiments 1 and 2, further discloses a method for gridded storage of data blocks, such as... Figure 3 As shown, the specific steps include the following: The determined computing center data can be obtained from Examples 1 and 2. and computing center data The associated computing center data block sequence In this embodiment, the data block sequence of the computing power center is taken. If any one data block is processed as an example, the remaining data blocks will be processed in the same and synchronous manner as this data block. Mark any of the determined data blocks as ,in, This is a counting index, with values ​​ranging from 1 to... ; Next, obtain the data block. The database that is about to undergo storage operations is marked as DB; Obtain the data category table pre-built by the operator from the content described in Example 1. And extract the total number of data categories from the data category table. ; According to the total number of data categories in the data category table Perform grid-based logical partitioning on the database (DB), which means dividing the database's storage space into equal parts. Each logical partition with the same storage space, and this Each logical partition is represented as: , where any logical partition Indicates the type of stored data. The data of the computing center and its associated data blocks (this partitioning operation is a persistent operation unless the data category table is modified or updated); Next, continue to determine the data blocks. Data from the associated computing center Related data categories According to data category Data blocks can be obtained The logical partition to be stored, that is ; Next, extract this logical partition. Then, perform a further partitioning operation on this logical partition, as follows: A two-dimensional coordinate system is constructed with 8 scales (0 to 7) on the horizontal axis and 8 scales (0 to 7) on the vertical axis. In this two-dimensional coordinate system, a total of 64 integer coordinate points are obtained, including those on the horizontal and vertical axes. Each integer coordinate point represents a logical partition. One of the storage units after partitioning; Thus, each storage unit corresponds to a two-dimensional coordinate in a two-dimensional coordinate system.

[0029] Next, data blocks are obtained according to the content described in Example 1. The physical location to be stored (before storage operations begin) is determined by combining data blocks with a hash algorithm and the addition of random numbers. The physical location to be stored is hashed to obtain the physical location hash value associated with this physical location (this physical location hash value must be greater than 6 bits. Generally, the SHA3-128 iterative hash algorithm will generate a 128-bit physical location hash value, which satisfies the condition of being greater than 6 bits). Extract the determined data blocks The first six bits of the physical location hash value are converted into an integer, resulting in a 6-bit integer. The first three bits are grouped into the first group, and the last three bits are grouped into the second group. Next, calculate the average of the 6-digit integers, and compare it with the average starting from the first integer. If the value of the corresponding integer is greater than or equal to the average, then assign the value 1 to this number. Similarly, if the value of the corresponding integer is less than the average value, then this number is assigned the value 0; Finally, we obtain the first and second groups, which are both sequences of three integers consisting of 0 or 1. We treat the first and second groups as binary (sequences of integers consisting of 0 or 1) and convert them to decimal to obtain the corresponding integers for the first and second groups respectively. That is, convert the first group into decimal representation, and the resulting integer is denoted as . ; Convert the second group to decimal representation, and the resulting integer is denoted as . ; Will As the x-axis, Using the vertical axis as the ordinate, we can summarize them to obtain a two-dimensional coordinate. , this two-dimensional coordinate As a data block The associated physical dimension coordinates are denoted as , is represented as: ; Define two-dimensional coordinates The corresponding integer coordinate points and storage units in the two-dimensional coordinate system, and the data block The data is stored in this storage unit, thus completing the gridded storage operation.

[0030] data blocks Logical partitions in the database DB And in logical partitions The associated physical dimension coordinates Merge the data to obtain data blocks. The associated quantized grid coordinates are denoted as: Subsequent query operations will be based on the quantized grid coordinates. Determine data blocks The stored unit is then used for further retrieval, thus avoiding the complexity of global retrieval.

[0031] This embodiment first randomly selects a data block from the data block sequence of the computing center for processing, determines the database in which it is stored, and performs grid-based logical partitioning of the database according to the data category table. Each partition corresponds to a specific data category and its data blocks. Next, the logical partition containing the specific data block is further partitioned to construct a two-dimensional coordinate system, where each coordinate point represents a storage unit. Then, the hash value of the physical location of the data block is obtained through a hash algorithm and random number calculation. The first six bits are extracted and converted into a binary sequence, and the corresponding two-dimensional coordinates are calculated. The data block is then stored in the corresponding storage unit. At the same time, the logical partition and the physical dimension coordinates are merged to generate quantized grid coordinates. In this way, the storage unit can be quickly located based on the quantized grid coordinates during subsequent queries, avoiding global retrieval, reducing complexity, and improving data access efficiency.

[0032] Example 4 This embodiment, based on embodiments 1, 2, and 3, further discloses a method for circuit breaking and coordinate randomization reconstruction of illegal access requests, specifically including the following steps: As can be seen from Example 3, the grid storage operation of the data block has been completed. When users make access requests in the future, they need to request the corresponding data block from the corresponding database. Monitor and acquire each access request in real time, and extract any one of the access requests (the remaining access requests are also processed in the same way as this access request). The computing center data that needs to be accessed in this access request is marked. To distinguish it from the computing center data in the above embodiments, it is marked here as... ; Next, the computing center data is obtained according to the method described in the above embodiments. The total number of associated data blocks is denoted as: (The methods described in the above embodiments are considered to have been fully completed, and any subsequent operations that use the methods described in the above embodiments will not be repeated.) Next, obtain the data from the computing center during the time interval from the start to the end of this access request. The total number of requested accesses to storage units in the logical partitions of the stored database is denoted as... ; The total number of times obtained Less than or equal to the total If it is a normal access, no action will be taken. If the total number of times Greater than the total If this is not done, it will be considered unauthorized access and the circuit breaker mechanism will be automatically triggered. If the total Greater than the total If the request is not accepted, it will be considered an unauthorized access, and the circuit breaker mechanism will be automatically triggered to interrupt the access request.

[0033] Next, obtain data from the computing center. The associated data block sequence of the computing center is represented as follows: ; Then from the determined data block sequence of the computing center Get any data block The example processing is performed, and the remaining data blocks are processed in the same synchronous manner as follows; Get data blocks When storing a storage unit, the originally associated quantization grid coordinates ; From the determined data block Associated quantized grid coordinates Separate physical dimension coordinates Next, a random function is used to generate two sets of three-digit integer binary numbers in sequence, and the corresponding two decimal numbers are calculated. These two decimal numbers are then used as data blocks in the order they were generated. The associated new physical dimension coordinates are denoted as ; Combine data blocks The associated new physical dimension coordinates and data blocks The associated logical partition Get data block The associated new quantized grid coordinates ; Quantize grid coordinates For data blocks Implement grid-based storage.

[0034] All data in the formulas described above are numerical calculations performed after removing their dimensions. Furthermore, any content not described in detail in this specification is existing technology known to those skilled in the art.

[0035] The above description is merely an example and illustration of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the invention or exceed the scope defined in the claims, all of which should fall within the protection scope of the present invention.

[0036] It should be stated that all user data collected in this application was collected with the user's consent and authorization. Furthermore, the uses of user data are legal and compliant, and the use and processing of user data comply with the relevant laws, regulations, and standards of the relevant regions.

Claims

1. A security control method for implementing data gridding in large-scale model computing centers, characterized in that, This method includes the following: Step 1: Determine any one of the computing center data from the several computing center data received by the large model, obtain the sensitivity index, access frequency, and data size of the computing center data, generate the optimal total number of blocks associated with the computing center data, perform block processing on the computing center data, and obtain the computing center data block sequence. Step 2: Obtain the pre-stored physical location hash value and logical partition of each data block in the data block sequence of the computing center, analyze them, construct the quantized grid coordinates corresponding to each data block, and store them in a grid. Step 3: Monitor the frequency of cross-region access when accessing data blocks in real time. When the frequency of cross-region access exceeds the cross-region access frequency limit, it is considered an illegal access and the circuit breaker mechanism is automatically triggered to interrupt the access. Step 4: When the circuit breaker mechanism is triggered, extract the data block sequence of the computing center where the data block corresponding to the triggering circuit breaker mechanism is located, add offsets to the original quantization grid coordinates associated with each data block in the computing center data block sequence, and regenerate the quantization grid coordinates associated with each data block for grid storage. After the gridded storage is completed, each data block in the computing center data block sequence at the original storage location will be cleared.

2. The security control method for realizing data gridding in a large-scale computing center according to claim 1, characterized in that, In step one, the specific method for generating the optimal total number of blocks associated with the computing center data is as follows: At the current moment, acquire data from several computing centers received by the large model, extract data from any one computing center, and denote it as... ; Acquiring large model reception The moment is recorded as the start moment. ; Use the current time as the end time. ,statistics to Internally, users The total number of visits is calculated by averaging them per unit time. The access frequency is marked as Wherein, the unit time is the time period preset by the operator; Determine based on the data category table pre-built by the operator Data categories, labeled as ; Then extract from the sensitivity index table pre-built by the operator The associated sensitivity index is denoted as ; Get the operator as the data category The preset target access frequency is marked as ; Remove , and The dimensions of the quantity are removed, and the dimensionless value is reassigned to the value. , and ; use ,get Total number of associated blocks ,in, This is the floor function.

3. The security control method for realizing data gridding in a large-scale computing center according to claim 2, characterized in that, In step one, the data category table pre-constructed by the operator is as follows: ,in, Represents the total number of all data categories and ; The sensitivity index table is as follows: , correspond Where i is the counting index, and its value ranges from 1 to... .

4. The security control method for realizing data gridding in a large-scale computing center according to claim 2, characterized in that, In step one, the specific method for dividing the computing center data into blocks to obtain the computing center data block sequence is as follows: Based on the determined total number of blocks Data from the computing center Average score A data block, denoted as the data block sequence of the computing center. .

5. The security control method for implementing data gridding in a large-scale computing center according to claim 1, characterized in that, In step two, the specific method for constructing the quantization grid coordinates corresponding to each data block is as follows: Extracting data block sequences from computing centers Determine any one of the data blocks ,in, This is a counting index, with values ​​ranging from 1 to... ; extract Pre-stored physical location, and generated using a hash algorithm The physical location hash value is used to obtain the data block. The first 6 bits of the physical location hash value are taken, converted to an integer, and then processed according to the base to obtain the data block. The associated physical dimension coordinates are denoted as ; Get data blocks Pre-stored database DB; Logically partition the database DB using a grid-based approach based on the data category table; Sure and the associated computing center data Data categories The logical partition in question is marked as ; merge Logical partitions and physical dimension coordinates Determine data blocks The associated quantized grid coordinates are denoted as: .

6. A security control method for implementing data gridding in a large-scale computing center according to claim 5, characterized in that, In step two, the specific method for performing gridded logical partitioning of the database DB is as follows: Get data category table Extract the total number of data categories. ; Divide DB into equal parts One logical partition, namely , where any logical partition Indicates the type of stored data. The computing center data and its associated data blocks, and .

7. A security control method for implementing data gridding in a large-scale computing center according to claim 5, characterized in that, In step two, data blocks are obtained through number base processing. The specific method for determining the associated physical dimension coordinates is as follows: Obtain the data after the database has been logically partitioned into grids. Logical partitions ; Get Any logical partition among the logical partitions ; Construct a two-dimensional coordinate system with 8 scales from 0 to 7 on the horizontal axis and 8 scales from 0 to 7 on the vertical axis, resulting in a total of 64 integer coordinate points on both the horizontal and vertical axes; Each integer coordinate point represents One of the storage units, that is, a logical partition It is divided into 64 storage units; Get The first 6 digits of the associated physical location hash value are converted to an integer; Calculate the average of the first 6 digits. Starting from the first digit, compare it with the average. If the digit is greater than or equal to the average, assign the value 1 to the digit; otherwise, assign the value 0. The first three digits are used as the first group, converted to decimal, and denoted as: ; The next three digits are grouped as the second set, converted to decimal, and denoted as . ; collection , have to The associated physical dimension coordinates are represented as ,express Two-dimensional coordinates stored in a two-dimensional coordinate system In the corresponding storage unit; Each data block in the process is processed using the method described above.

8. A security control method for implementing data gridding in a large-scale computing center according to claim 1, characterized in that, In step three, the circuit breaker mechanism is automatically triggered, and access is denied in the following way: Real-time monitoring of any access request, and locking onto the computing center data accessed by that request, marking it as... ; Determine the data of the computing center The total number of associated data blocks is denoted as ; Get the total number of times the database storage units were accessed during the time period from the start to the end of this access request. ; This is considered a normal access and will not be processed. Conversely, if the access is deemed unauthorized, the circuit breaker mechanism will be automatically triggered, interrupting the access request.

9. A security control method for implementing data gridding in a large-scale computing center according to claim 8, characterized in that, In step four, the specific method for regenerating the quantized grid coordinates associated with each data block and storing them in a grid is as follows: If access to the computing center data is confirmed If the access request is deemed invalid and triggers the circuit breaker mechanism, then... The associated computing center data block sequence ; Get any data block and its associated quantized grid coordinates ; Separate Physical dimension coordinates Two sets of binary arrays of three-digit integers are generated sequentially using a random function; Convert the two binary arrays to decimal to obtain two integers; The two integers obtained are used in the order of their generation. The associated physical dimension coordinates are denoted as ; Combination as well as Get data block Associated quantized grid coordinates ; by right Implement grid-based storage; Each data block in the process is processed in the same way as described above.

10. A security control method for implementing data gridding in a large-scale computing center according to claim 9, characterized in that, In step four, after the gridded storage is completed, the specific method for clearing each data block in the computing center data block sequence at the original storage location is as follows: The central data block sequence any data block In its associated quantized grid coordinates Clear the data blocks stored in the middle; Similarly, for Repeat the above steps for each data block to complete the cleanup.

Images