Centralized management method for data center cluster

Abstract

The invention discloses a centralized management method for a data center cluster, and relates to the technical field of data management. The method comprises the following steps: acquiring an original data set in a database cluster, and performing deduplication processing on an important score original data set to obtain a deduplicated data set; carrying out random division according to the sample number of the data in the important score de-duplication data set to obtain a plurality of granularity balls, carrying out grading processing on the data through the granularity balls, carrying out compression encryption processing on the data subjected to important score grading to obtain encrypted data, and carrying out migration and storage on the encrypted data according to the important score; redundant data are reduced by deduplication of a large amount of data in a database, and the data are more compact through granularity ball division and dispersion calculation; the data are correspondingly stored according to the importance level after being graded, high-efficiency data management is achieved, the storage cost is reduced through data compression, data security is protected through encryption, and high-efficiency data management is achieved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of data management, and particularly relates to a centralized management method of a data center cluster. BACKGROUND

[0002] A data center cluster is composed of multiple geographically dispersed but interconnected data centers, which cooperatively provide computing, storage and network services. Through resource sharing, load balancing and fault switching, high availability and scalability are achieved to support cloud services and enterprise applications. Centralized management of a data center cluster is a mode of operation and maintenance in which a unified management platform integrates, schedules and monitors resources such as servers, storage and networks distributed in multiple locations; it abstracts physical resources into a resource pool, combines virtualization and automated operation and maintenance, and realizes dynamic allocation of computing power, unified control of data flow, centralized alarm and rapid recovery of faults; centralized management can improve resource utilization, simplify operation and maintenance processes, reduce labor costs, and through real-time monitoring and policy configuration, ensure system stability and data security, and is suitable for business expansion needs of large-scale enterprises and cloud service providers.

[0003] The prior art (publication number: CN112052287A) discloses a centralized management method of a data center cluster, the data center cluster includes multiple data centers, the method includes: obtaining environmental control data of at least one data center in the data center cluster; converting the obtained environmental control data into standardized environmental control data according to a preset environmental control data standard; classifying and storing the standardized environmental control data according to the importance of the data.

[0004] However, in actual use, the method does not perform deduplication processing on the data before data grading, which may cause some repeated and redundant data to occupy computing resources, and the operation of classifying the data according to the importance score is only a simple introduction, without a detailed solution, and the data is not stored in different storage media according to different importance scores, which may result in poor effect of data in the grading storage, thereby affecting the efficiency of users in the data centralized management process. SUMMARY

[0005] The purpose of the present application is to solve the problem of low efficiency of grading storage in the centralized management process of a data cluster, and a centralized management method of a data center cluster is proposed.

[0006] In the implementation of the present application, a centralized management method of a data center cluster is first proposed, the method includes:

[0007] Obtaining an original data set in a database cluster, and performing deduplication processing on the original data set to obtain a deduplicated data set;

[0008] a plurality of granularity balls are obtained by randomly dividing the sample number of data in the deduplicated data set according to the granularity ball algorithm;

[0009] a dispersion of the target granularity ball is calculated, if the dispersion is greater than 0, the target granularity ball is first split to obtain two first granularity balls, the sum of the dispersions of the two first granularity balls is minimum, and each first granularity ball is added to the first granularity ball set; the target granularity ball is any one of the plurality of granularity balls;

[0010] the mean and median of the radius of each first granularity ball in the first granularity ball set are calculated, if the target first granularity ball meets the preset condition, the target first granularity ball is second split to obtain two second granularity balls, the sum of the dispersions of the two second granularity balls is minimum, and each second granularity ball of each first granularity ball is added to the second granularity ball set; the target first granularity ball is any one of the first granularity ball set;

[0011] the importance score of each second granularity ball in the second granularity ball set is calculated, and the graded data is obtained by grading processing according to the importance score;

[0012] the encrypted data is obtained by compression and encryption processing according to the graded data; and the migration and storage are performed according to the encrypted data.

[0013] Optionally, the dispersion calculation process of the granularity ball is:

[0014] the distribution metric and the weighted distribution metric of the granularity ball are calculated, the dispersion is obtained by subtracting the distribution metric from the weighted distribution metric;

[0015] the graded data is obtained by grading processing according to the importance score, including:

[0016] the Euclidean distance between two granularity balls is calculated as the fuzzy similarity;

[0017] the data importance score is calculated according to the fuzzy similarity;

[0018] the graded data is obtained by important level division according to the data importance score.

[0019] Optionally, the data importance score is calculated according to the fuzzy similarity, including:

[0020] a fuzzy relation matrix is constructed by the fuzzy similarity between the granularity balls; the information granule cardinality of each granularity ball is calculated according to the fuzzy relation matrix; and the importance degree of each attribute is calculated according to the information granule cardinality;

[0021] an ordered attribute sequence is obtained by ascending according to the importance degree of the attribute, and an attribute simplification sequence and an attribute complex sequence are constructed according to the ordered attribute sequence;

[0022] For each attribute subset in the attribute simplified sequence, the information granules of the granular ball are extracted from its fuzzy relation matrix to form an attribute simplified granular sequence;

[0023] For each attribute subset in the attribute complex sequence, the information granules of the granular ball are extracted from its fuzzy relation matrix to form an attribute complex granular sequence;

[0024] The weight of each granular ball is calculated according to the cardinality of the information granules of a single attribute;

[0025] The mean of the cardinality of the information granules in the attribute simplified granular sequence and the attribute complex granular sequence is fused to calculate the importance score of the granular ball.

[0026] Optionally, the mean of the cardinality of the information granules in the attribute simplified granular sequence and the attribute complex granular sequence is fused to calculate the importance score of the granular ball, and specifically includes:

[0027]

[0028] wherein, is the importance score of the i-th granular ball, is the weight of the i-th granular ball, denotes the i-th granular ball, denotes the cardinality of the q-th information granule in the attribute simplified granular sequence, denotes the cardinality of the q-th information granule in the attribute complex granular sequence; m refers to the total number of attributes in the attribute set; g represents the total number of granular balls in the granular ball set.

[0029] Optionally, the hierarchical data is obtained by classifying the data according to the importance score, and specifically includes:

[0030] The importance score is compared with a threshold interval:

[0031] If the importance score is less than the lower limit of the threshold interval, the data importance is classified into a third level;

[0032] If the importance score is not less than the lower limit of the threshold interval and not greater than the upper limit of the threshold interval, the data importance is classified into a second level;

[0033] The importance score is compared with a threshold interval, and if the importance score is greater than the upper limit of the threshold interval, the data importance is classified into a first level.

[0034] Optionally, the hierarchical data is compressed, and specifically includes:

[0035] Step 1: Obtain the original binary sequence of the hierarchical data, and for a to-be-compressed original binary sequence with a length of n, count the number of 1s h in the binary sequence, and convert the binary sequence into an h-balance code combination;

[0036] Step two: from the least significant bit of the h-balance code combination, remove consecutive 0s until the first 1 is encountered, obtaining a binomial number Y with a length of y, and y < n;

[0037] Step three: convert the binomial number into a natural number system number with a length of t through a binomial numbering function;

[0038] Step four: determine whether to compress according to the compression ratio formula, and only compress sequences with a compression ratio greater than a threshold value, otherwise directly transmit the original binary sequence.

[0039] Optionally, the hierarchical data is encrypted to obtain encrypted data, specifically including:

[0040] The sensitive data after the hierarchical processing is encrypted using the advanced encryption standard AES-256, and the encryption key is securely generated, distributed, and rotated using the key management service KMS to obtain the encrypted data; during data transmission, the data channel is encrypted using the transport layer security protocol TLS; the sensitive data includes user identity authentication information, business critical data, and system configuration data.

[0041] Optionally, the encrypted data is migrated and stored, specifically including:

[0042] For the data of the first level, it is stored in a first storage space and is redundantly backed up in multiple copies;

[0043] For the data of the second level, it is stored in a second storage space and is redundantly backed up;

[0044] For the data of the third level, it is stored in a third storage space;

[0045] The read-write speed of the first storage space > the read-write speed of the second storage space > the read-write speed of the third storage space.

[0046] Optionally, after migrating and storing the encrypted data, the method further includes:

[0047] The hash value of the data is checked at regular intervals to ensure that the integrity of the data is not damaged during storage and transmission;

[0048] The data access behavior patterns of users and systems are monitored in real time, and once an abnormal access pattern is detected, an alarm is triggered and isolation is started.

[0049] Optionally, according to the importance, a differentiated backup strategy is formulated:

[0050] An object storage service is used to store backup data to multiple physical locations;

[0051] ​Establishing a disaster recovery center library, synchronously copying key data to the disaster recovery center library, and quickly switching to the disaster recovery center to resume business operation when the main data center fails.

[0052] The present application has the following beneficial effects:

[0053] The present application provides a centralized management method of data center cluster, which reduces redundant data by deduplication of a large amount of data in the database, and makes the data more compact through granularity sphere division and dispersion calculation; the hierarchical data is stored on the corresponding storage medium according to the importance level, realizing efficient data management, reducing storage cost through data compression, protecting data security through encryption, controlling authority, and improving data availability through disaster recovery; making the data processing efficient, safe and reliable, and meeting the higher demand for data management. BRIEF DESCRIPTION OF DRAWINGS

[0054] The present application will be further described below with reference to the accompanying drawings.

[0055] Figure 1 A flowchart of the centralized management method of data center cluster provided by the present application embodiment is shown in the figure.

[0056] Figure 2 A block diagram of another centralized management method of data center cluster provided by the present application embodiment is shown in the figure. DETAILED DESCRIPTION

[0057] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. The term "and / or" in this paper is only to describe the association relationship of the associated objects, which means that there are three kinds of relationships, for example, A and B, which can represent the existence of A, the existence of A and B, and the existence of B. In addition, the description of "first", "second" and the like in the present application is only for the purpose of description, and cannot be understood as indicating or implying the relative importance of the indicated technical features or implicitly indicating the number of the indicated technical features. Therefore, the features limited by "first", "second" can explicitly or implicitly include at least one of the features. In addition, the technical solutions of each embodiment can be combined with each other, but it must be based on the realization of ordinary skilled in the art, when the combination of technical solutions appears contradictory or unachievable, it should be considered that the combination of technical solutions does not exist, and is not within the scope of protection claimed by the present application.

[0058] Based on the embodiments in the present application, all other embodiments obtained by ordinary skilled in the art without creative labor belong to the scope of protection of the present application.

[0059] The embodiment of the present application provides a centralized management method of a data center cluster. Figure 1 , Figure 1 The embodiment of the present application provides a flow chart of the centralized management method of the data center cluster. The method comprises the following steps:

[0060] S101, obtaining an original data set in a database cluster, and performing deduplication processing on the original data to obtain a deduplicated data set;

[0061] S102, performing random division on the deduplicated data set according to a sample quantity of data in the deduplicated data set by using a granularity ball algorithm to obtain a plurality of granularity balls;

[0062] S103, calculating a dispersion degree of a target granularity ball, if the dispersion degree is greater than 0, performing first splitting on the target granularity ball to obtain two first sub-granularity balls, making the sum of the dispersion degrees of the two first sub-granularity balls minimum, and adding the first sub-balls of each granularity ball to a first sub-ball set;

[0063] S104, calculating a radius mean value and a radius median of each first sub-ball in the first sub-ball set, if a target first sub-ball satisfies a preset condition, performing second splitting on the target first sub-ball to obtain two second sub-balls, making the sum of the dispersion degrees of the two second sub-balls minimum, and adding the second sub-balls of each first sub-ball to a second sub-ball set;

[0064] S105, calculating an important score of each second sub-ball in the second sub-ball set, and performing hierarchical processing according to the important score to obtain hierarchical data;

[0065] S106, performing compression and encryption processing according to the hierarchical data to obtain encrypted data, and performing migration and storage according to the encrypted data.

[0066] The target granularity ball is any one of the plurality of granularity balls; and the target first sub-ball is any one of the first sub-balls.

[0067] According to the centralized management method of the data center cluster, the present application reduces redundant data by deduplicating a large amount of data in a database, makes the data more compact by granularity ball division and dispersion degree calculation, stores the data according to importance levels after data classification, realizes efficient data management, reduces storage cost by data compression, protects data security by encryption, improves data availability by permission control and disaster recovery, and makes data processing efficient, safe and reliable, thereby meeting the higher management requirement for data.

[0068] In an implementation, the deduplication of the original data includes calculating a hash value of a data block to identify duplicate data before the data is written to the storage medium; the deduplication of data saves storage space, improves performance and data reliability; the data set is randomly divided into multiple granularity balls according to the number of samples of the data in the deduplicated data set; the specific number of samples is an optimal division method obtained according to historical data, and can also be adjusted by relevant personnel.

[0069] In an implementation, the preset condition is that the radius of the target first sub-ball is greater than the radius of the target first sub-ball . is the average radius of each first sub-ball, is the median radius of each first sub-ball; if the dispersion is not greater than 0, the target granularity ball is not subjected to the first splitting, and the target granularity ball that is not subjected to the first splitting is added to the first sub-ball set as a first sub-ball; if the target first sub-ball does not satisfy the radius , the target first sub-ball is not subjected to the second splitting, and the target first sub-ball that is not subjected to the second splitting is added to the second sub-ball set as a second sub-ball.

[0070] In an implementation, the dispersion of the granularity ball is used for splitting to optimize data distribution and make the data more compact; the radius average and the radius median are combined in the two splitting operations to effectively control the size of the granularity ball and avoid excessive segmentation or insignificant processing; finally, the fuzzy similarity between the granularity balls is calculated by the Euclidean distance to accurately reflect the relative relationship of the data and improve the efficiency and accuracy of the data classification processing.

[0071] In an embodiment, the dispersion calculation process of the granularity ball is as follows:

[0072] The distribution metric and the weighted distribution metric of the granularity ball are calculated, and the dispersion is obtained by subtracting the distribution metric from the weighted distribution metric;

[0073] The data after the classification processing includes:

[0074] The Euclidean distance between the two granularity balls is calculated as the fuzzy similarity;

[0075] The data importance score is calculated according to the fuzzy similarity;

[0076] The data after the classification processing is obtained according to the importance level division of the data importance score.

[0077] In an implementation, the calculation of the distribution metric and the weighted distribution metric of the granularity ball includes temporarily dividing the target granularity ball to generate two temporary sub-balls and .

[0078]

[0079]

[0080]

[0081] where, is the distribution metric of the tth target granularity ball, is the weighted distribution metric of the target granularity ball, denotes the tth target granularity ball, is the number of samples in the tth target granularity ball, denotes the sum of Euclidean distances from all samples in the tth target granularity ball to its center, denotes the ith sample in the target granularity ball, denotes the center of the tth target granularity ball; G is the target granularity ball to be split, G1 and G2 are two temporary sub-balls, , , are the number of samples of the target granularity ball and two temporary sub-balls, respectively;

[0082] The dispersion μ is obtained by subtracting the distribution metric from the weighted distribution metric:

[0083]

[0084]

[0085] where c represents the average value of samples in the ball and is the center of the granularity ball, and k is the total number of samples in the ball;

[0086] First, calculate the Euclidean distance between the centers of the two balls and

[0087]

[0088] Then calculate the fuzzy similarity , the process is:

[0089]

[0090] where B represents the attribute set of the granularity ball, is the fuzzy similarity of the centers of the two balls under the attribute set, is the Euclidean distance of the centers of the two balls under the attribute set, is the center of the ith granularity ball under the qth attribute, is the center of the jth granularity ball under the qth attribute, is the qth attribute in the attribute set, ∈[0,1] is a hyperparameter, which is a threshold value used to control the similarity, and is set according to the experiment, is the center of the ith granularity ball,​ is the center of the jth granularity ball.

[0091] In an implementation, the essence of a granularity ball is an information unit covering a local data region and having semantics; the radius of the granularity ball is a measure of the uncertainty or generalization range of the information unit, quantifying the average dispersion degree of data samples within the ball; the dispersion degree is obtained by calculating the difference between the distribution measure and the weighted distribution measure of the granularity ball, which can optimize data distribution and improve organization compactness; the Euclidean distance between the centers of granularity balls is used as the fuzzy similarity, and then the data importance score is calculated and the importance level is divided, which can accurately distinguish data value, improve grading accuracy and efficiency, and effectively handle complex target detection tasks.

[0092] In one embodiment, the data importance score is calculated according to the fuzzy similarity, comprising:

[0093] A fuzzy relation matrix is constructed by the fuzzy similarity between the granularity balls; the information granule cardinality of each granularity ball is calculated according to the fuzzy relation matrix; and the importance of each attribute is calculated according to the information granule cardinality.

[0094] An ordered attribute sequence is obtained in ascending order of the importance of the attributes, and an attribute simplification sequence and an attribute complex sequence are constructed according to the ordered attribute sequence.

[0095] For each attribute subset in the attribute simplification sequence, the information granules of the granularity balls are extracted by the fuzzy relation matrix to form an attribute simplification granular sequence.

[0096] For each attribute subset in the attribute complex sequence, the information granules of the granularity balls are extracted by the fuzzy relation matrix to form an attribute complex granular sequence.

[0097] The weight of each granularity ball is calculated according to the information granule cardinality of a single attribute.

[0098] The mean of the information granule cardinality in the attribute simplification granular sequence and the attribute complex granular sequence is fused, and the importance score of the granularity ball is calculated.

[0099] In an implementation, the fuzzy relation matrix is constructed as follows:

[0100]

[0101] wherein, is the fuzzy similarity between the granularity ball and the granularity ball , and g is the total number of granularity balls in the granularity ball set.

[0102] According to the fuzzy relation matrix, the information granule cardinality of each granularity ball is calculated, and the importance of the attribute is calculated according to the information granule cardinality. , the process is:

[0103]

[0104]

[0105] According to a single attribute The weight of each granular ball is calculated based on the information granule cardinality, including:

[0106]

[0107] wherein, is the weight of the i-th granular ball, is the information granule cardinality of the granular ball under a single attribute , g is the total number of granular balls in the granular ball set, and m is the total number of attributes in the attribute set B. In an implementation, a fuzzy relationship matrix is constructed by fuzzy similarity, the information granule cardinality and attribute importance are calculated, and the attributes are sorted to construct a simplified and complex sequence; information granules are extracted to form a granular sequence for different sequences, and the importance score of the granular ball is calculated in combination with the mean value of the information granule cardinality, which can more accurately measure the importance of data, improve the efficiency and accuracy of data processing, and enhance the explainability of data.

[0108] In one embodiment, the mean value of the information granule cardinality in the attribute simplified granular sequence and the attribute complex granular sequence is fused, and the importance score of the granular ball is calculated, specifically including:

[0109]

[0110]

[0111] wherein, is the importance score of the i-th granular ball, is the weight of the i-th granular ball, denotes the i-th granular ball, denotes the cardinality of the q-th information granule in the attribute simplified granular sequence, denotes the cardinality of the q-th information granule in the attribute complex granular sequence; m refers to the total number of attributes in the attribute set B; g represents the total number of granular balls in the granular ball set.

[0112] In an implementation, the precision and efficiency of anomaly detection are effectively improved by constructing a fuzzy relationship matrix and calculating an information granule cardinality in combination with the fusion of attribute simplified and complex sequences; it uses multi-granularity feature analysis to capture data anomalies from different dimensions, and the rationality of the dependence and weight design of attribute importance measurement becomes an important operation for performance improvement.

[0113] ​​In one embodiment, the data after grading is obtained according to the data importance score, specifically comprising:

[0114] The importance score is compared with the threshold interval:

[0115] If the importance score is less than the lower limit of the threshold interval, the data importance is divided into the third grade;

[0116] If the importance score is not less than the lower limit of the threshold interval and not greater than the upper limit of the threshold interval, the data importance is divided into the second grade;

[0117] The importance score is compared with the threshold interval, and if the importance score is greater than the upper limit of the threshold interval, the data importance is divided into the first grade.

[0118] In one implementation, the threshold interval is an optimal interval summarized based on historical data, which can well provide a basis for data grading; by setting the threshold interval of the importance score, the data is divided into three grades, which can quickly and clearly filter out the data area with high importance, realize the priority sorting of the data, and provide hierarchical management for data governance; it is convenient to take targeted measures according to different grades, and improve the data management efficiency.

[0119] In one embodiment, the data after grading is compressed, specifically comprising:

[0120] Step one: obtain the original binary sequence of the data after grading, count the number of 1s h in the original binary sequence to be compressed with a length of n, and convert the binary sequence into an h-balance code combination;

[0121] Step two: from the least significant bit of the h-balance code combination, remove consecutive 0s until the first 1 is encountered, to obtain a binomial number Y with a length of y, and y<n;

[0122] Step three: convert the binomial number into a natural number system number with a length of t through a binomial numbering function;

[0123] Step four: judge whether to compress according to the compression ratio formula, only compress the sequence with a compression ratio greater than a threshold value, otherwise directly transmit the original binary sequence.

[0124] In one implementation, referring to Figure 2 , Figure 2 a block diagram of a centralized management method of a data center cluster provided by an embodiment of the present application; the threshold value is obtained according to experiments; the compression ratio is calculated as follows:

[0125]

[0126] The compressed data is recovered, comprising: ​

[0127] After knowing the length t and h of the compressed binary number, the binary number is converted into the binomial number Y obtained in step two by inverse operation of the binomial numbering function;

[0128] According to whether the end of the binomial number Y ends with 1 or 0, n-h 0s or h 1s are padded to the least significant bit to restore the original binary sequence with a length of n.

[0129] In an implementation, the number of 0s in the binary sequence is counted and converted into a h-balanced code combination to remove consecutive 0s or 1s to form a binomial number, and then converted into a natural number by the binomial numbering function to realize data compression; at the same time, by judging the compression ratio, only the sequences with significant compression effect are compressed, otherwise the original sequence is directly transmitted to ensure the compression efficiency; the recovery process accurately restores the original sequence by inverse operation to realize lossless compression; it can effectively reduce the data storage and transmission amount, improve the efficiency, and is suitable for various data processing scenarios.

[0130] In one embodiment, the hierarchical data is encrypted to obtain encrypted data, specifically including:

[0131] The hierarchical sensitive data is encrypted by using the advanced encryption standard AES-256, and the encryption key is securely generated, distributed and replaced by using the key management service KMS to obtain the encrypted data; in the data transmission process, the data channel is encrypted by using the transport layer security protocol TLS; the sensitive data includes user identity authentication information, business critical data and system configuration data.

[0132] In an implementation, the confidentiality of sensitive data is protected by AES-256 encryption, the key is securely generated, distributed and replaced by combining KMS, and the data transmission channel is encrypted by using TLS to ensure the high security of data storage and transmission.

[0133] In one embodiment, the encrypted data is migrated and stored, specifically including:

[0134] The data of the first level is stored in the first storage space and is redundantly backed up in multiple copies;

[0135] The data of the second level is stored in the second storage space and is redundantly backed up;

[0136] The data of the third level is stored in the third storage space;

[0137] The read-write rate of the first storage space > the read-write rate of the second storage space > the read-write rate of the third storage space.

[0138] In an implementation, different storage strategies are adopted for data of different levels:

[0139] The first level of data is stored in a tape library or cloud archive storage, and is deeply compressed and stored in multiple copies, greatly reducing storage costs, while preventing permanent data loss through multiple copies, and adapting to archiving and long-term storage needs.

[0140] The second level of data is stored in a traditional mechanical hard disk or hybrid storage array, and is moderately backed up with redundancy, balancing storage costs and reliability, and ensuring that data is not lost under moderate performance conditions.

[0141] The third level of data is stored in a solid state disk or memory, and is configured with a low-latency data access path, which can quickly respond to high-frequency access requirements, reduce business delay, and improve key task efficiency.

[0142] In one embodiment, after migrating and storing the encrypted data, the method further comprises:

[0143] By checking the hash value of the data at regular intervals, the integrity of the data during storage and transmission is ensured not to be destroyed;

[0144] Real-time monitoring of user and system data access behavior patterns, once abnormal access patterns are detected, immediately trigger an alarm and start isolation.

[0145] In one implementation, the abnormal access pattern includes excessive data extraction, excessive access rate, remote login access, and abnormal IP address, etc. By checking the hash value of the data at regular intervals, it can effectively detect whether the data has been tampered with during storage and transmission, ensuring data integrity. At the same time, real-time monitoring of user and system data access behavior patterns can timely detect abnormal access or suspicious operations, quickly trigger an alarm and take isolation or blocking measures, thereby enhancing the security of the data access process and reducing the impact range and loss of security incidents.

[0146] In one embodiment, a differentiated backup strategy is developed:

[0147] Object storage services are used to store backup data in multiple physical locations;

[0148] A remote disaster recovery center library is established to synchronize and replicate critical data to the disaster recovery center library, which can quickly switch to the disaster recovery center to resume business operation in the event of a failure in the main data center.

[0149] In an implementation, the physical location can be a data center in another city, such as a data center in Qingpu District of Shanghai, a cloud computing center in Xiaoshan District of Hangzhou, Zhejiang Province, and a data center in Suzhou Industrial Park, Jiangsu Province; the differentiated backup strategy can effectively improve data security and business continuity; the object storage multi-location backup can prevent local physical damage, reduce cost, and be easy to expand; the off-site disaster recovery center synchronously replicates key data, can quickly recover business, and reduces the impact of extreme situations; through the combination of the two, a deep defense is formed to ensure high data availability.

[0150] The above describes one embodiment of the present application in detail, but the content is only the preferred embodiment of the present application, and cannot be considered as limiting the implementation range of the present application. Any equivalent changes and improvements made within the scope of the present application should still belong to the scope of the present application.

Claims

1. A method for centralized management of a data center cluster, characterized by, The method comprises: obtaining an original data set in a database cluster, and performing deduplication processing on the original data set to obtain a deduplicated data set; performing random division on the deduplicated data set according to a sample quantity of data in the deduplicated data set by a granularity ball algorithm to obtain a plurality of granularity balls; calculating a dispersion degree of a target granularity ball, and if the dispersion degree is greater than 0, performing first splitting on the target granularity ball to obtain two first sub-granularity balls, so that the sum of the dispersion degrees of the two first sub-granularity balls is minimum, and adding the first sub-balls of each granularity ball to a first sub-ball set; the target granularity ball is any one of the plurality of granularity balls; calculating a radius mean value and a radius median value of each first sub-ball in the first sub-ball set, and if a target first sub-ball meets a preset condition, performing second splitting on the target first sub-ball to obtain two second sub-balls, so that the sum of the dispersion degrees of the two second sub-balls is minimum, and adding the second sub-balls of each first sub-ball to a second sub-ball set; the target first sub-ball is any one of the first sub-balls in the first sub-ball set; calculating an important score of each second sub-ball in the second sub-ball set, and performing hierarchical processing according to the important score to obtain hierarchical data; performing compression and encryption processing on the hierarchical data to obtain encrypted data, and performing migration and storage according to the encrypted data.

2. The centralized management method of a data center cluster according to claim 1, characterized in that, The dispersion degree calculation process of the granularity ball comprises: calculating a distribution metric and a weighted distribution metric of the granularity ball, and subtracting the distribution metric from the weighted distribution metric to obtain the dispersion degree; the hierarchical processing according to the important score to obtain hierarchical data comprises: calculating the Euclidean distance between two granularity balls as a fuzzy similarity; calculating a data important score according to the fuzzy similarity; performing important level division according to the data important score to obtain hierarchical data.

3. The method of claim 2, wherein, The calculation of the data important score according to the fuzzy similarity comprises: constructing a fuzzy relation matrix through the fuzzy similarity between the granularity balls; calculating an information granule cardinality of each granularity ball according to the fuzzy relation matrix; and then calculating the importance of each attribute according to the information granule cardinality; arranging in ascending order according to the importance of the attributes to obtain an ordered attribute sequence, and constructing an attribute simplification sequence and an attribute complex sequence according to the ordered attribute sequence; for each attribute subset in the attribute simplification sequence, extracting an information granule of the granularity ball through the fuzzy relation matrix to obtain an attribute simplification granule sequence; for each attribute subset in the attribute complex sequence, extracting an information granule of the granularity ball through the fuzzy relation matrix to obtain an attribute complex granule sequence; calculating the weight of each granularity ball according to the information granule cardinality of a single attribute; fusing the mean value of the information granule cardinalities in the attribute simplification granule sequence and the attribute complex granule sequence to calculate the important score of the granularity ball.

4. The centralized management method of a data center cluster according to claim 3, characterized in that, The fusing of the mean value of the information granule cardinalities in the attribute simplification granule sequence and the attribute complex granule sequence to calculate the important score of the granularity ball specifically comprises: ； wherein, is the important fraction of the i-th granularity ball, is the weight of the i-th granularity ball, denotes the i-th granularity ball, denotes the cardinality of the q-th information granule in the sequence of attribute-reduced granules, denotes the cardinality of the q-th information granule in the sequence of attribute-complex granules; m refers to the total number of attributes in the attribute set; g denotes the total number of granularity balls in the set of granularity balls.

5. The centralized management method of a data center cluster according to claim 2, characterized in that, performing level division according to the data important score to obtain hierarchical data, specifically comprising: comparing the important score with a threshold interval; if the important score is less than the lower limit of the threshold interval, dividing the data importance into a third level; if the important score is not less than the lower limit and not greater than the upper limit of the threshold interval, dividing the data importance into a second level; The important score is compared with a threshold interval, and if the important score is greater than the upper limit of the threshold interval, the data importance is classified into a first level.

6. The method of claim 1, wherein, The method further comprises the following steps of: Step one: obtaining the original binary sequence of the classified data, counting the number of 1s in the original binary sequence of length n, and converting the binary sequence into an h-balance code combination; Step two: starting from the least significant bit of the h-balance code combination, removing consecutive 0s until the first 1 is encountered, obtaining a binomial number Y of length y, and y < n; Step three: converting the binomial number into a natural number system number with a length of t through a binomial numbering function; Step four: determining whether to compress according to a compression ratio formula, and only compressing the sequence with a compression ratio φ greater than a threshold value, otherwise directly transmitting the original binary sequence.

7. The method of claim 1, wherein, The classified data is encrypted to obtain encrypted data, specifically including: The sensitive data after classification is encrypted by using an advanced encryption standard AES-256, and a key management service KMS is used to securely generate, distribute and replace the encryption key to obtain the encrypted data; in the data transmission process, the data channel is encrypted by using a transport layer security protocol TLS; the sensitive data includes user identity authentication information, business critical data and system configuration data.

8. The centralized management method of a data center cluster according to claim 5, wherein, The encrypted data is migrated and stored, specifically including: For the data of the first level, the data is stored in a first storage space and is redundantly backed up in multiple copies; For the data of the second level, the data is stored in a second storage space and is redundantly backed up; For the data of the third level, the data is stored in a third storage space; The read-write speed of the first storage space > the read-write speed of the second storage space > the read-write speed of the third storage space.

9. The method of claim 1, wherein, After the encrypted data is migrated and stored, the method further comprises: The hash value of the data is checked at regular intervals to ensure that the integrity of the data is not damaged during storage and transmission; Real-time monitoring of user and system data access behavior patterns, and once an abnormal access pattern is detected, an alarm is triggered immediately.

10. The method of claim 1, wherein, The method further comprises formulating a differentiated backup strategy: Object storage services are used to store backup data to multiple physical locations; A remote disaster recovery center library is established, and critical data is synchronously copied to the disaster recovery center library, so that in the event of a failure of the main data center, the business can be quickly switched to the disaster recovery center to resume operation.

Images