Cloud storage data access method and device

Abstract

This application discloses a cloud storage data access method and apparatus, applied in the field of cloud service technology. The method includes: acquiring historical behavior information of a target terminal, wherein the historical behavior information consists of parameters associated with the target terminal's cloud storage data access behavior within a historical time period; obtaining target behavior information and several target access data of the target terminal based on the historical behavior information, wherein the target behavior information consists of parameters associated with the target terminal's cloud storage data access behavior within a future time period; determining multiple target preloaded data based on the target behavior information and the several target access data; performing preload processing on each target preloaded data using the target behavior information to obtain each preloaded target preloaded data; and performing access processing on each preloaded target preloaded data in response to an access command from the target terminal. This application can effectively reduce data latency in cloud storage services and improve the user experience of cloud storage services.

Description

Technical Field

[0001] This application relates to the field of cloud service technology, and in particular to cloud storage data access methods and apparatus. Background Technology

[0002] Cloud storage services are a technology that stores data on remote servers via the internet. By storing data on remote servers, it provides users with a cost-effective, flexible, and scalable storage solution. Currently, although cloud storage services offer users a convenient way to store and access data, data latency remains a significant issue, thus degrading the user experience. Summary of the Invention

[0003] This application provides cloud storage data access methods and apparatus to reduce data latency in cloud storage services and improve the user experience of cloud storage services.

[0004] On the one hand, embodiments of this application provide a cloud storage data access method, including the following steps:

[0005] Obtain historical behavior information of the target terminal; wherein, the historical behavior information is a parameter associated with the cloud storage data access behavior of the target terminal within a historical time period;

[0006] Based on the historical behavior information, target behavior information of the target terminal and several target access data are obtained; wherein, the target behavior information are parameters associated with the cloud storage data access behavior of the target terminal in a future time period;

[0007] Based on the target behavior information and several target access data, multiple target preload data are determined;

[0008] The target behavior information is used to preload the target preload data to obtain the preloaded target preload data.

[0009] In response to the access command of the target terminal, access processing is performed on each preloaded target preloaded data.

[0010] On the other hand, embodiments of this application provide a cloud storage data access device, including:

[0011] The acquisition module is used to acquire historical behavior information of the target terminal; wherein, the historical behavior information is a parameter associated with the cloud storage data access behavior of the target terminal within a historical time period;

[0012] The first processing module is used to obtain target behavior information of the target terminal and several target access data based on the historical behavior information; wherein, the target behavior information is a parameter associated with the cloud storage data access behavior of the target terminal in a future time period;

[0013] The second processing module is used to determine multiple target preload data based on the target behavior information and several target access data.

[0014] The third processing module is used to preload each target preload data using the target behavior information to obtain each preloaded target preload data.

[0015] The fourth processing module is used to perform access processing on the pre-loaded target data in response to the access command of the target terminal.

[0016] According to the cloud storage data access method and apparatus of this application, firstly, historical behavior information of the target terminal is obtained, which refers to parameters associated with the cloud storage data access behavior of the target terminal within a historical time period; secondly, based on the historical behavior information, target behavior information and several target access data of the target terminal are obtained, which refers to parameters associated with the cloud storage data access behavior of the target terminal within a future time period; then, based on the target behavior information and several target access data, multiple target preloaded data are determined; subsequently, each target preloaded data is preloaded using the target behavior information to obtain each preloaded target preloaded data; finally, in response to the access command of the target terminal, each preloaded target preloaded data is accessed. This effectively reduces the data latency of the cloud storage service, speeds up the response to data access requests of the cloud storage service, and thus improves the user experience of the cloud storage service.

[0017] Other features and advantages of this application will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the description, claims and drawings. Attached Figure Description

[0018] Figure 1 This is a flowchart of a cloud storage data access method provided in this application;

[0019] Figure 2 This is a flowchart of the prediction of future access needs of the target terminal provided in this application;

[0020] Figure 3 This is a flowchart of a preloading process provided in this application;

[0021] Figure 4 This is a schematic diagram of the principle for determining the preloading parameters provided in this application;

[0022] Figure 5 This is another flowchart of the preloading process provided in this application;

[0023] Figure 6 This is a schematic diagram illustrating the relationship between data, queues, media, and terminals provided in this application;

[0024] Figure 7 This is a flowchart of the preloading optimization provided in this application;

[0025] Figure 8 This is a structural diagram of the cloud storage data access device provided in this application;

[0026] Figure 9 This is a schematic diagram of the cloud storage data access method provided in this application;

[0027] Figure 10 This is another flowchart of the cloud storage data access method provided in this application. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0029] The present application will be further described below with reference to the accompanying drawings and specific embodiments. The described embodiments should not be considered as limitations on the present application, and all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of the present application.

[0030] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0031] 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 this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0032] With the development of Internet Technology (IT) and the significant decrease in storage and computing costs, cloud computing has gradually become widespread and a mainstream Internet technology service model. The core concept of cloud computing is to provide computing resources such as servers, storage, networks, and applications as a service to users via the Internet. Users can dynamically adjust computing resources according to their actual needs without pre-configuring hardware. This model not only reduces the cost of Internet services for users but also improves the utilization and flexibility of computing resources.

[0033] Among the many services offered by cloud computing, cloud storage services play a crucial role. With the acceleration of digital transformation, data volumes are exploding, and users' demand for storage is increasing daily. Cloud storage services provide users with a cost-effective, flexible, and scalable storage solution by storing data on remote servers managed by cloud service providers. Cloud storage services offer reliable security, data redundancy, and disaster recovery capabilities, ensuring continuous data availability and security.

[0034] Cloud storage services have a broad user base, including individual users, SMEs, large enterprises, and government agencies. The needs and data access behaviors of different users vary significantly. Data access behavior refers to the patterns of user behavior when accessing resources such as computer systems, network services, and databases. It typically includes data such as the time, frequency, purpose, and operations of data access, which can reveal a user's data access habits. In practical applications, individual users often focus more on storage space size, price, and data synchronization capabilities, while enterprise users tend to have higher requirements for data security, access speed, and application programming interface (API) integration. Furthermore, even within the same user group, there are significant differences in access behavior among different individuals. For example, within an enterprise user group, some departments may frequently access financial statements, while others may focus more on product design documents.

[0035] Due to the complexity of data access behavior, data access exhibits non-uniformity and volatility. Specifically, firstly, data access is not evenly distributed; it is non-uniform, often exhibiting distinct hotspots and access peaks. For example, on weekday mornings and afternoons, enterprise users' access demand increases significantly, while on holidays and weekends, individual users' access to multimedia content such as photos and videos increases. Secondly, data access behavior typically changes dynamically with time or specific events; that is, data access is volatile. For instance, specific events such as corporate restructuring, new business launches, and holiday promotions can easily cause changes in data access behavior.

[0036] As users' expectations for data access speeds increase, data latency has become a key factor affecting the user experience of cloud storage services. The non-uniformity and volatility of data access negatively impact data latency. Currently, cloud storage services still suffer from significant data latency issues, leading to a degraded user experience, especially in scenarios with high concurrency or large data transfers.

[0037] In view of this, embodiments of this application provide a cloud storage data access method and apparatus, which aim to predict the data access behavior of a target terminal in a future time period through intelligent prediction technology, and then preload the data that the target terminal needs to access in the future time period into storage media such as cache servers and edge nodes based on the prediction results, thereby reducing the data latency of cloud storage services, improving the response speed of data access, and providing users with a smoother and faster data access experience.

[0038] First, the implementation steps of the cloud storage data access method provided in the embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0039] The cloud storage data access method provided in this application can be applied to terminals, servers, or software running on either terminal or server. Terminals can be tablets, laptops, desktop computers, etc., but are not limited to these. Servers can be independent physical servers, server clusters or distributed systems composed of multiple physical servers, or cloud servers providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. Furthermore, a server can be a node server in a blockchain network, but is not limited to this. Blockchain is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms.

[0040] Reference Figure 1 , Figure 1 This is a flowchart of a cloud storage data access method provided in this application, which may include the following steps S101-S105:

[0041] S101, Obtain historical behavior information of the target terminal; wherein, the historical behavior information is a parameter associated with the cloud storage data access behavior of the target terminal within a historical time period;

[0042] S102, Based on historical behavior information, obtain target behavior information of the target terminal and several target access data; wherein, the target behavior information is a parameter associated with the cloud storage data access behavior of the target terminal in the future time period;

[0043] S103, Based on the target behavior information and several target access data, determine multiple target preload data;

[0044] S104, Use target behavior information to preload the preloaded data of each target to obtain the preloaded data of each target;

[0045] S105, in response to the access command of the target terminal, performs access processing on each preloaded target preloaded data.

[0046] In this embodiment, firstly, historical behavior information of the target terminal is obtained, which refers to parameters associated with the target terminal's cloud storage data access behavior within a historical time period. Secondly, based on the historical behavior information, target behavior information and several target access data are obtained. The target behavior information refers to parameters associated with the target terminal's cloud storage data access behavior within a future time period, and the target access data refers to the data objects that the target terminal expects to access within the future time period. Then, based on the target behavior information and several target access data, multiple target preloaded data are determined, which refers to preloaded data objects. Subsequently, the target behavior information is used to preload each target preloaded data to obtain each preloaded target preloaded data. Finally, in response to the target terminal's access command, each preloaded target preloaded data is accessed.

[0047] As can be seen, the embodiments of this application determine the target behavior information and several target access data of the target terminal through the historical behavior information of the target terminal. Then, using the target behavior information and several target access data, multiple target preloaded data are determined. Before the data access request of the target terminal arrives, the target behavior information is used to perform preloading processing on each target preloaded data. Finally, when the data access request of the target terminal arrives, the preloaded target preloaded data is accessed, thereby realizing data access to the cloud storage service. In this way, the data latency of the cloud storage service can be effectively reduced, especially the data latency in scenarios with high concurrency access or large data volume transmission, and the response to the data access request of the cloud storage service can be accelerated, thereby improving the user experience of the cloud storage service.

[0048] In step S101 above, parameters associated with the cloud storage data access behavior of the target terminal within a historical time period are obtained as historical behavior information, so as to accurately capture the access behavior and access needs of the target terminal in future time periods through historical behavior information in subsequent steps.

[0049] The aforementioned target terminal refers to a terminal applicable to the cloud storage data access method in the embodiments of this application. The type of terminal can be flexibly set according to the actual situation. For example, the terminal can be a tablet computer, a laptop computer, a desktop computer, etc., but is not limited to this.

[0050] The aforementioned historical behavior information refers to parameters associated with the cloud storage data access behavior of the target terminal within a historical time period. Cloud storage data access behavior refers to the behavioral patterns of users when accessing cloud storage service resources.

[0051] The aforementioned historical behavior information can be set according to actual circumstances, and this application embodiment does not impose specific limitations on it.

[0052] For example, in some embodiments, historical behavior information may include behavioral characteristic data, time characteristic data, user characteristic data, and context characteristic data of the target terminal within a historical time period. Specifically, the behavioral characteristic data of the target terminal may include the target terminal's access records, access data types, access data parameters, access operation types, clickstream data, and search query counts, etc., wherein the access operation types may include, but are not limited to, read operations, download operations, and upload operations; clickstream data is used to indicate the access trajectory of the target terminal when accessing data, and may be a Pageviews table or a Visits table, etc., but is not limited to these. The time characteristic data of the target terminal may include the target terminal's data access time value, data access date, and data access holidays, etc., and these parameters can be used to capture the periodic changes in the target terminal's data access behavior. The user characteristic data of the target terminal may include user information of the target terminal, such as the user's occupation, gender, and preference information. The context characteristic data of the target terminal may include the target terminal's device type, browser version number, etc., and these parameters can help understand the hardware and software environment of the target terminal when accessing data.

[0053] The method for obtaining the aforementioned historical behavior information can be set according to the actual situation, and this application embodiment does not impose specific limitations on it. For example, the historical behavior information of the target terminal can be obtained by analyzing the log files of the target terminal within a historical time period; or, the historical behavior information of the target terminal can be directly obtained through a database used to store the historical behavior information of several terminals, but is not limited to this.

[0054] In step S102 above, after obtaining historical behavior information, the historical behavior information and intelligent prediction technology are used to obtain target behavior information and several target access data, so as to accurately capture the access behavior and access needs of the target terminal in the future time period.

[0055] The aforementioned target access data refers to the data objects that the target terminal may expect to access within a future time period.

[0056] The aforementioned target behavior information refers to parameters associated with the cloud storage data access behavior of the target terminal within a future time period. Cloud storage data access behavior refers to the behavioral patterns of users when accessing cloud storage service resources.

[0057] The aforementioned target behavior information can be set according to actual circumstances, and this application embodiment does not impose specific limitations on it.

[0058] For example, in some embodiments, target behavior information may include the access frequency of each target access data, the access pattern of each target access data, the data type of each target access data, the peak data access time of the target terminal, the potential demand data of the target terminal, and the churn risk data of the target terminal. Specifically, the access frequency of target access data refers to the difference between the access volume of target access data in a future time period and the duration of that future time period; the access pattern of target access data may include the access path of the target access data; the data type of target access data can be flexibly set according to actual conditions, for example, the data type of target access data may be image type, text type, video type, etc., but is not limited to these; the peak data access time of the target terminal refers to the sub-time period with the highest data access volume in a future time period, with the unit of the sub-time period being hours; the potential demand data of the target terminal refers to the type of product or service that the target terminal expects in a future time period, which can indicate the products or services that the target terminal may be interested in in a future time period; the churn risk data of the target terminal refers to the probability that the target terminal will not use cloud storage services in a future time period, which can indicate the target terminal that may churn.

[0059] The above-mentioned acquisition of target behavior information and several target access data of the target terminal based on historical behavior information may include, but is not limited to, acquiring target behavior information and several target access data of the target terminal based on historical behavior information combined with machine learning methods.

[0060] The machine learning methods described above can be set according to the actual situation, and the embodiments of this application do not impose specific limitations on them.

[0061] For example, the machine learning method described above could be logistic regression; or, it could be support vector machine, but it is not limited to these.

[0062] In step S103 above, the target behavior information and several target access data are processed in order to determine multiple target preload data from the several target access data.

[0063] The aforementioned target preloaded data refers to the data objects that participate in the preload processing. It can also be understood as the data objects that the target terminal expects to access within a future time period.

[0064] The above-mentioned determination of multiple target preload data based on target behavior information and several target access data may include obtaining the access probability of several target access data by combining the target behavior information and several target access data with machine learning methods. The access probability refers to the probability that the target access data will be accessed in the future time period. Then, target access data with access probabilities greater than a preset probability threshold are selected as target preload data, but it is not limited to this.

[0065] In step S104 above, after determining multiple target preloaded data, the target behavior information is used to preload each target preloaded data, which aims to preload each target preloaded data in the cloud storage service to storage media such as cache servers and edge nodes before the data access request of the target terminal arrives.

[0066] The aforementioned preloading process refers to loading data into storage media such as cache servers and edge nodes in advance.

[0067] The aforementioned preloaded target data refers to the target preloaded data that has been preloaded to storage media such as cache servers and edge nodes.

[0068] The above-mentioned preloading process of preloading data for each target using target behavior information may include determining the preloading time based on the target behavior information and combining it with machine learning methods, and loading the preloading data of each target to the preset edge node when the preloading time of the target terminal arrives, but is not limited to this.

[0069] In step S105 above, the target preloaded data of the target terminal has been preloaded before the data access request from the target terminal arrives. Therefore, when the access instruction from the target terminal is received, it indicates that the data access request from the target terminal has arrived. At this time, in response to the access instruction from the target terminal, each preloaded target preloaded data is obtained and sent to the target terminal. It can be understood that after obtaining each preloaded target preloaded data, the target terminal can perform operations such as decryption and decompression on each preloaded target preloaded data and render and display it.

[0070] The access instructions for the target terminal mentioned above are used to indicate the data access request of the target terminal.

[0071] The steps described above will be explained in further detail below.

[0072] In some implementations, refer to Figure 2 In step S102 above, the specific implementation process of obtaining the target behavior information of the target terminal and several target access data based on historical behavior information may include the following steps S1021-S1023:

[0073] S1021, Preprocess the historical behavior information to obtain preprocessed historical behavior information;

[0074] S1022, Extract features from the preprocessed historical behavior information to obtain historical behavior features;

[0075] S1023, predict historical behavioral characteristics to obtain target behavior information and several target access data.

[0076] In this embodiment, firstly, historical behavior information is preprocessed to effectively ensure its accuracy and reduce the interference of redundant and useless data on intelligent prediction, thereby obtaining preprocessed historical behavior information. Then, feature extraction is performed on the preprocessed historical behavior information to capture feature information associated with the target terminal's cloud storage data access behavior within a historical time period from various types of historical behavior information, thus obtaining historical behavior features. Next, the historical behavior features are input into a pre-trained prediction model, which accurately captures the target terminal's access behavior and access needs within a future time period, thereby obtaining target behavior information and several target access data. Thus, this embodiment can effectively and accurately capture the target terminal's access behavior and access needs within a future time period, enabling data preloading in subsequent steps based on the target terminal's access behavior and access needs within that future time period.

[0077] The aforementioned historical behavior information may include, but is not limited to, behavioral characteristic data, time characteristic data, user characteristic data, and context characteristic data of the target terminal within a historical time period.

[0078] The implementation method of the above preprocessing can be set according to the actual situation, and this embodiment does not limit it in detail. For example, preprocessing may include data cleaning such as missing value handling, outlier handling and duplicate value handling, and data transformation such as normalization, discretization and mathematical transformation, but is not limited to these.

[0079] The implementation method of the above feature extraction can be set according to the actual situation, and this implementation method does not make specific limitations on it.

[0080] For example, in some embodiments, the aforementioned historical behavioral features include behavioral features, temporal features, user features, and contextual features. Feature extraction is performed by extracting features from the preprocessed behavioral characteristic data, preprocessed temporal characteristic data, preprocessed user characteristic data, and preprocessed contextual characteristic data, respectively, to obtain behavioral features, temporal features, user features, and contextual features. Feature extraction can be implemented using natural language processing techniques such as bag-of-words models and word embeddings, or it can be implemented using other methods such as convolutional neural networks and one-hot encoding, but is not limited to these.

[0081] For example, in some embodiments, historical behavioral features include a behavioral temporal feature sequence, user features, and contextual features. The feature extraction method is as follows: first, the preprocessed behavioral feature data and preprocessed temporal characteristic data are integrated into behavioral temporal data, and features are extracted from the behavioral temporal data to obtain a behavioral temporal feature sequence. Then, features are extracted from the preprocessed user characteristic data to obtain user features. Finally, features are extracted from the preprocessed contextual characteristic data to obtain contextual features. Feature extraction can be implemented using natural language processing techniques such as bag-of-words models and word embeddings, or other methods such as convolutional neural networks and one-hot encoding, but is not limited to these.

[0082] The aforementioned target behavior information may include, but is not limited to, the access frequency of each target's access data, the access pattern of each target's access data, the data type of each target's access data, and the peak data access time of the target terminal.

[0083] The implementation method of the above prediction can be set according to the actual situation, and this implementation method does not make specific limitations on it.

[0084] For example, in some embodiments, historical behavioral features may include behavioral features, time features, user features, and contextual features. The prediction method is as follows: the behavioral features, time features, user features, and contextual features are integrated into a feature sequence and input into a first prediction model. The first prediction model is used for prediction processing to obtain several target access data, the access frequency of each target access data, the access pattern of each target access data, the data type of each target access data, and the peak data access time of the target terminal.

[0085] The first prediction model can be a model trained from multiple preset first feature sequences and the corresponding label information of each preset first feature sequence. The preset first feature sequences can be represented as {preset behavioral features, preset time features, preset user features, and preset context features}. The label information corresponding to the first feature sequences can be represented as {multiple preset access data, access frequency of each preset access data, access pattern of each preset access data, data type of each preset access data, and preset peak access time}. For example, if the preset access data includes access data 1 and access data 2, then the label information corresponding to the first feature sequences can be represented as {access data 1, access frequency of access data 1, access pattern of access data 1, data type of access data 1; access data 2, access frequency of access data 2, access pattern of access data 2, data type of access data 2, and peak access time for both access data 1 and access data 2}. Furthermore, the first prediction model can be a neural network model such as Long Short-Term Memory (LSTM) or Recurrent Neural Network (RNN), or other machine learning models such as Random Forest or Support Vector Machine, but is not limited to these.

[0086] For example, in some embodiments, intelligent prediction can be achieved by using model ensemble to target the characteristics of different features. For instance, a prediction model adept at handling time-series data can be used to predict time-series data, while a prediction model adept at handling non-time-series data can be used to predict non-time-series data, thereby fully leveraging the advantages of each prediction model and improving overall prediction performance. Furthermore, a phased prediction strategy can be adopted. First, preliminary predictions can be made using some historical behavioral features to obtain some future data. Then, more accurate subsequent predictions can be made using the remaining historical behavioral features and some future data. Based on this, the model ensemble approach is combined with the phased prediction strategy. The prediction method is as follows: First, a second prediction model is used to predict the sequence of behavioral time-series features to obtain several target access data, the access frequency of each target access data, the data type of each target access data, and the peak data access time of the target terminal. Then, a third prediction model is used to predict the access frequency of each target access data, the peak data access time of the target terminal, user characteristics, and contextual features to obtain the access pattern of each target access data.

[0087] The second prediction model can be a model trained from multiple preset behavioral time-series feature sequences and the corresponding label information of each preset behavioral time-series feature sequence. The label information corresponding to the behavioral time-series feature sequences can be represented as {multiple preset access data, access frequency of each preset access data, data type of each preset access data, and preset peak data access time}. For example, if the preset access data are access data 1 and access data 2, then the label information corresponding to the behavioral time-series feature sequences can be represented as {access data 1, access frequency of access data 1, data type of access data 1, access data 2, access frequency of access data 2, data type of access data 2, and peak data access time for access data 1 and access data 2}. Furthermore, the second prediction model can be a neural network model that is adept at processing time-series data, such as a Long Short-Term Memory network or a Recurrent Neural Network, but is not limited to these.

[0088] The third prediction model can be a model trained from multiple preset second feature sequences and the corresponding label information of each preset second feature sequence. The preset second feature sequence can be represented as {access frequency of each preset access data, preset peak access time, user characteristics, context features}, and the corresponding label information can be represented as {access pattern of each preset access data}. For example, if the preset access data are access data 1 and access data 2, then the second feature sequence can be represented as {access frequency of access data 1, access frequency of access data 2, peak access time of access data 1 and access data 2, user characteristics, context features}, and the corresponding label information can be represented as {access pattern of access data 1, access pattern of access data 2}. Furthermore, the third prediction model can be a machine learning model that excels at handling non-time-series data, such as random forests, decision trees, support vector machines, etc., but is not limited to these.

[0089] In some implementations, the target behavior information of the target terminal may include the access frequency of each target access data; in step S103, the specific implementation process of determining multiple target preloaded data based on the target behavior information and several target access data may include the following step S1031:

[0090] S1031, Based on the access frequency of each target access data, and combined with the timeliness coefficient and user preference coefficient of each target access data, determine multiple target preloaded data from several target access data.

[0091] In this embodiment, the access frequency of each target access data predicted in the aforementioned steps is used, combined with the timeliness coefficient and user preference coefficient of each target access data, to select target access data that meets preset conditions from several target access data sets as target preloaded data, thereby obtaining multiple target preloaded data sets. Thus, this embodiment, by fully considering the access behavior and access needs of the target terminal in the future time period, as well as multi-dimensional factors such as data timeliness and data user preference, accurately selects the data objects that the target terminal expects to access in the future time period from several target access data sets, i.e., the target preloaded data. This effectively improves the accuracy of the target preloaded data, making it more closely aligned with the data access needs of the target terminal.

[0092] The timeliness coefficient of the target access data is positively correlated with the effectiveness of the target access data; that is, the higher the timeliness coefficient of the target access data, the more effective the target access data.

[0093] The user propensity coefficient of the target access data is used to indicate the similarity between the target access data and the data access needs of the target terminal. The higher the user propensity coefficient of the target access data, the closer the target access data is to the data access needs of the target terminal.

[0094] The timeliness coefficient and user preference coefficient of the target access data can be pre-calibrated according to the actual situation, and this embodiment does not impose specific limitations on them.

[0095] The selection method for the aforementioned target preloaded data can be set according to the actual situation, and this implementation method does not limit it.

[0096] For example, in some embodiments, target access data that meets the preloading conditions is selected from a plurality of target access data as target preloading data, thereby obtaining a plurality of target preloading data; wherein, the preloading conditions include at least one of the following: access frequency greater than a preset frequency threshold, timeliness coefficient greater than a preset first coefficient threshold, or user preference coefficient greater than a preset second coefficient threshold.

[0097] For example, in some embodiments, firstly, based on the access frequency, timeliness coefficient, and user preference coefficient of each target access data, and combined with machine learning methods, the importance coefficient of each target access data is obtained, wherein the importance coefficient is used to indicate the importance of the target preloaded data to the target terminal; then, target access data with an importance coefficient greater than a preset importance threshold are selected as target preloaded data, thereby obtaining multiple target preloaded data.

[0098] In some implementations, refer to Figure 3In step S104 above, the specific implementation process of preloading the preloaded data of each target using the target behavior information to obtain the preloaded data of each target may include the following steps S1041-S1042:

[0099] S1041, Based on the target behavior information and preset comprehensive parameters, the preloading parameters are obtained; wherein, the preloading parameters include the target storage medium of each target preloaded data, the target storage format of each target preloaded data, and the preloading time of the target terminal;

[0100] S1042, preload the preload data of each target using the preload parameters to obtain the preloaded target data.

[0101] In this embodiment, after determining the target preloaded data, target behavior information and preset comprehensive parameters are used as the parameter acquisition benchmark to determine the target storage medium, target storage format, and preload time of the target terminal for each target preloaded data. These preload parameters are used to preload the target preloaded data. By fully considering the access behavior and access needs of the target terminal within a future time period, as well as other related parameters, the determination of the preload parameters can be achieved, effectively improving the accuracy of the preload parameters and thus improving the precision of the preload processing. After obtaining the preload parameters, each target preloaded data is preloaded using these parameters to obtain the preloaded target preloaded data. This allows the target preloaded data in the cloud storage service to be preloaded to storage media such as cache servers and edge nodes before the data access request from the target terminal arrives, effectively reducing data latency in the cloud storage service and improving the user experience.

[0102] The aforementioned comprehensive parameters refer to the parameters associated with the target terminal, the preloaded data of each target, and each preset storage medium. These parameters can be pre-calibrated or acquired in real time according to the actual situation, and this embodiment does not impose specific limitations on them.

[0103] The aforementioned preloading parameters may include, but are not limited to, the target storage medium for each target preloaded data, the target storage format for each target preloaded data, and the preloading time of the target terminal.

[0104] The target storage medium for the aforementioned target preloaded data refers to the storage medium used to store the target preloaded data. It is understood that the type of target storage medium can be set according to actual circumstances, and this implementation does not impose specific limitations on it. For example, the target storage medium can be an edge node of a cloud storage service. An edge node refers to a business platform built at the network edge close to the user, providing storage, computing, and network resources, and offloading some critical business applications to the edge of the access network to reduce bandwidth and latency losses caused by network transmission and multi-level forwarding. As another example, the target storage medium can be a high-speed cache server for a cloud storage service. A high-speed cache server is a hardware device that stores and provides data to accelerate data access and improve system performance.

[0105] The target storage format of the aforementioned target preloaded data refers to the data format of the target preloaded data during preloading processing. It is understood that the target storage format can be set according to actual circumstances, and this embodiment does not impose specific limitations on it. For example, the target storage format can be image format, text format, and video format, but it is not limited to these.

[0106] The preloading time of the target terminal mentioned above refers to the time value for preloading the target preload data.

[0107] The above-mentioned preloaded parameters are obtained based on the target behavior information and the preset comprehensive parameters. This can include obtaining the preloaded parameters by combining the target behavior information and the preset comprehensive parameters with machine learning methods, but is not limited to this.

[0108] In some implementations, refer to Figure 4 The above comprehensive parameters may include the importance coefficient of each target preloaded data, the usage frequency of each target preloaded data, the terminal geographical location of the target terminal, the network characteristic data of the target terminal, the load rate of each preset storage medium, and the terminal geographical location of each preset storage medium.

[0109] In step S1041 above, the specific implementation process of obtaining the preload parameters based on the target behavior information and preset comprehensive parameters may include the following steps:

[0110] Based on the importance coefficient and usage frequency of each target preloaded data, combined with the target terminal's geographical location, network characteristic data of the target terminal, load rate of each preset storage medium, and terminal geographical location of each preset storage medium, the target storage medium for each target preloaded data is determined from multiple preset storage media.

[0111] In this embodiment, the selection of target storage media involves several steps. First, acquiring the target terminal's geographical location and network characteristics data, as well as the terminal geographical location of each preset storage medium, allows for dynamic selection of the storage medium closest to the target terminal to reduce latency. Next, acquiring the load rate of each preset storage medium allows for dynamic selection of storage media with lower load; the load rate indicates the load level of the preset storage medium. Finally, acquiring the importance coefficient and usage frequency of each target preloaded data allows for the allocation of storage media with lower load and sufficient storage space to high-priority target preloaded data. Then, based on this data, corresponding target storage media are allocated to each target preloaded data.

[0112] As can be seen, this implementation method, by fully considering multiple factors such as data importance, data usage frequency, terminal location, terminal network performance, node load rate, and node location, determines the target storage medium for each target preloaded data. This not only improves the compatibility between the target preloaded data and the target storage medium, balances the network load of the target storage medium, reduces data latency, and optimizes the storage space utilization efficiency of the target storage medium, reducing resource waste caused by excessive preloading, but also improves the preloading processing and accuracy, which is conducive to achieving better preloading results. This reduces the data latency of cloud storage services, meets users' data access needs, and provides a solid foundation for cloud storage data access.

[0113] It should be noted that different target preloaded data are allowed to share the same target storage medium.

[0114] The aforementioned comprehensive parameters may include, but are not limited to, the importance coefficient of each target preloaded data, the usage frequency of each target preloaded data, the geographical location of the target terminal, the network characteristic data of the target terminal, the load rate of each preset storage medium, and the geographical location of each preset storage medium.

[0115] The importance coefficient of the aforementioned target preloaded data is used to indicate the degree of importance of the target preloaded data to the target terminal. The higher the importance coefficient of the target preloaded data, the more important the target preloaded data is to the target terminal. Furthermore, the method for obtaining the importance coefficient of the target preloaded data can be set according to actual circumstances, and this embodiment does not impose specific limitations on it. For example, the importance coefficient can be a pre-calibrated value. Or, for example, the importance coefficient of the target preloaded data can be obtained based on the access frequency, timeliness coefficient, and user preference coefficient of the target preloaded data, combined with machine learning methods, but it is not limited to these methods.

[0116] The usage frequency of the target preloaded data mentioned above refers to the usage rate of the target preloaded data per unit time. It can be a pre-calibrated value, but is not limited to it.

[0117] The aforementioned target terminal's geographical location refers to the location of the target terminal. It is understood that the target terminal's geographical location can be flexibly obtained based on actual circumstances, and this implementation method does not impose specific limitations on it. For example, the target terminal's geographical location can be determined using technologies such as Geographic Information System (GIS) or information such as Internet Protocol (IP) addresses and Global Positioning System (GPS), but it is not limited to these methods.

[0118] The network characteristic data of the target terminal mentioned above is used to indicate the network transmission performance of the target terminal. The higher the network characteristic data of the target terminal, the better the network transmission performance of the target terminal. It is understood that the network characteristic data of the target terminal can be real-time acquired network characteristic data of the target terminal, or it can be the average value of the network characteristic data of the target terminal over a historical period, but it is not limited to these. In addition, the specific type of the network characteristic data of the target terminal can be set according to the actual situation, and this embodiment does not impose specific limitations on it. For example, the network characteristic data of the target terminal can be the uplink speed and downlink speed of the target terminal, but it is not limited to these.

[0119] The aforementioned preset storage medium refers to the storage medium applicable to the cloud storage data access method in this application embodiment. It can be a preset storage medium, but is not limited to it. Furthermore, the type of the preset storage medium can be flexibly set according to actual conditions. For example, the preset storage medium can be a high-speed cache server, or other storage media such as edge nodes. This embodiment does not specifically limit this.

[0120] The aforementioned terminal geographic location of the preset storage medium refers to the location of the preset storage medium. It is understood that the terminal geographic location of the preset storage medium can be flexibly obtained or preset according to actual circumstances, and this embodiment does not impose specific limitations on this. For example, the terminal geographic location of the preset storage medium can be determined through technologies such as Geographic Information Systems (GIS) and Global Positioning Systems (GPS), but is not limited to these methods.

[0121] The load rate of the aforementioned preset storage medium is used to indicate the load level of the storage medium. The higher the load rate of the preset storage medium, the more severe the load on the preset storage medium. In addition, the load rate of the aforementioned preset storage medium can be the load rate of the preset storage medium obtained in real time, or it can be the average load rate of the preset storage medium over a historical period of time, but it is not limited to these two.

[0122] The method for determining the target storage medium can be set according to the actual situation, and this embodiment does not impose specific limitations on it.

[0123] For example, in some embodiments, firstly, based on the importance coefficient and usage frequency of each target preloaded data, and combined with machine learning methods, the priority of each target preloaded data is obtained. The priority of the target preloaded data is positively correlated with the terminal need and timeliness coefficient of the target preloaded data. The terminal need refers to the degree to which the target terminal requires the target preloaded data; that is, the higher the priority of the target preloaded data, the more the target terminal needs the target preloaded data, and the higher the timeliness coefficient of the target preloaded data. Then, optimization algorithms such as ant colony optimization and genetic algorithms are used to optimize the priority of each target preloaded data, the geographical location of the target terminal, the network characteristic data of the target terminal, and the load rate of each preset storage medium to obtain the target storage medium for each target preloaded data. It can be understood that the optimization object of the optimization algorithm is the optimal preset storage medium for each target preloaded data, i.e., the target storage medium for each target preloaded data. Furthermore, the objective function and constraints of the objective function of the optimization algorithm can be flexibly set according to actual conditions.

[0124] For example, the objective function can include three sub-objective functions. The first sub-objective function minimizes the location distance between the target storage medium and the target terminal for each target preloaded data. This distance can be obtained by presetting the terminal's geographical location on both the storage medium and the target terminal. The second sub-objective function maximizes the network characteristic data of the target terminal. The third sub-objective function minimizes the load rate of the target storage medium for each target preloaded data. The first and second sub-objective functions can reduce data latency, while the third sub-objective function can achieve node load balancing, reducing latency and performance degradation caused by overloaded data access requests on high-load nodes.

[0125] The constraints of the objective function can include a positive correlation between the priority of the target preloaded data and the storage space of the preset storage medium; that is, the higher the priority of the target preloaded data, the larger the storage space of the preset storage medium. Furthermore, the constraints can be negatively correlated with the load rate of the preset storage medium; that is, the higher the priority of the target preloaded data, the lower the load rate of the preset storage medium. This constraint can be understood as meaning that high-priority data should be preferentially stored in storage media with lower load and sufficient storage space.

[0126] For example, in some embodiments, for each target preloaded data, based on the importance coefficient of the target preloaded data, the usage frequency of the target preloaded data, the terminal geographical location of the target terminal, the network characteristic data of the target terminal, the load rate of each preset storage medium and the terminal geographical location of each preset storage medium, combined with machine learning methods, a storage coefficient of each preset storage medium relative to the target preloaded data is obtained. The storage coefficient can map the performance of the preset storage medium when the preset storage medium stores the target preloaded data, and then the preset storage medium with the highest storage coefficient is selected as the target storage medium for the target preloaded data.

[0127] In some implementations, refer to Figure 4 The target behavior information of the aforementioned target terminal may include the data type, access mode, and access frequency of each target preloaded data; the aforementioned comprehensive parameters may also include the sensitivity coefficient and media load rate of each target preloaded data; in step S1041, the specific implementation process of obtaining the preloaded parameters based on the target behavior information and the preset comprehensive parameters may further include the following steps:

[0128] Based on the data type, access mode, and access frequency of each target preloaded data, and combined with the sensitivity coefficient and media load rate of each target preloaded data, the target storage format of each target preloaded data is determined.

[0129] In this embodiment, the determination of the target storage format involves first obtaining the data type, access mode, and access frequency of each target preloaded data predicted in the aforementioned steps, and simultaneously obtaining the sensitivity coefficient and media load rate of each target preloaded data. The sensitivity coefficient refers to the severity of the consequences caused by the leakage of target access data; the higher the sensitivity coefficient, the more severe the consequences of the leakage of target access data. The media load rate refers to the load rate of the target storage medium. Then, the target storage format of each target preloaded data is determined based on these data.

[0130] Specifically, the target storage format is mainly divided into two aspects: encryption and compression methods, and storage type.

[0131] For encrypted compression methods, a suitable compression algorithm can be dynamically selected based on the data type and access frequency of the target preloaded data. This compresses the target preloaded data, saving storage space on the target storage medium and improving preload speed and subsequent data access speed. The compression algorithm can be flexibly configured according to actual conditions. For example, for target preloaded data with an access frequency higher than a preset threshold, a lower compression ratio can be selected as the target storage format to improve decompression speed during subsequent data access, thus ensuring data access speed.

[0132] Simultaneously, the sensitivity coefficient of the target preloaded data can be used to dynamically select a suitable encryption algorithm to encrypt the data, thereby ensuring its security. The encryption algorithm can be flexibly configured according to the actual situation. For example, for target preloaded data with a sensitivity coefficient higher than a preset sensitivity threshold, a strong encryption algorithm such as Advanced Encryption Standard (AES) can be used for encryption, while for target preloaded data with a sensitivity coefficient lower than the threshold, a lightweight encryption algorithm can be used to ensure good performance.

[0133] Regarding storage type, the appropriate storage type can be dynamically selected based on the data type, access mode, and media load rate of the target preloaded data to optimize preload speed and storage efficiency. For example, the target preloaded data can be preloaded and stored as a lower-resolution image to improve preload speed and subsequent data access speed. Furthermore, the original type of target preloaded data can be used on high-bandwidth storage media, while compressed target preloaded data can be used on storage media with limited bandwidth.

[0134] As can be seen, this implementation method, by fully considering multi-dimensional factors such as data type, data access mode, data access frequency, data sensitivity coefficient, and node load rate, determines the target storage format for each target preloaded data. This not only improves the compatibility between the target preloaded data and the target storage medium and reduces resource waste caused by excessive preloading, but also improves the efficiency and accuracy of preloading processing, which is conducive to achieving better preloading results. This reduces data latency in cloud storage services, meets users' data access needs, and provides a solid foundation for cloud storage data access.

[0135] The aforementioned comprehensive parameters may also include the sensitivity coefficients and media load rates of the preloaded data for each target, but are not limited to these.

[0136] The sensitivity coefficient mentioned above refers to the severity of the consequences caused by the leakage of target access data. The higher the sensitivity coefficient of the target access data, the more severe the consequences caused by the leakage of target access data. It can be a pre-defined value, but is not limited to it.

[0137] The aforementioned media load rate refers to the load rate of the target storage medium. It can be real-time data, but is not limited to this.

[0138] The method for determining the target storage format can be set according to the actual situation, and this embodiment does not impose specific limitations on it.

[0139] For example, in some embodiments, for each target preloaded data, the target storage format of the target preloaded data is obtained by combining machine learning methods based on the data type, access mode, sensitivity coefficient, and media load rate of the target preloaded data.

[0140] The target storage format includes, but is not limited to, encrypted image format, compressed image format, encrypted text format, compressed text format, encrypted video format, compressed video format, and raw format. It can be understood that encrypted image format refers to an image type processed by an encryption algorithm, and the same applies to encrypted text and encrypted video formats. Compressed image format refers to an image type processed by a compression algorithm, and the same applies to compressed text and compressed video formats. Ordinary image format refers to the raw format without encryption or compression processing.

[0141] For example, in some embodiments, for each target preloaded data, the encryption and compression method of the target preloaded data is determined based on the data type, access frequency, and sensitivity coefficient of the target preloaded data, combined with machine learning methods. The encryption and compression method includes any one of encryption, compression, or no processing. The no processing method can be understood as a method that does not perform any encryption and compression processing. Then, based on the encryption and compression method of the target preloaded data, the data type of the target preloaded data, the access mode of the target preloaded data, and the media load rate of the target preloaded data, combined with machine learning methods, the storage type of the target preloaded data is obtained. The storage type includes any one of image type, text type, or video type. Finally, based on the storage type and encryption and compression method of the target preloaded data, the target storage format of the target preloaded data is obtained.

[0142] The target storage formats include, but are not limited to, encrypted image formats, plain image formats, compressed image formats, encrypted text formats, plain text formats, compressed text formats, encrypted video formats, plain video formats, and compressed video formats. It can be understood that encrypted image formats refer to storage types that are image types and encryption / compression methods that are encrypted; the same applies to encrypted text and encrypted video formats. Compressed image formats refer to storage types that are image types and encryption / compression methods that are compressed; the same applies to compressed text and compressed video formats. Plain image formats refer to storage types that are image types and encryption / compression methods that are not processed; the same applies to plain text and compressed video formats.

[0143] In some implementations, refer to Figure 4 The target behavior information of the target terminal may also include the peak data access period of the target terminal; the comprehensive parameters may also include the network characteristic data of the target terminal; the specific implementation process of obtaining the preloaded parameters based on the target behavior information and the preset comprehensive parameters in step S1041 may also include the following steps:

[0144] Based on the peak data access times of the target terminal and combined with the network characteristics data of the target terminal, the preloading time of the target terminal is obtained.

[0145] In this embodiment, the determination of the preloading time involves first obtaining the peak data access period of the target terminal predicted in the aforementioned steps, and simultaneously acquiring the network characteristic data of the target terminal. Then, based on the peak data access period and network characteristic data, the preloading time of the target terminal is dynamically selected. For example, large-scale data preloading is performed during periods when network characteristic data is optimal and not during peak data access times, thereby reducing network occupancy and the impact on user experience. It is evident that this embodiment determines the preloading time by fully considering multiple factors such as peak data access times and terminal network performance. This improves the compatibility between preloading processing and cloud storage data access timing, reduces resource waste caused by excessive preloading, and enhances the efficiency and accuracy of preloading processing, leading to better preloading results. This, in turn, reduces data latency in cloud storage services, meets user data access needs, and provides a solid foundation for cloud storage data access.

[0146] The method for determining the preloading time can be set according to the actual situation, and this embodiment does not impose specific limitations on it.

[0147] For example, in some embodiments, the preloading time of the target terminal is obtained by combining machine learning methods with peak data access times and network characteristic data of the target terminal.

[0148] For example, in some embodiments, a preset time period with the highest network characteristic data is selected from several preset time periods as a candidate time period, and then the candidate time period closest to the peak data access time period is selected as the preloading time of the target terminal.

[0149] In some implementations, refer to Figure 5 and Figure 6 In step S1042 above, the specific implementation process of preloading each target preload data using preload parameters to obtain each preloaded target preload data may include the following steps S01-S03:

[0150] S01, transform the storage format of the target preloaded data into the target storage format of the target preloaded data;

[0151] S02, load the target preloaded data into the preload queue corresponding to the target preloaded data;

[0152] S03, when the preloading time of the target terminal arrives, the target preload data is mapped from the preload queue corresponding to the target preload data to the target storage medium of the target preload data to obtain the preloaded target preload data; wherein, the preloaded target preload data is stored in the target storage medium of the target preload data in the target storage form of the target preload data.

[0153] In this embodiment, during the preloading process, for each target preloaded data, the following applies:

[0154] First, the target preloaded data undergoes a data storage format transformation based on its target storage format. For example, if the current storage format of the target preloaded data is an unencrypted or uncompressed image, while the target storage format is text processed by an encryption algorithm, the preloaded data is transformed from an unencrypted or uncompressed image to a text-processed image during preloading. Similarly, if the current storage format of the target preloaded data is an unencrypted or uncompressed image, while the target storage format is an image processed by a compression algorithm, the preloaded data is transformed from an unencrypted or uncompressed image to an image-processed image during preloading.

[0155] After the data storage format transformation is completed, several preloading queues are obtained. Each preloading queue is used to preload the corresponding target preloading data. The preloading queue corresponding to the target preloading data is then selected from these queues, and the target preloading data is loaded into the corresponding queue. It is understood that the selection of preloading queues can be flexibly configured according to actual conditions. For example, idle preloading queues can be randomly assigned to the target preloading data, while full preloading queues are not allocated. Furthermore, the preloading queues corresponding to each target preloading data run in parallel to ensure that different target preloading data can be processed concurrently, shortening the overall preloading processing time and improving the user experience.

[0156] Next, the system waits for the target terminal's preloading time to arrive. When the target terminal's preloading time arrives, the target preloaded data is mapped from the preload queue corresponding to the target preloaded data to the target storage medium of the target preloaded data, so that the target preloaded data is stored in the target storage medium in the target storage format, thus obtaining the preloaded target preloaded data. By traversing all the target preloaded data, multiple preloaded target preloaded data sets can be obtained.

[0157] As can be seen, this implementation transforms the storage format of the target preloaded data into its target storage format before the preload time arrives, and uses a preload queue as an intermediate carrier to load the target preloaded data into the preload queue corresponding to the target preloaded data. When the preload time of the target terminal arrives, the target preloaded data is mapped from the preload queue corresponding to the target preloaded data to the target storage medium of the target preloaded data, thereby realizing the preload processing of the target preloaded data. This can effectively improve the efficiency of the preload processing, ensuring that the preload processing of each target preloaded data is completed before the target terminal accesses the data. At the same time, it can effectively improve the accuracy of the preload processing, ensuring that each target preloaded data can be preloaded into the corresponding target storage medium, thereby helping to reduce the data latency of the cloud storage service, speed up the response to the data access request of the cloud storage service, and improve the user experience of the cloud storage service.

[0158] It should be noted that, when there is only one target terminal and the target terminal corresponds to at least one target preloaded data, the relationship between the target preloaded data, the preload queue, the target storage medium, and the target terminal is as follows: One target preloaded data corresponds to one preload queue, and a preload queue is allowed to preload one or more target preloaded data, meaning that different target preloaded data can correspond to the same preload queue. A preload queue can correspond to the target storage medium of one or more target preloaded data, that is, the same preload queue is allowed to have a mapping relationship with one or more target storage media. A target terminal can correspond to the target storage medium of one or more target preloaded data, that is, a target terminal is allowed to be configured with one or more target storage media. Furthermore, when facing multiple target terminals, one target storage medium can correspond to one or more target terminals, that is, different target terminals are allowed to share the same target storage medium.

[0159] Exemplarily, in some embodiments, reference is made to Figure 7 Terminal 1 corresponds to target preloaded data 1 and target preloaded data 2. Target preloaded data 1 and target preloaded data 2 can share preload queue 1. The target storage medium for target preloaded data 1 is edge node 1, and the target storage medium for target preloaded data 2 is edge node 2. During preloading, target preloaded data 1 and target preloaded data 2 are preloaded into preload queue 1 and wait for the preload time to arrive. When the preload time arrives, target preloaded data 1 in preload queue 1 is stored in edge node 1, and target preloaded data 2 in preload queue 1 is stored in edge node 2. During data access, terminal 1 can access target preloaded data 1 through edge node 1 and target preloaded data 2 through edge node 2.

[0160] Terminal 2 corresponds to target preloaded data 3 and target preloaded data 4. Target preloaded data 3 is configured with preload queue 2, and target preloaded data 4 is configured with preload queue 3. The target storage medium for target preloaded data 3 is edge node 2, and the target storage medium for target preloaded data 4 is cache server 1. During preloading, target preloaded data 3 is preloaded into preload queue 2, and target preloaded data 4 is preloaded into preload queue 3, waiting for the preload time to arrive. When the preload time arrives, target preloaded data 3 in preload queue 2 is stored in edge node 2, and target preloaded data 4 in preload queue 3 is stored in cache server 1. During data access, terminal 2 can access target preloaded data 3 through edge node 2 and target preloaded data 4 through cache server 1.

[0161] The mapping method between the preload queue and the target storage medium can be set according to the actual situation, and this embodiment does not impose a specific limitation on it. For example, the mapping method can be a hash mapping method; or, the mapping method can be a tree mapping method, a linear mapping method, or other mapping methods, but it is not limited to these.

[0162] In some implementations, refer to Figure 8 In step S104 above, after preloading the target preload data using target behavior information to obtain the preloaded target data, the method may further include at least one of the following steps S11-S12:

[0163] S11, based on the media usage rate of each target preloaded data, perform data migration processing on the target storage media of each target preloaded data.

[0164] In this step, after the preloading process is completed, the target storage media for each target preloaded data are monitored in real time to migrate data from overloaded target storage media, thereby freeing up some storage space. Specifically, for each target storage media for the target preloaded data, firstly, the media utilization rate is obtained; then, based on the media utilization rate, it is determined whether the target storage media is overloaded; if so, data migration is performed on the target storage media to free up storage space; otherwise, no action is taken. In this way, this step can improve the resource utilization rate of target storage media with sufficient storage space and alleviate the load on target storage media with insufficient storage space, thereby balancing the storage space among target storage media and ensuring the efficiency of subsequent cloud storage data access.

[0165] The aforementioned media utilization rate refers to the ratio of the used storage space parameter of the target storage medium to the total storage space parameter of the target storage medium, and it can be data obtained in real time.

[0166] The specific implementation method of the above data migration process can be set according to the actual situation, and this implementation method does not impose specific limitations on it.

[0167] For example, in some embodiments, if the target storage medium for each target preloaded data has a media utilization rate greater than a preset utilization rate threshold, it indicates that the target storage medium is overloaded. In this case, the target storage medium with a utilization rate less than the preset utilization rate threshold is first identified as the storage medium to be migrated. Then, a portion of the target preloaded data stored in the target storage medium is randomly selected as the preloaded data to be migrated. After that, the preloaded data to be migrated is migrated to the storage medium to be migrated so that the media utilization rate drops to the utilization rate threshold, and it is ensured that the utilization rate of the storage medium to be migrated after the data migration remains less than the utilization rate threshold. Finally, the target storage medium for the preloaded data to be migrated is updated to the storage medium to be migrated.

[0168] For example, in some embodiments, if the target storage medium for each target preloaded data has a media utilization rate greater than a preset utilization rate threshold, it indicates that the target storage medium is overloaded. In this case, the target storage medium with a utilization rate less than the preset utilization rate threshold is first identified as the storage medium to be migrated. Then, from the target preloaded data stored in the target storage medium, the target preloaded data with a utilization rate lower than the preset utilization rate threshold and a timeliness coefficient lower than the preset timeliness threshold is selected as the preloaded data to be migrated. After that, the preloaded data to be migrated is migrated to the storage medium to be migrated so that the media utilization rate drops to the utilization rate threshold, and ensures that the utilization rate of the storage medium to be migrated after the data migration remains less than the utilization rate threshold. Finally, the target storage medium for the preloaded data to be migrated is updated to the storage medium to be migrated.

[0169] The above-mentioned usage rate threshold can be set according to the actual situation, and this implementation method does not impose a specific limitation on it. For example, the usage rate threshold is 70%, but it is not limited to this.

[0170] S12, according to the priority of each target preloaded data, perform data clearing processing on the target storage medium of each target preloaded data.

[0171] In this step, after the preloading process is completed, the target storage media for each target preloaded data are monitored in real time to delete expired or unnecessary data, thereby optimizing the use of storage resources on the target storage media. Specifically, for each target preloaded data, firstly, the priority of the target preloaded data is obtained; then, the target preloaded data with lower priority is deleted from its corresponding target storage media to free up storage space. In this way, this step can effectively reduce the probability of target storage media overload, improve the resource utilization of target storage media, thereby balancing the storage space among target storage media and ensuring the efficiency of subsequent cloud storage data access.

[0172] The priority of the target preloaded data is positively correlated with the terminal need and timeliness coefficient of the target preloaded data. The terminal need refers to the degree to which the target terminal needs the target preloaded data. That is, the higher the priority of the target preloaded data, the more the target terminal needs the target preloaded data and the higher the timeliness coefficient of the target preloaded data. This means that the target preloaded data with low priority is expired or not needed.

[0173] The method for obtaining the priority of the aforementioned target preloaded data can be preset according to the actual situation, and this embodiment does not impose specific limitations on it.

[0174] For example, in some embodiments, for each target preloaded data, the priority of the target preloaded data is obtained based on the importance coefficient and usage frequency of the target preloaded data, combined with machine learning methods.

[0175] For example, in some embodiments, for each target preloaded data, the priority corresponding to the importance coefficient and usage frequency of the target preloaded data is found from the fourth mapping data as the priority of the target preloaded data. The fourth mapping data may include multiple preset comprehensive parameters and the priority corresponding to each preset comprehensive parameter. The preset comprehensive parameters may include preset importance coefficients and preset usage frequencies.

[0176] The fourth mapping data mentioned above can be chart data or tabular data, but is not limited to these.

[0177] The implementation method of the above data cleaning process can be preset according to the actual situation, and this embodiment does not impose specific limitations on it.

[0178] For example, in some embodiments, for each target preloaded data, if the priority of the target preloaded data is lower than a preset priority threshold, the target preloaded data is regarded as preloaded data to be deleted; all preloaded data to be deleted is deleted from its corresponding target storage medium.

[0179] For example, in some embodiments, the target preloaded data is sorted according to the priority of each target preloaded data to obtain sorted target preloaded data, wherein the higher the priority of the target preloaded data, the earlier the target preloaded data is sorted. Then, from all sorted target preloaded data, the last N target preloaded data are selected as preloaded data to be deleted, and all preloaded data to be deleted are deleted from their corresponding target storage media.

[0180] Secondly, refer to Figure 9 This application provides a cloud storage data access device, which may include:

[0181] The acquisition module 201 is used to acquire historical behavior information of the target terminal; wherein, the historical behavior information is parameters associated with the cloud storage data access behavior of the target terminal within a historical time period;

[0182] The first processing module 202 is used to obtain target behavior information of the target terminal and several target access data based on historical behavior information; the target behavior information consists of parameters associated with the cloud storage data access behavior of the target terminal in a future time period.

[0183] The second processing module 203 is used to determine the preloading parameters and multiple target preloading data based on the target behavior information and several target access data;

[0184] The third processing module 204 is used to preload the preloaded data of each target according to the preloaded parameters to obtain the preloaded data of each target.

[0185] The fourth processing module 205 is used to process the preloaded target data in response to the access command of the target terminal.

[0186] The content of the above method embodiments is applicable to the device embodiments. The specific functions implemented by the device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0187] To facilitate understanding of the cloud storage data access method and apparatus described in the embodiments of this application, examples of actual application scenarios of the cloud storage data access method and apparatus described in this application are provided for illustration. (Refer to...) Figure 10 For a single terminal, the specific process for accessing cloud storage data in this application embodiment is shown in steps S301-S306 below.

[0188] S301, Historical Data Collection and Analysis: Obtain parameters related to the cloud storage data access behavior of the target terminal within a historical time period through the application programming interface, i.e., historical behavior information. This includes behavioral characteristic data, time characteristic data, user characteristic data, and context characteristic data of the target terminal within the historical time period.

[0189] S302, Data Processing and Feature Engineering: First, the historical behavior information is preprocessed to obtain preprocessed historical behavior information; then, features are extracted from the preprocessed historical behavior information to obtain historical behavior features.

[0190] S303: Intelligent Prediction: Predicts and processes historical behavioral characteristics to capture the access behavior and access needs of the target terminal in the future time period, thereby obtaining target behavior information and several target access data. The target behavior information includes the access frequency of each target access data, the access pattern of each target access data, the data type of each target access data, and the peak data access time of the target terminal.

[0191] S304, Analysis of Preloading Parameters: Based on the prediction results, a dynamic preloading strategy is designed, including the selection of target preloaded data, the determination of preload time, the determination of storage medium for target preloaded data, and the determination of storage format for target preloaded data. In this way, multiple target preloaded data, the target storage medium for each target preloaded data, the target storage format for each target preloaded data, and the preload time for the target terminal can be obtained.

[0192] S305, Implementation of Data Preloading Process: First, the target preloaded data is transformed according to its target storage format to match its target storage format. After the transformation, several preloading queues are obtained, each used to preload the corresponding target preloaded data. The preloading queue corresponding to the target preloaded data is determined from these queues, and the target preloaded data is then loaded into the appropriate queue. Next, the preloading time for the target terminal is awaited. When the preloading time arrives, the target preloaded data is mapped from its corresponding preloading queue to its target storage medium, ensuring it is stored in the target storage format, thus obtaining the preloaded target preloaded data. By traversing all target preloaded data, multiple preloaded target preloaded data sets can be obtained. Furthermore, to optimize the storage space and resource utilization of each target storage medium, optimization processes such as data migration and data cleaning can be performed after data preloading.

[0193] S306, Access to cloud storage data: Receives access instructions from the target terminal, responds to the access instructions from the target terminal, retrieves each pre-loaded target pre-loaded data from the target storage medium, and sends each pre-loaded target pre-loaded data to the target terminal so that the target terminal can perform operations such as decryption and decompression on each pre-loaded target pre-loaded data and render and display it.

[0194] In some application scenarios, cloud storage service users are enterprise users who need to access a large number of financial reports and project reports from the previous month at the beginning of each month. Historical data shows that the peak access period always occurs in the first week of each month. In this application scenario, intelligent prediction technology is used to predict and analyze the historical behavior information of this enterprise user, and the preloading strategy is designed based on the prediction results. In this enterprise user's preloading strategy, the first target preloaded data is the financial report, the target storage medium for this target preloaded data is edge node A, and the target storage format for this target preloaded data is encrypted text; the second target preloaded data is the project report, the target storage medium for this target preloaded data is cache server B, and the target storage format for this target preloaded data is compressed text; the preloading time is the week before the peak data access period, which is predicted to be the first week of the following month. Based on this, in the last week of this month, the financial report is preloaded and stored in encrypted text format on edge node A through a preloading queue, and the project report is preloaded and stored in compressed text format on cache server B through a preloading queue. When the first week of next month arrives, enterprise users can access financial reports preloaded on edge node A and project reports preloaded on cache server B. The average access latency for financial reports and project reports can be reduced from 2.5 seconds to 0.5 seconds, significantly improving the access speed of cloud storage data and the user experience.

[0195] In other application scenarios, cloud storage service users are entertainment platform users who typically watch popular movies and TV series on weekend and holiday evenings. Historical data shows that these users prefer to watch specific types of programs in the evening. In this application scenario, intelligent prediction technology is used to predict and analyze the historical behavior information of these entertainment platform users, and the prediction results are used to design a preloading strategy for them. In this preloading strategy, the target preloaded data includes program 1, program 2, ..., program N. The target storage medium for all target preloaded data is edge node C, and the target storage format for all target preloaded data is ordinary video. The preloading time is the first two hours of the peak data access period, which is predicted to be from 5 PM to 8 PM today. Based on this, at 3 PM today, program 1, program 2, ..., program N are preloaded and stored in ordinary video format on edge node C through the preloading queue. At 5 PM today, entertainment platform users accessed and watched various programs pre-loaded on edge node C. The video loading speed of various programs can be improved by 50%, and the video stuttering issues reported by users are reduced by 70%, thus improving the video viewing experience.

[0196] In some application scenarios, the cloud storage service users are online education platform users who typically provide students with access to a large amount of course materials during peak class times. Historical data shows that students begin accessing course materials extensively 10 minutes before class. In this application scenario, intelligent prediction technology is used to predict and analyze the historical behavior information of the online education platform users, and the prediction results are used to design a preloading strategy for the online education platform users. In this preloading strategy, the target preloaded data are course material 1, course material 2, course material 3, ..., course material N. The target storage medium for all target preloaded data is a high-speed cache server D, and the target storage format for all target preloaded data is plain text. The preloading time is 30 minutes during the peak data access period, which is predicted to be from 3 PM to 5 PM today. Based on this, at 2:30 PM today, course material 1, course material 2, course material 3, ..., course material N are preloaded and stored on the high-speed cache server D through the preloading queue. At 3 PM today, users of the online education platform accessed various course materials pre-loaded on the cache server D. The average loading time for various course materials was reduced from 5 seconds to 1 second, and the problem of slow loading of course materials reported by students was basically solved, improving the learning experience.

[0197] In summary, the embodiments of this application can effectively improve the data access efficiency of cloud storage services and optimize user experience. They not only alleviate data latency issues but also improve the resource utilization of cloud storage services, demonstrating their enormous potential and value in cloud storage service optimization.

[0198] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.

[0199] The above is a detailed description of the preferred embodiments of this application, but this application is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this application, and these equivalent modifications or substitutions are all included within the scope defined by the claims of this application.

Claims

1. A cloud storage data access method, characterized in that, Includes the following steps: Obtain historical behavior information of the target terminal; wherein, the historical behavior information is a parameter associated with the cloud storage data access behavior of the target terminal within a historical time period; Based on the historical behavior information, target behavior information of the target terminal and several target access data are obtained; wherein, the target behavior information are parameters associated with the cloud storage data access behavior of the target terminal in a future time period; Based on the target behavior information and several target access data, multiple target preload data are determined; The target behavior information is used to preload the target preload data to obtain the preloaded target preload data. In response to the access command of the target terminal, access processing is performed on each preloaded target preloaded data; The step of obtaining the target behavior information and several target access data of the target terminal based on the historical behavior information includes: The historical behavior information is preprocessed to obtain the preprocessed historical behavior information; Feature extraction is performed on the preprocessed historical behavior information to obtain historical behavior features; The historical behavioral characteristics are predicted to obtain the target behavior information and several target access data. The target behavior information of the target terminal includes the access frequency of each of the target access data; the step of determining multiple target preloaded data based on the target behavior information and several of the target access data includes: Based on the access frequency of each target access data, and combined with the timeliness coefficient and user preference coefficient of each target access data, multiple target preloaded data are determined from a number of target access data.

2. The cloud storage data access method according to claim 1, characterized in that, The step of preloading each target preloaded data using the target behavior information to obtain each preloaded target preloaded data includes: Based on the target behavior information and preset comprehensive parameters, preloading parameters are obtained; wherein, the preloading parameters include the target storage medium of each target preloaded data, the target storage format of each target preloaded data, and the preloading time of the target terminal; The target preload data is preloaded using the preload parameters to obtain the preloaded target preload data.

3. The cloud storage data access method according to claim 2, characterized in that, The comprehensive parameters include the importance coefficient of each target preloaded data, the usage frequency of each target preloaded data, the terminal geographical location of the target terminal, the network characteristic data of the target terminal, the load rate of each preset storage medium, and the terminal geographical location of each preset storage medium. The process of obtaining pre-loaded parameters based on the target behavior information and preset comprehensive parameters includes: Based on the importance coefficient and usage frequency of each target preloaded data, combined with the terminal geographical location of the target terminal, the network characteristic data of the target terminal, the load rate of each preset storage medium and the terminal geographical location of each preset storage medium, the target storage medium for each target preloaded data is determined from among the multiple preset storage media.

4. The cloud storage data access method according to claim 2, characterized in that, The target behavior information of the target terminal includes the data type, access mode, and access frequency of each target preloaded data; the comprehensive parameters include the sensitivity coefficient and media load rate of each target preloaded data. The process of obtaining pre-loaded parameters based on the target behavior information and preset comprehensive parameters includes: Based on the data type, access mode, and access frequency of each target preloaded data, and in conjunction with the sensitivity coefficient and media load rate of each target preloaded data, the target storage format of each target preloaded data is determined.

5. The cloud storage data access method according to claim 2, characterized in that, The target terminal's target behavior information includes the target terminal's peak data access periods; the comprehensive parameters include the target terminal's network characteristic data; The process of obtaining pre-loaded parameters based on the target behavior information and preset comprehensive parameters includes: Based on the peak data access times of the target terminal and combined with the network characteristic data of the target terminal, the preloading time of the target terminal is obtained.

6. The cloud storage data access method according to claim 2, characterized in that, The step of preloading each of the target preloaded data using the preload parameters to obtain each preloaded target preloaded data includes: Transform the storage format of the target preloaded data into the target storage format of the target preloaded data; Load the target preloaded data into the preload queue corresponding to the target preloaded data; In the case of the preloading time arriving at the target terminal, the target preloaded data is mapped from the preload queue corresponding to the target preloaded data to the target storage medium of the target preloaded data to obtain the preloaded target preloaded data; The preloaded target preloaded data is stored in the target storage medium of the target preloaded data in the target storage format of the target preloaded data.

7. The cloud storage data access method according to claim 1, characterized in that, After preloading each target preloaded data using the target behavior information to obtain each preloaded target preloaded data, the method further includes at least one of the following: Based on the media usage rate of each target preloaded data, data migration processing is performed on the target storage media of each target preloaded data; Based on the priority of each target preloaded data, the target storage medium of each target preloaded data is subjected to data clearing processing.

8. A data access device, characterized in that, include: The acquisition module is used to acquire historical behavior information of the target terminal; wherein, the historical behavior information is a parameter associated with the cloud storage data access behavior of the target terminal within a historical time period; The first processing module is used to obtain target behavior information of the target terminal and several target access data based on the historical behavior information; wherein, the target behavior information is a parameter associated with the cloud storage data access behavior of the target terminal in a future time period; The second processing module is used to determine multiple target preload data based on the target behavior information and several target access data. The third processing module is used to preload each target preload data using the target behavior information to obtain each preloaded target preload data. The fourth processing module is used to respond to the access command of the target terminal and perform access processing on each preloaded target preloaded data; The step of obtaining the target behavior information and several target access data of the target terminal based on the historical behavior information includes: The historical behavior information is preprocessed to obtain the preprocessed historical behavior information; Feature extraction is performed on the preprocessed historical behavior information to obtain historical behavior features; The historical behavioral characteristics are predicted to obtain the target behavior information and several target access data. The target behavior information of the target terminal includes the access frequency of each of the target access data; the step of determining multiple target preloaded data based on the target behavior information and several of the target access data includes: Based on the access frequency of each target access data, and combined with the timeliness coefficient and user preference coefficient of each target access data, multiple target preloaded data are determined from a number of target access data.

Images