A data security disaster recovery method and system for a data center

Abstract

The application discloses a data security disaster recovery method and system of a data center, and relates to the technical field of data processing. The method comprises the following steps: acquiring the data amount of the data center in the current day and the data amount of each historical day in a historical time period; dividing one day into multiple time stages based on the data amount of each historical day in the historical time period; predicting future data pressure based on the data repetition feature and the data growth feature of a target time stage; in the case that the future data pressure is less than or equal to a preset pressure threshold, comparing the data amount of the current day in the target time stage with the data amount of each historical day in the target time stage to determine the first data concentration degree of the current day in the target time stage; and determining a data synchronization strategy of the current moment based on the first data concentration degree, so that the data center is subjected to data security disaster recovery through the data synchronization strategy. According to the application, the data synchronization effect can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a data security disaster recovery method and system of a data center. BACKGROUND

[0002] In the daily operation and management of a data center, data security disaster recovery strategy plays a crucial role, which is the key to ensuring data integrity, availability and business continuity. Traditionally, data security disaster recovery mainly relies on physical level disaster recovery measures, such as configuring backup servers, storage devices and building disaster recovery rooms, etc. The core goal is to quickly recover data and maintain normal operation of business in the event of hardware failure, natural disasters or human errors and other sudden conditions.

[0003] At present, in order to realize efficient data security disaster recovery, the industry generally adopts full synchronization or incremental synchronization and other data synchronization technologies. These technologies have their own advantages and can effectively guarantee the consistency and real-time performance of data between different systems or storage locations.

[0004] However, the existing data synchronization scheme can only make a single choice between full synchronization and incremental synchronization, and cannot dynamically adjust between the two modes according to actual needs. This lack of flexibility often leads to poor data synchronization effect. SUMMARY

[0005] The embodiments of the present application provide a data security disaster recovery method and system of a data center, which can improve the data synchronization effect.

[0006] In a first aspect of the embodiments of the present application, a data security disaster recovery method of a data center is provided, comprising:

[0007] Obtaining the data volume of the data center in the current day and the data volume of each historical day in a historical time period;

[0008] Based on the data volume of each historical day in the historical time period, the day is divided into multiple time stages, and the data volume of each hour in the time stage presents consistent volatility;

[0009] Based on the data repetition feature and the data growth feature of the target time stage, the future data pressure is predicted, and the target time stage is the time stage in which the current time of the current day is located;

[0010] In the case where the future data pressure is less than or equal to the preset pressure threshold, the data volume of the target time stage of the current day is compared with the data volume of the target time stage of each historical day to determine the first data concentration of the target time stage of the current day;

[0011] Determine a data synchronization strategy for the current time based on the first data concentration degree, so as to perform data security disaster recovery on the data center through the data synchronization strategy.

[0012] In some possible implementation manners, the day is divided into a plurality of time stages based on the data amount of each historical day in the historical time period, which can specifically include:

[0013] Determine the second data concentration degree of each hour based on the data amount of each hour of each historical day in the historical time period;

[0014] Curve fitting is performed on the second data concentration degree of each hour to obtain a target curve;

[0015] Adjacent maximum points in the target curve are divided into a time stage to obtain a plurality of time stages.

[0016] In some possible implementation manners, the second data concentration degree of each hour is determined based on the data amount of each hour of each historical day in the historical time period, which can specifically include:

[0017] The proportion of the data amount of the nth hour of each historical day in the total data amount of the historical day is determined as the data amount prominence of the nth hour of each historical day, and n is a positive integer;

[0018] The difference between the data amount prominences of adjacent historical days in the nth hour is used to determine a data concentration degree coefficient of the nth hour;

[0019] The minimum value in the data amount prominences and the data concentration degree coefficient are used to determine the second data concentration degree of the nth hour.

[0020] In some possible implementation manners, the future data pressure is predicted based on the data repetition feature and the data growth feature of the target time stage, which can specifically include:

[0021] Determine a target data repetition degree of the target time stage based on the data repetition feature of the target time stage;

[0022] Perform mean value processing on the second data concentration degree of each hour in the target time stage to obtain a first data growth length of the target time stage;

[0023] The future data pressure is predicted by using the target data repetition degree and the first data growth length.

[0024] In some possible implementation manners, the target data repetition degree of the target time stage is determined based on the data repetition feature of the target time stage, which can specifically include:

[0025] For each historical day, the following is performed: the appearance frequency of each data type of the historical day in the target time stage is subjected to standard deviation processing to obtain a first repetition coefficient of the historical day;

[0026] The first repetition coefficients of each historical day are subjected to mean processing to obtain a first data repetition degree of the target time stage;

[0027] For each data type, the following is performed: the appearance frequency of the data type in each historical day is subjected to standard deviation processing to obtain a second repetition coefficient of the data type;

[0028] The second repetition coefficients of each data type are subjected to mean processing to obtain a second data repetition degree of the target time stage;

[0029] The first data repetition degree and the second data repetition degree are used to determine a target data repetition degree of the target time stage.

[0030] In some possible implementation manners, the data amount of the current day in the target time stage is compared with the data amount of each historical day in the target time stage to determine a first data concentration degree of the current day in the target time stage, which can specifically include:

[0031] The second data concentration degree of each hour of the current day in the target time stage is compared with the corresponding second data concentration degree of each historical day in the target time stage to obtain a data concentration degree difference value;

[0032] The data concentration degree difference values are subjected to mean processing to obtain the first data concentration degree of the current day in the target time stage.

[0033] In some possible implementation manners, after the data amount of the current day in the target time stage is compared with the data amount of each historical day in the target time stage to determine the first data concentration degree of the current day in the target time stage, the data center data security disaster recovery method can further include:

[0034] Based on the number of packet losses of each historical day in the target time stage, a future packet loss rate is predicted;

[0035] Based on the future packet loss rate, a frequency adjustment coefficient of the data synchronization sending frequency of the current day in the target time stage is determined;

[0036] The frequency adjustment coefficient is used to adjust the data synchronization sending frequency of the current day in the target time stage.

[0037] In some possible implementation manners, based on the number of packet losses of each historical day in the target time stage, a future packet loss rate is predicted, which can specifically include:

[0038] obtaining the number of packet losses of each historical day in the target time stage, and the second data increment length of the current day in the target time stage;

[0039] averaging the number of packet losses of each historical day in the target time stage to obtain the average number of packet losses of the historical day;

[0040] comparing the second data increment length with the first data increment length to obtain an increment length difference value;

[0041] determining the future packet loss rate by using the increment length difference value and the average number of packet losses.

[0042] In some possible implementation manners, based on the future packet loss rate, a frequency adjustment coefficient of the data synchronization sending frequency of the current day in the target time stage is determined, and specifically can include the following steps:

[0043] obtaining the bandwidth occupation degree of the current moment and the packet loss data sequence of the target time stage, the packet loss data sequence including each packet loss moment in the target time stage;

[0044] determining a packet loss time interval feature value of the target time stage by using the packet loss data sequence of the target time stage;

[0045] determining the frequency adjustment coefficient by using the bandwidth occupation degree, the packet loss time interval feature value and the future packet loss rate.

[0046] A second aspect of the embodiment of the present application provides a data security disaster recovery system of a data center, which includes:

[0047] a data amount obtaining module, configured to obtain the data amount of the data center in the current day and the data amount of each historical day in a historical time period;

[0048] a stage dividing module, configured to divide one day into a plurality of time stages based on the data amount of each historical day in the historical time period, and the data amount of each hour in the time stage presents consistent volatility;

[0049] a pressure predicting module, configured to predict future data pressure based on the data repetition feature and the data growth feature of the target time stage, the target time stage being a time stage in which the current moment of the current day is located;

[0050] a concentration degree determining module, configured to, in the case that the future data pressure is less than or equal to a preset pressure threshold, compare the data amount of the current day in the target time stage with the data amount of each historical day in the target time stage to determine a first data concentration degree of the current day in the target time stage;

[0051] a strategy determining module, configured to determine a data synchronization strategy of the current moment based on the first data concentration degree, so as to perform data security disaster recovery on the data center by using the data synchronization strategy.

[0052] The present application has the following advantages:

[0053] In the data security disaster recovery method of the data center provided by the embodiment of the present application, the time period is divided according to the data volume of each historical day in the historical time period. Then, the future data pressure is predicted based on the data repetition feature and the data growth feature of the target time period corresponding to the current time of the current day. In the case that the future data pressure is less than or equal to the preset pressure threshold, the data volume of the target time period of the current day is compared with the data volume of the target time period of each historical day to determine the first data concentration degree of the target time period of the current day. Thus, the data synchronization strategy of the current time is selected according to the first data concentration degree of the target time period of the current day, so that the data security disaster recovery is performed on the data center through the data synchronization strategy. In this way, the present application predicts the concentration degree of the data of the target time period of the current day, selects a reasonable data synchronization strategy according to the actual demand, and dynamically adjusts the data synchronization strategy flexibly, thereby improving the data synchronization effect. BRIEF DESCRIPTION OF DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art and the advantages thereof, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of these drawings.

[0055] Figure 1 The flowchart of the first data security disaster recovery method of the data center provided by an embodiment of the present application;

[0056] Figure 2 The flowchart of S102 provided by an embodiment of the present application;

[0057] Figure 3 The time period diagram provided by an embodiment of the present application;

[0058] Figure 4 The flowchart of S103 provided by an embodiment of the present application;

[0059] Figure 5 The flowchart of S104 provided by an embodiment of the present application;

[0060] Figure 6 The flowchart of the second data security disaster recovery method of the data center provided by an embodiment of the present application;

[0061] Figure 7A structural diagram of a data security disaster recovery system of a data center provided by one embodiment of the present application. DETAILED DESCRIPTION

[0062] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined object of the application, the following describes in detail the specific embodiments, structure, features and effects of a data security disaster recovery method and system according to the present application, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0063] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0064] It should be noted that the acquisition, storage, use, processing, etc. of data in the technical solutions of the present application comply with the relevant provisions of laws and regulations.

[0065] In the complex and meticulous daily operation and management system of a data center, a data security disaster recovery strategy undoubtedly occupies a core position and is a solid cornerstone for guaranteeing data integrity, availability and business continuity. In today's highly digitized era, data has become one of the most valuable assets of an enterprise, and any loss, damage or unavailability of data can bring immeasurable losses to the enterprise, from financial losses to reputation damage, and even possibly leading to the complete paralysis of the business of the enterprise. Therefore, it is crucial for the survival and development of an enterprise to build a perfect and efficient data security disaster recovery strategy.

[0066] With the rapid development of information technology and the explosive growth of data volume, traditional physical disaster recovery measures have been difficult to meet the increasingly complex data security needs of enterprises. In order to achieve more efficient data security disaster recovery, the industry actively explores and widely adopts data synchronization technologies such as full synchronization or incremental synchronization. Full synchronization technology is like a comprehensive "data moving", which will copy all the data in the source data system to the target data system, ensuring that the target data system and the source data system are completely consistent at a certain moment. The advantage of this technology is simple operation and high data consistency, but the disadvantage is that when the data volume is large, the synchronization process will consume a lot of network bandwidth and storage resources, and the synchronization time is longer. Incremental synchronization technology is more "frugal", which only synchronizes the data that has changed since the last synchronization in the source data system. The advantage of this technology is fast synchronization speed and less resource consumption, but the disadvantage is that an additional mechanism is needed to record the changes in the data, and when the data is recovered, the incremental data needs to be applied in order according to the change order, the process is relatively complex. These two data synchronization technologies have their own advantages, enterprises can choose the appropriate synchronization technology according to their business needs, data characteristics and resource conditions, to effectively protect the consistency and real-time of data between different systems or storage locations.

[0067] However, the existing data synchronization scheme has a significant limitation, that is, it can usually only make a single choice between full synchronization and incremental synchronization, and cannot dynamically adjust between the two modes according to actual needs. In actual business scenarios, the data change situation is complex and variable. For example, during some business peak periods, the data change frequency will increase significantly, at this time, using incremental synchronization technology may cause the delay of data synchronization to increase, affecting the consistency of data; while in the business trough period, the data changes less, and using full synchronization technology will cause waste of resources. This lack of flexibility makes the data synchronization scheme unable to adapt to the data change characteristics in different business scenarios, often leading to poor data synchronization results, and failing to fully play the advantages of data synchronization technology.

[0068] The application aims to provide a data security disaster recovery method and system for a data center. The data security disaster recovery method provided by the application divides time stages according to the data volume of each historical day in a historical time period. Then, the future data pressure is predicted based on the data repetition feature and data growth feature of the target time stage corresponding to the current time of the current day. In the case that the future data pressure is less than or equal to a preset pressure threshold, the data volume of the current day in the target time stage is compared with the data volume of each historical day in the target time stage to determine the first data concentration degree of the current day in the target time stage. Then, the data synchronization strategy of the current time is selected according to the first data concentration degree of the current day in the target time stage, so that the data security disaster recovery of the data center is performed through the data synchronization strategy. In this way, the application can predict the concentration degree of the data of the current day in the target time stage, select a reasonable data synchronization strategy according to the actual demand, dynamically adjust the data synchronization strategy flexibly, and thus improve the data synchronization effect.

[0069] The following describes specific embodiments of the data security disaster recovery method and system for a data center provided by the application.

[0070] Figure 1 A flowchart of the data security disaster recovery method for a data center is provided. The data security disaster recovery method for a data center can be applied to a server. The data security disaster recovery method for a data center can include the following S101 to S105.

[0071] S101, obtaining the data volume of the data center in the current day and the data volume of each historical day in a historical time period.

[0072] In this embodiment, the historical time period is a time period set in advance. For example, the historical time period can be one month in the past.

[0073] As an example, special data collection agent programs can be deployed in the data center. These agent programs can be distributed on different nodes of the data center and are responsible for collecting the data volume information generated locally. For example, an agent program is installed on a database server to count the number of newly added data records and the size change of data files of the database every hour.

[0074] At the same time, a special data storage library is established to store the collected data volume information. A relational database (such as MySQL or Oracle) or a non-relational database (such as MongoDB or Redis) can be used to store the data volume information. The collected data volume information is stored in categories. The current day's data volume and the data volume of each historical day in the past month are stored in different data tables according to the date, which facilitates subsequent query and analysis.

[0075] S102, dividing a day into multiple time phases based on the data volume of each historical day in the historical time period, the data volume of each hour in a time phase presents consistent volatility.

[0076] In this embodiment, a time phase includes multiple hours with consistent volatility in data volume.

[0077] As an example, the server first cleans the data volume information of each historical day in the historical time period, removing outliers and error information. For example, for data volume information that deviates significantly from the normal range, it can be identified and removed by statistical analysis methods (such as the 3σ principle). Then, the average data volume of each hour in the historical time period is calculated by averaging the data volume of each hour in the historical time period.

[0078] Then, the clustering algorithm (such as K-means clustering algorithm) is used to analyze the clustering of the average data volume of each hour in the historical day. The average data volume of each hour is used as a feature vector, and the clustering center is adjusted by iteration. The 24 hours of a day are divided into multiple categories, each of which corresponds to a time phase.

[0079] Further, a volatility evaluation index such as standard deviation, coefficient of variation, etc. can be set to calculate the volatility of the average data volume of each hour in each time phase. According to the size of the volatility index, adjust the division of the time phase, and ensure that the average data volume of each hour in each time phase presents consistent volatility.

[0080] S103, predicting future data pressure based on the data repetition feature and data growth feature of the target time phase, the target time phase being the time phase in which the current time of the current day is located.

[0081] In this embodiment, the future data pressure is used to represent the flow pressure degree corresponding to the future data. As an example, the smaller the future data pressure, the smaller the flow pressure degree corresponding to the future data, and at this time, full data synchronization can be considered.

[0082] As an example, the server analyzes the data volume of each historical day in the target time phase, and counts the repeated data patterns, data blocks, etc. For example, in some business systems, a large number of similar transaction data records will be generated at a certain time period every day, and the data can be divided into blocks by using a hash algorithm, and the frequency of the same data block is counted.

[0083] At the same time, the growth rate of the data volume of each historical day in the target time phase is calculated, and the data growth trend is analyzed. Linear regression, exponential smoothing, etc. can be used to model the growth rate of the data volume to predict the growth of the future data volume.

[0084] Finally, a prediction model is constructed in combination with the data repetition feature and the data growth feature. For example, a neural network model (such as an LSTM neural network) can be used, the extracted features are taken as inputs, and the model is trained to predict the data pressure in the future for the current day.

[0085] S104, in the case where the future data pressure is less than or equal to the preset pressure threshold, comparing the data amount of the target time stage of the current day with the data amount of the target time stage of each historical day to determine the first data concentration degree of the target time stage of the current day.

[0086] In this embodiment, the first data concentration degree is used to represent the degree of concentrated distribution of data in the target time stage of the current day.

[0087] As an example, the server sets a reasonable preset pressure threshold according to the hardware resources (such as storage capacity, network bandwidth, processing capacity, etc.) of the data center and the business requirements.

[0088] In the case where the future data pressure is less than or equal to the preset pressure threshold, it means that the full-amount synchronization of data can be considered at this time. At this time, the similarity of the data amount of the target time stage of the current day and the data amount of the target time stage of each historical day is calculated. Specifically, cosine similarity, Euclidean distance, etc. can be used for similarity calculation.

[0089] Then, according to the similarity calculation result, the first data concentration degree of the target time stage of the current day is determined. For example, if the average similarity of the data amount of the current day and the data amount of each historical day is high, it means that the data distribution of the current day is more similar to the data distribution of the historical day, and thus the first data concentration degree of the target time stage of the current day is determined according to the data concentration distribution of the target time stage of the historical day.

[0090] As another example, in the case where the future data pressure is greater than the preset pressure threshold, it means that the future data corresponds to a greater traffic pressure, and at this time the incremental synchronization strategy can be directly determined to be used for data security disaster recovery.

[0091] S105, determining a data synchronization strategy for the current time based on the first data concentration degree, so as to perform data security disaster recovery on the data center through the data synchronization strategy.

[0092] In this embodiment, the data synchronization strategy includes a full-amount synchronization strategy and an incremental synchronization strategy. The full-amount synchronization strategy completely replicates all data, and the incremental synchronization strategy only synchronizes the data that has changed since the last synchronization.

[0093] As an example, when the first data concentration degree is less than or equal to a set threshold value, it indicates that the data performance based on the target time stage at the current time presents a relatively small concentration trend compared with the historical time stage, and therefore, under the premise of analyzing the future data pressure at the current time, a full-amount synchronization strategy can be selected for the future data to synchronize all data in the data center to the disaster recovery system, so as to ensure that the data in the disaster recovery system is completely consistent with the master data center; otherwise, if the first data concentration degree is greater than the set threshold value, only an incremental synchronization strategy is adopted to synchronize only the data changed on the current day, so as to reduce the time and resource consumption of data synchronization and avoid network delay.

[0094] As an optional embodiment, as shown in Figure 2 S102 can specifically include the following S201 to S203.

[0095] S201, determining a second data concentration degree of each hour based on the data amount of each hour in each historical day in a historical time period;

[0096] S202, performing curve fitting on the second data concentration degree of each hour to obtain a target curve;

[0097] S203, dividing adjacent maximum value points in the target curve into a time stage to obtain a plurality of time stages.

[0098] In this embodiment, the second data concentration degree is used to represent the average degree of concentration distribution of data in the corresponding hours of each historical day in the historical time period.

[0099] As an example, the server calculates the proportion of the data amount of each hour in the total amount of data of the corresponding historical day, and then performs mean processing on the proportion values corresponding to the same hours in each historical day to obtain the second data concentration degree of the corresponding hours.

[0100] Then, a suitable fitting function is selected. Common fitting functions include polynomial functions, exponential functions, logarithmic functions, etc. Specifically, the distribution trend of the data can be preliminarily judged by observing the scatter plot of the second data concentration degree, so as to select a suitable fitting function. For example, if the data presents a clear periodic fluctuation, a sine function or a cosine function can be considered for fitting; if the growth or decay trend of the data is relatively obvious, an exponential function can be considered for fitting.

[0101] Then, using the selected fitting function, the parameters of the fitting function are determined by using an optimization algorithm such as the least square method, so that the error between the fitting curve and the actual data points is minimized.

[0102] Finally, in the target curve obtained by fitting, the hour corresponding to the maximum value point found is recorded. The hour range between the adjacent two maximum value points is divided into a time phase. For example, if the maximum value points appear at the 3rd hour and the 15th hour respectively, the time period between the 3rd hour and the 15th hour (including the 3rd hour and excluding the 15th hour) is a time phase.

[0103] For example, as shown in the schematic diagram of a target curve provided in Figure 3 , the X axis is time and the Y axis is the concentration of the second data. The points B, D and F are maximum value points in the target curve, and the time phase between the points B and D and the time phase between the points D and F are divided. Meanwhile, the time phase between the remaining points A and B is also divided, so that the target curve is divided into multiple time phases.

[0104] According to the embodiment, the concentration of the second data is determined based on the data amount of each historical day per hour in the historical time period, which can comprehensively consider the data of the same time period on different days and avoid the one-sidedness that may be caused by only relying on single-day data. In this way, the accuracy of the division of the time phase can be improved.

[0105] As an optional embodiment, S201 can specifically include:

[0106] The proportion of the data amount of each historical day at the nth hour in the total data amount of the historical day is determined as the data amount prominence of each historical day at the nth hour, and n is a positive integer;

[0107] The difference between the data amount prominences of adjacent historical days at the nth hour is used to determine the data concentration degree coefficient of the nth hour;

[0108] The minimum value in the data amount prominences and the data concentration degree coefficient are used to determine the second data concentration degree of the nth hour.

[0109] In the embodiment, the data amount prominence of the historical day at the nth hour is the proportion of the data amount of the historical day at the nth hour in the total data amount of the historical day. For example, the total data amount of the historical day is Z, and the data amount of the historical day at the nth hour is z, and then the data amount prominence of the historical day at the nth hour is z divided by Z.

[0110] As an example, the data concentration degree coefficient can be determined by the following formula 1:

[0111]

[0112] In formula 1, U n is used to represent the data concentration degree coefficient of the nth hour, and M is used to represent the total number of days in the historical time period. qm,n q is used to represent the data amount prominence of the mth day at the nth hour. m+1,n norm is used to represent normalization processing. It should be noted that, in order to ensure that the calculation result is meaningful, when performing fractional operation, if the denominator is 0, a parameter adjustment factor greater than 0 is added to the denominator to prevent the denominator from being 0. The value of the parameter adjustment factor is set by the implementer according to the actual situation, and the present application does not make special limitations.

[0113] For a large-scale data change scenario, it usually maintains the same data amount feature in a specific period, so the balanced trend of the data amount prominence of each historical day at the same hour in the historical period can be considered. This means that the more obvious the data amount prominence of the same hour in multiple days is, the more the data at the same hour in multiple days presents the trend of data concentration at the time.

[0114] Then, the second data concentration degree can be determined by the following formula 2:

[0115] Q n = (min m∈[1,M] q m,n )*(1+U n ) Formula 2

[0116] In formula 2, Q n is used to represent the second data concentration degree of the nth hour, U n is used to represent the data concentration degree coefficient of the nth hour, and q m,n is used to represent the data amount prominence of the mth day at the nth hour. M is used to represent the total number of historical days in the historical period, min m∈[1,M] q m,n is used to represent the minimum value of the data amount prominence of each historical day.

[0117] Through the present embodiment, the data amount of each hour of each historical day in the historical period can be accurately determined, so as to accurately determine the second data concentration degree of each hour. Therefore, an accurate target curve can be generated, which is helpful for subsequent reasonable division of the time phase and improves the accuracy of the division of the time phase.

[0118] As an optional embodiment, as shown in FIG. 4, S103 can specifically include the following S401 to S403. Figure 4

[0119] S401, determining a target data repetition degree of the target time phase based on the data repetition feature of the target time phase;

[0120] ​S402, average the second data concentration per hour in the target time period to obtain a first data growth length of the target time period;

[0121] S403, predict the future data pressure using the target data repetition degree and the first data growth length.

[0122] In this embodiment, the target data repetition degree is used to represent the degree of repeated occurrence of data in the target time period, and the first data growth length is used to represent the degree of data growth of the historical day in the target time period in the historical period.

[0123] As an example, the server calculates the frequency and regularity of each data repetition for the target time period, such as the number of times of repeated occurrence in the target time period every day. Specifically, the proportion of repeated data to total data can be used to represent the degree of repeated occurrence, that is, the number of repeated data of each historical day in the target time period is divided by the total amount of data of the historical day in the target time period, and then the average value is calculated, that is, the target data repetition degree of the target time period can be obtained.

[0124] Then, the second data concentration per hour in the target time period is extracted, and the second data concentration per hour in the target time period is averaged, that is, the first data growth length of the target time period can be obtained.

[0125] Then, a suitable prediction model is selected, such as a linear regression model, a neural network model, etc. For example, if there is a linear relationship between the future data pressure and the target data repetition degree and the first data growth length, a linear regression model is selected. The historical data is used to train the prediction model, the historical target data repetition degree and the first data growth length are used as input, and the actual data pressure is used as output, and the model parameters are determined by least square method or the like.

[0126] Finally, the target data repetition degree and the first data growth length calculated at present are substituted into the trained prediction model to calculate the predicted value of the future data pressure.

[0127] As another example, the future data pressure can be determined by the following formula 3:

[0128] Y=D*(1-C)+(1-D)*Z Formula 3

[0129] In formula 3, Y is used to represent the future data pressure, D is used to represent the bandwidth occupation degree at the current time, C is used to represent the target data repetition degree, and Z is used to represent the first data growth length.

[0130] The greater the bandwidth occupation degree D at the current moment and the smaller the target data repetition degree at the target time stage, the greater the future data pressure; the smaller the bandwidth occupation degree D at the current moment and the greater the first data increase length at the target time stage, the greater the future data pressure.

[0131] Through the embodiment, the future data pressure is predicted in combination with the target data repetition degree and the first data increase length, the repetition feature of data and the data growth trend in the time stage are fully considered, deviation caused by single factor prediction is avoided, and the accuracy of future data pressure prediction is improved.

[0132] As an optional embodiment, S401 can specifically include:

[0133] For each historical day, the following is performed: the appearance frequency of each data type of the historical day in the target time stage is subjected to standard deviation processing to obtain a first repetition coefficient of the historical day;

[0134] The first repetition coefficients of each historical day are subjected to mean value processing to obtain a first data repetition degree of the target time stage;

[0135] For each data type, the following is performed: the appearance frequency of the data type in each historical day is subjected to standard deviation processing to obtain a second repetition coefficient of the data type;

[0136] The second repetition coefficients of each data type are subjected to mean value processing to obtain a second data repetition degree of the target time stage;

[0137] The target data repetition degree of the target time stage is determined by using the first data repetition degree and the second data repetition degree.

[0138] In the embodiment, the first data repetition coefficient reflects the dispersion degree of the appearance frequency of the data type in the target time stage of the historical day, and the second data repetition coefficient reflects the dispersion degree of the appearance frequency of the data type in each historical day in the target time stage.

[0139] The first data repetition degree reflects the fluctuation of the data type in each historical day, and the second data repetition degree reflects the stability of the data type between different historical days, and the two can more accurately describe the repetition mode of data.

[0140] As an example, the server collects data of each historical day in the target time stage, and divides the data into multiple data types according to different data.

[0141] For each historical day, the number of occurrences of each data type in the target time stage is counted, and the occurrence frequency is calculated. The occurrence frequency = the number of occurrences of a certain data type / the total number of data in the target time stage of the historical day. Then, for each historical day, the standard deviation of the occurrence frequency of each data type in the target time stage is calculated to obtain the first repetition coefficient of each historical day. Then, the first repetition coefficients of the historical days are averaged to obtain the first data repetition degree of the target time stage.

[0142] For each data type, the occurrence frequency of each data type in each historical day is sorted. Then, for each data type, the standard deviation of the occurrence frequency in each historical day is calculated to obtain the second repetition coefficient of each data type. Then, the second repetition coefficients of each data type are averaged to obtain the second data repetition degree of the target time stage.

[0143] Finally, the first data repetition degree and the second data repetition degree can be combined to determine the target data repetition degree of the target time stage by using a weighted average method. Alternatively, the first data repetition degree and the second data repetition degree can be multiplied to obtain the target data repetition degree of the target time stage.

[0144] Through the embodiment, the first data repetition degree and the second data repetition degree are calculated from two dimensions of historical days and data types, which can more comprehensively depict the repetition characteristics of data in the target time stage. In this way, the calculation accuracy of the data repetition degree can be improved.

[0145] As an optional embodiment, as shown in Figure 5 S104 can specifically include the following S501-S502.

[0146] S501, compare the second data concentration degree of each hour in the target time stage of the current day with the second data concentration degree corresponding to each historical day in the target time stage, respectively, to obtain a data concentration degree difference value;

[0147] S502, average the data concentration degree difference values to obtain the first data concentration degree of the current day in the target time stage.

[0148] In the embodiment, the data concentration degree difference value is used to represent the difference between the second data concentration degree of the current day in the target time stage and the second data concentration degree corresponding to the historical day in the target time stage. For example, the second data concentration degree of the nth hour in the target time stage of the current day is 0.7, and the second data concentration degree of the nth hour in the target time stage of the historical day is 0.6, and the corresponding data concentration degree difference value is 0.1.

[0149] As an example, the server calculates the data concentration difference between the data concentration of the current day and the second data concentration of each historical day for each hour in the target time period.

[0150] Finally, the average of the data concentration differences is calculated to obtain the first data concentration of the current day in the target time period.

[0151] Through this embodiment, by calculating the data concentration difference, the difference between the data concentration of the current day and the corresponding hour of each historical day in the target time period can be intuitively reflected. This quantitative method helps to better understand the uniqueness of the current day data and provides a basis for the selection of subsequent data synchronization strategy.

[0152] As an optional embodiment, as shown in Figure 6 After S105, the data center data security disaster recovery method can further include S601-S603.

[0153] S601, based on the number of packet loss of each historical day in the target time period, predicting the future packet loss rate;

[0154] S602, based on the future packet loss rate, determining a frequency adjustment coefficient of the data synchronization sending frequency of the current day in the target time period;

[0155] S603, using the frequency adjustment coefficient to adjust the data synchronization sending frequency of the current day in the target time period.

[0156] In this embodiment, the frequency adjustment coefficient is used to represent the coefficient of adjusting the data synchronization sending frequency of the current day in the target time period.

[0157] As an example, the server first collects the number of packet loss of each historical day in the target time period. For example, in a network communication scenario, the number of data packet loss of each historical day in a specific time period (such as 9:00-11:00 every day) is recorded. The average number of packet loss of each historical day in the target time period is calculated.

[0158] Then, according to the characteristics of the data and the prediction requirements, a suitable prediction model is selected. Common models include time series analysis models (such as ARIMA model), regression models (such as linear regression, polynomial regression) or machine learning models (such as neural network). For example, assuming that there is a linear relationship between the future packet loss rate and the average number of packet loss of the historical day, a linear regression model is selected.

[0159] Then, the selected prediction model is trained using historical data to determine the parameters of the prediction model. Finally, the trained prediction model is applied to the current average number of packet losses to predict the future packet loss rate.

[0160] Then, an adjustment rule for the data synchronization sending frequency is defined according to the size of the future packet loss rate. For example, a threshold range can be set, and when the future packet loss rate is below a certain threshold, the data synchronization sending frequency is appropriately increased; when the future packet loss rate is above a certain threshold, the data synchronization sending frequency is appropriately reduced.

[0161] Finally, the data synchronization sending frequency in the target time stage of the current day is adjusted according to the frequency adjustment coefficient to obtain the adjusted data synchronization sending frequency. Specifically, the data synchronization sending frequency in the target time stage of the current day can be adjusted by the following formula 4:

[0162] P = p1 + (p2 - p1) * (1 - F) Formula 4

[0163] In formula 4, P is used to represent the adjusted data synchronization sending frequency, p1 is used to represent the minimum value of the data synchronization sending frequency, p2 is used to represent the maximum value of the data synchronization sending frequency, (p2 - p1) is used to represent the adjustable range of the data synchronization sending frequency, and F is used to represent the frequency adjustment coefficient.

[0164] Through this embodiment, based on predicting the future packet loss rate and adjusting the data synchronization sending frequency, the data transmission can better adapt to the changes in network conditions. When the future packet loss rate is predicted to be high, reducing the sending frequency can reduce the conflict and loss of data packets, and improve the success rate of data transmission.

[0165] As an optional embodiment, S601 can specifically include:

[0166] Obtaining the number of packet losses in the target time stage of each historical day, and the second data increment length in the target time stage of the current day;

[0167] The number of packet losses in the target time stage of each historical day is averaged to obtain the average number of packet losses of the historical day;

[0168] The second data increment length is compared with the first data increment length to obtain an increment length difference;

[0169] The increment length difference and the average number of packet losses are used to determine the future packet loss rate.

[0170] In this embodiment, the second data growth length is used to represent the degree of data growth of the target time stage of the current day. Specifically, the second data set density of each hour contained in the target time stage of the current day is extracted, and the second data set density of each hour contained in the target time stage is averaged, that is, the second data growth length of the target time stage of the current day is obtained.

[0171] As an example, the server collects the number of packet loss times of each historical day in the target time stage, and the second data growth length of the target time stage of the current day.

[0172] Then, the number of packet loss times of each historical day in the target time stage is averaged to calculate the average number of packet loss times of the historical day; at the same time, the second data growth length is compared with the first data growth length to obtain a length difference value.

[0173] Then, the length difference value is multiplied by the average number of packet loss times to obtain the initial packet loss rate of the target time stage of the current day. Finally, the initial packet loss rate is multiplied by the future data pressure to obtain the future packet loss rate.

[0174] Through this embodiment, the data growth and historical packet loss are comprehensively considered, which can more comprehensively evaluate the future packet loss rate. The change of the data growth length reflects the change of the network load or the traffic, and the historical packet loss times provide historical information of network stability, and the combination of the two can more accurately predict the future packet loss, thereby improving the calculation accuracy of the future packet loss rate.

[0175] As an optional embodiment, S602 can specifically include:

[0176] Obtain the bandwidth occupation degree of the current time and the packet loss data sequence of the target time stage, and the packet loss data sequence includes each packet loss time in the target time stage;

[0177] Determine the packet loss time interval feature value of the target time stage by using the packet loss data sequence of the target time stage;

[0178] Determine the frequency adjustment coefficient by using the bandwidth occupation degree, the packet loss time interval feature value, and the future packet loss rate.

[0179] In this embodiment, the packet loss data sequence includes each packet loss time in the target time stage, and each packet loss time is arranged in the packet loss data sequence in the order from early to late.

[0180] As an example, the server first obtains the bandwidth occupation degree of the current time, and constructs the corresponding packet loss data sequence according to the historical time period and each packet loss time of the current day.

[0181] Then, the packet loss time interval between two adjacent packet losses in the packet loss data sequence is obtained, so as to obtain a plurality of packet loss time interval types, and the occurrence number of each packet loss time interval type is counted.

[0182] The packet loss time interval representative degree of each packet loss time interval type and the interval length of each packet loss time interval type are combined to determine the packet loss time interval characteristic value of the target time stage by using the weighted average method and norm normalization.

[0183] Finally, the bandwidth occupation degree, the packet loss time interval characteristic value and the future packet loss rate are used to determine the frequency adjustment coefficient of the data synchronization sending frequency in the target time stage of the current day by using the following formula 5:

[0184] F=T*H+(1-T)*D Formula 5

[0185] In formula 5, F is used to represent the frequency adjustment coefficient of the data synchronization sending frequency in the target time stage of the current day, T is used to represent the packet loss time interval characteristic value of the target time stage, H is used to represent the future packet loss rate, and D is used to represent the bandwidth occupation degree at the current time.

[0186] Wherein, the greater the packet loss time interval characteristic value T of the target time stage and the greater the future packet loss rate, the greater the frequency adjustment coefficient should be; the smaller the packet loss time interval characteristic value T of the target time stage, but the greater the bandwidth occupation degree at the current time, the greater the frequency adjustment coefficient should be.

[0187] By the embodiment, the bandwidth occupation degree, the packet loss time interval characteristic value and the future packet loss rate can be used to accurately determine the frequency adjustment coefficient of the data synchronization sending frequency in the target time stage of the current day. In this way, it is helpful to adjust the data synchronization sending frequency in the target time stage of the current day according to the frequency adjustment coefficient, so that the data transmission can better adapt to the change of network condition.

[0188] A data security disaster recovery method based on a data center is provided. Accordingly, specific embodiments of a data security disaster recovery system of a data center are also provided.

[0189] As shown in Figure 7 A structural schematic diagram of a data security disaster recovery system of a data center is provided. The data security disaster recovery system 700 of the data center includes a data amount acquisition module 710, a stage division module 720, a pressure prediction module 730, a concentration degree determination module 740 and a strategy determination module 750.

[0190] The data volume acquisition module 710 is configured to acquire data volumes of the data center on a current day and historical days in a historical time period.

[0191] The stage division module 720 is configured to divide a day into a plurality of time stages based on the data volumes of the historical days in the historical time period, and the data volume in each time stage presents consistent volatility every hour.

[0192] The pressure prediction module 730 is configured to predict future data pressure based on the data repetition feature and the data growth feature of a target time stage, the target time stage being a time stage in which the current time of the current day is located.

[0193] The concentration degree determination module 740 is configured to, in a case where the future data pressure is less than or equal to a preset pressure threshold, compare the data volume of the target time stage of the current day with the data volumes of the target time stage of the historical days, and determine a first data concentration degree of the target time stage of the current day.

[0194] The strategy determination module 750 is configured to determine a data synchronization strategy of the current time based on the first data concentration degree, so as to perform data security disaster recovery on the data center through the data synchronization strategy.

[0195] It should be noted that the present application is not limited to the specific configurations and processes described above and shown in the drawings. For the sake of brevity, detailed descriptions of well-known methods are omitted herein. In the above embodiments, several specific steps are described and shown as examples. However, the method processes of the present application are not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order of the steps, after understanding the spirit of the present application.

[0196] It should also be noted that the exemplary embodiments mentioned in the present application describe some methods or systems based on a series of steps or devices. However, the present application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiments, or in an order different from the embodiments, or several steps can be performed simultaneously.

[0197] The above is merely specific implementation of the present application, and those skilled in the art can clearly understand that, for the sake of convenience and brevity of description, the specific working processes of the above-described systems, modules and units can refer to the corresponding processes in the foregoing method embodiments, which will not be described herein. It should be understood that the protection scope of the present application is not limited thereto, and any skilled person in the art can easily think of various equivalent modifications or replacements within the technical range disclosed by the present application, and these modifications or replacements should be covered within the protection scope of the present application.

Claims

1. A data security disaster recovery method for a data center, characterized in that, The method includes: Get the data volume of the data center on the current day and the data volume of each historical day within the historical time period; Based on the data volume of each historical day within the historical time period, a day is divided into multiple time stages, and the data volume of each hour within the time stage exhibits consistent fluctuation. Based on the data repetition and growth characteristics of the target time period, the future data pressure is predicted, where the target time period is the time period in which the current moment of the current day is located. If the future data pressure is less than or equal to a preset pressure threshold, the amount of data on the current day in the target time period is compared with the amount of data on each of the historical days in the target time period to determine the first data concentration of the current day in the target time period. Based on the first data centrality, a data synchronization strategy for the current moment is determined so that the data center can be protected against data security disaster recovery through the data synchronization strategy.

2. The data security disaster recovery method for data centers according to claim 1, characterized in that, Based on the amount of data for each historical day within the historical time period, a day is divided into multiple time segments, including: Based on the amount of data per hour for each historical day within the historical time period, determine the second data concentration for each hour; Curve fitting is performed on the hourly concentration of the second data to obtain the target curve; The time period is divided into two segments by dividing the time between adjacent maximum points in the target curve. This results in multiple time segments.

3. The data security disaster recovery method for data centers according to claim 2, characterized in that, The determination of the second data concentration for each hour based on the amount of data per hour for each historical day within the historical time period includes: The proportion of the data volume of each historical day in the nth hour to the total data volume of the historical day is determined as the prominence of the data volume of each historical day in the nth hour, where n is a positive integer; The data concentration coefficient for the nth hour is determined by using the difference in the data salience of adjacent historical days in the nth hour. The second data concentration for the nth hour is determined by using the minimum value of each of the data prominence values ​​and the data concentration coefficient.

4. The data security disaster recovery method for data centers according to claim 1, characterized in that, The prediction of future data pressure based on the data repetition and growth characteristics of the target time period includes: Based on the data repetition characteristics of the target time period, the target data repetition degree of the target time period is determined; The average value of the second data concentration in each hour within the target time period is processed to obtain the first data growth rate of the target time period. The future data pressure is predicted by using the target data redundancy and the first data growth rate.

5. The data security disaster recovery method for data centers according to claim 4, characterized in that, Determining the target data repetition degree for the target time period based on the data repetition characteristics of the target time period includes: For each historical day, the following steps are performed: the frequency of occurrence of each data type on the historical day within the target time period is processed by standard deviation to obtain the first repetition coefficient of the historical day; The first repetition coefficient of each of the historical days is averaged to obtain the first data repetition degree of the target time period. For each of the data types, the following steps are performed: the frequency of occurrence of the data type on each of the historical days is processed by standard deviation to obtain the second repetition coefficient of the data type; The second repetition coefficient of each data type is averaged to obtain the second data repetition degree of the target time period. The target data repetition for the target time period is determined by using the first data repetition and the second data repetition.

6. The data security disaster recovery method for data centers according to claim 1, characterized in that, The step of comparing the amount of data for the current day in the target time period with the amount of data for each of the historical days in the target time period to determine the first data concentration of the current day in the target time period includes: The second data concentration of each hour of the current day within the target time period is compared with the second data concentration of each historical day within the target time period to obtain the data concentration difference. The mean of the differences in the centrality of each dataset is used to obtain the first centrality of the dataset for the current day in the target time period.

7. The data security disaster recovery method for data centers according to claim 1, characterized in that, After comparing the amount of data for the current day in the target time period with the amount of data for each of the historical days in the target time period to determine the first data concentration of the current day in the target time period, the method further includes: Based on the number of packet losses on each historical day within the target time period, predict the future packet loss rate; Based on the future packet loss rate, determine the frequency adjustment coefficient for the data synchronization transmission frequency of the current day within the target time period; The frequency adjustment coefficient is used to adjust the data synchronization transmission frequency of the current day within the target time period.

8. The data security disaster recovery method for data centers according to claim 7, characterized in that, The prediction of future packet loss rate based on the number of packet losses on each of the historical days within the target time period includes: Obtain the number of packet losses for each historical day within the target time period, and the second data growth rate of the current day within the target time period; The average number of packet losses for each historical day within the target time period is calculated by averaging the number of packet losses for that historical day. The growth rate of the second data is compared with the growth rate of the first data to obtain the growth rate difference; The future packet loss rate is determined by using the growth rate difference and the average number of packet losses.

9. The data security disaster recovery method for data centers according to claim 7, characterized in that, The step of determining the frequency adjustment coefficient for the data synchronization transmission frequency of the current day within the target time period based on the future packet loss rate includes: Obtain the bandwidth usage at the current moment and the packet loss data sequence for the target time period, wherein the packet loss data sequence includes each packet loss moment within the target time period; Using the packet loss data sequence of the target time period, determine the packet loss time interval characteristic value of the target time period; The frequency adjustment coefficient is determined using the bandwidth occupancy, the packet loss time interval characteristic value, and the future packet loss rate.

10. A data security disaster recovery system for a data center, characterized in that, The system includes: The data volume acquisition module is used to acquire the data volume of the data center on the current day and the data volume of each historical day within the historical time period; The phase division module is used to divide a day into multiple time phases based on the data volume of each historical day within the historical time period, and the data volume of each hour within the time phase exhibits consistent fluctuation. The pressure prediction module is used to predict future data pressure based on the data repetition characteristics and data growth characteristics of the target time stage, wherein the target time stage is the time stage in which the current moment of the current day is located. The concentration determination module is used to compare the amount of data of the current day in the target time period with the amount of data of each of the historical days in the target time period when the future data pressure is less than or equal to a preset pressure threshold, and determine the first data concentration of the current day in the target time period. The strategy determination module is used to determine the data synchronization strategy at the current moment based on the first data concentration, so as to perform data security disaster recovery on the data center through the data synchronization strategy.

Images