Object automatic hierarchical storage method and device, equipment and storage medium

Abstract

The invention relates to the technical field of data storage, can be applied to the medical field and the financial field, and discloses an object automatic hierarchical storage method, device and equipment and a storage medium, the method comprises the following steps: obtaining object historical access data, and extracting an access popularity value according to the object historical access data; judging whether a storage level migration decision is triggered or not according to the access popularity value; when it is judged that the storage level migration decision is triggered, the time sequence prediction model is used for conducting access demand prediction; and based on the result of the access demand prediction, executing an incremental migration operation through an intelligent scheduling algorithm, and carrying out hierarchical migration on the object according to a preset hierarchical strategy template. According to the method, whether the storage level migration decision is triggered or not is judged through the access heat value, the access demand prediction is performed, hierarchical migration is performed in combination with the intelligent scheduling algorithm and the preset hierarchical strategy template, and the problems of low storage efficiency, difficulty in cost control, poor user experience and the like in the prior art can be effectively solved.

Description

Technical Field

[0001] This invention relates to the field of data storage technology, and in particular to an automatic hierarchical storage method, apparatus, device, and storage medium for objects. Background Technology

[0002] In today's digital age, data is growing explosively, making object storage increasingly important. To meet users' data storage management needs, many cloud service providers have launched automated storage tier management functions. However, these existing technologies have revealed many key flaws and shortcomings in practical applications, severely hindering improvements in storage efficiency, cost control, and user experience. Most existing automated hierarchical storage solutions rely on simple time-based strategies. For example, in the medical field, hospital image data storage typically involves periodically migrating image data that hasn't been accessed for a period of time from standard storage to infrequently accessed storage. A large tertiary hospital, for instance, moves image data that hasn't been accessed for more than six months to infrequently accessed storage every three months. However, this static strategy has significant drawbacks. In real-world medical scenarios, patients' conditions are complex and constantly changing, and doctors may need to access images from years ago to compare disease progression. If the images have already been migrated to infrequently accessed storage, the retrieval speed will be significantly reduced, severely impacting diagnostic efficiency. Similarly, in the financial sector, some financial institutions use similar time-based storage strategies for transaction data. For example, a securities trading platform migrates historical transaction data of customers who haven't traded for more than a year to infrequently accessed storage. However, during periods of high market volatility, investors may suddenly need to access transaction records from years ago to formulate investment strategies. Access delays caused by data migration may cause investors to miss trading opportunities, harming the user experience. This static, time-driven strategy lacks intelligent awareness of actual access patterns, making it difficult to adapt to complex and ever-changing business scenarios. This results in unreasonable resource allocation and an inability to effectively balance storage costs and access performance. Current access pattern analysis mechanisms in most systems are relatively simplistic. They typically focus only on the frequency of object access, neglecting crucial information such as the temporal distribution of access, the continuity of access patterns, and the changing trends of data hotspots. Furthermore, existing technologies generally lack dynamic adaptation mechanisms. Once hierarchical rules are set, even significant changes in the business scenario make it difficult for the system to adjust in a timely manner. Traditional solutions also lack the ability to fine-grainedly distinguish between different access patterns, failing to accurately identify and respond to complex user access behaviors. For example, during e-commerce promotional events, consumer access to product detail pages exhibits a sudden surge in access, concentrated within a short period after the event begins. For some subscription services, user access to service content is periodic, such as checking subscription reports at fixed monthly times. Existing storage systems struggle to accurately differentiate and optimize storage management for these complex access patterns. Additionally, many existing systems lack effective predictive mechanisms when performing storage level conversions, failing to anticipate future data access demands. This can lead to frequent data migration operations, increasing system overhead and impacting service stability. In summary, existing object storage technologies have significant shortcomings in storage efficiency, cost control, and user experience, all of which are problems that those skilled in the art need to address. Summary of the Invention

[0003] This invention provides an automatic hierarchical storage method, apparatus, computer device, and storage medium for objects, aiming to solve problems such as low storage efficiency, difficulty in cost control, and poor user experience in the prior art.

[0004] In a first aspect, embodiments of the present invention provide an automatic hierarchical storage method for objects, comprising: Obtain historical access data for an object, and extract access popularity values ​​based on the historical access data. Determine whether to trigger a storage-level migration decision based on the access popularity value; When a storage-level migration decision is triggered, a time-series forecasting model is used to predict access demand. Based on the results of access demand prediction, incremental migration operations are performed through intelligent scheduling algorithms, and objects are migrated in a hierarchical manner according to a preset hierarchical strategy template.

[0005] Secondly, embodiments of the present invention provide an automatic hierarchical storage device for objects, comprising: The data acquisition unit is used to acquire historical access data of an object and extract access popularity values ​​based on the historical access data of the object. A triggering judgment unit is used to determine whether to trigger a storage level migration decision based on the access popularity value. The demand forecasting unit is used to forecast access demand using a time series forecasting model when a storage-level migration decision is triggered. The hierarchical migration unit is used to perform incremental migration operations based on the results of access demand prediction through an intelligent scheduling algorithm, and to migrate objects hierarchically according to a preset hierarchical strategy template.

[0006] Thirdly, embodiments of the present invention provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the automatic hierarchical storage method for objects as described in the first aspect.

[0007] Fourthly, embodiments of the present invention provide a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, it implements the automatic hierarchical storage method for objects as described in the first aspect.

[0008] This invention provides an automatic hierarchical storage method, apparatus, computer device, and storage medium for objects. The method includes: acquiring historical access data of objects and extracting access popularity values ​​based on the historical access data; determining whether a storage level migration decision is triggered based on the access popularity values; when a storage level migration decision is triggered, predicting access demand using a time series forecasting model; and performing incremental migration operations using an intelligent scheduling algorithm based on the predicted access demand results, while simultaneously performing hierarchical migration of objects according to a preset hierarchical strategy template. This invention first acquires historical access data of objects and extracts access popularity values, then determines whether a storage level migration decision is triggered based on these values. If triggered, it uses a time series forecasting model to predict access demand, and performs incremental migration operations using an intelligent scheduling algorithm based on the prediction results, while simultaneously performing hierarchical migration according to a preset hierarchical strategy template. Based on the aforementioned automatic hierarchical storage method for objects, embodiments of the present invention can ensure that hot data is always stored in high-performance storage categories, thereby significantly improving object access response speed and overall system throughput. At the same time, it reduces unnecessary storage migration operations and lowers system resource consumption. Furthermore, by accurately identifying data access patterns and dynamically adjusting storage levels, compared to traditional time-driven strategies, embodiments of the present invention can promptly transfer cold data to lower-cost storage categories while preventing hot data from affecting access performance due to premature migration, thus achieving optimal configuration of storage costs. This effectively solves the problems of low storage efficiency, difficulty in cost control, and poor user experience in existing technologies. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 A flowchart illustrating an automatic hierarchical storage method for objects provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a sub-process of step S101 in an automatic hierarchical storage method for objects provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of a sub-process of step S102 in an automatic hierarchical storage method for objects provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of a sub-process of step S103 in an automatic hierarchical storage method for objects provided in an embodiment of the present invention; Figure 5 A schematic block diagram of an automatic hierarchical storage device for objects provided in an embodiment of the present invention; Figure 6 This is a first sub-schematic block diagram of an automatically hierarchical storage device for objects provided in an embodiment of the present invention; Figure 7 This is a second sub-schematic block diagram of an automatically hierarchical storage device for objects provided in an embodiment of the present invention; Figure 8 This is a third schematic block diagram of an automatic hierarchical storage device for objects provided in an embodiment of the present invention; Figure 9 This is a schematic block diagram of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0011] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0012] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0013] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0014] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0015] The gradient-guided image classification method provided in this invention can be applied in client-server interaction environments, where the client communicates with the server via a network. The server acquires historical access data for objects and extracts access popularity values ​​based on this data. It then determines whether a storage-level migration decision is triggered based on the access popularity values. When a storage-level migration decision is triggered, it uses a time-series prediction model to predict access demand. Based on the predicted access demand, it executes incremental migration operations using an intelligent scheduling algorithm and performs hierarchical migration of objects according to a preset hierarchical strategy template. The client can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server can be implemented using a standalone server or a server cluster consisting of multiple servers. The invention will be described in detail below through specific embodiments.

[0016] Please see below. Figure 1 This invention provides an automatic hierarchical storage method for objects, specifically including steps S101 to S105.

[0017] Step S101: Obtain historical access data of the object, and extract the access popularity value based on the historical access data of the object; Step S102: Determine whether to trigger a storage level migration decision based on the access popularity value; Step S103: When it is determined that a storage level migration decision is triggered, access demand is predicted using a time series forecasting model. Step S104: Based on the results of access demand prediction, perform incremental migration operation through intelligent scheduling algorithm, and migrate objects in a hierarchical manner according to the preset hierarchical strategy template.

[0018] This embodiment first acquires historical access data for the object and extracts the access popularity value. Then, based on the access popularity value, it determines whether a storage-level migration decision should be triggered. If triggered, a time series prediction model is used to predict access demand, and based on the prediction results, an incremental migration operation is executed through an intelligent scheduling algorithm. At the same time, a tiered migration is performed according to a preset tiered strategy template.

[0019] The automatic hierarchical storage method for objects provided in this embodiment can effectively solve the problems of low storage efficiency, difficulty in cost control, and poor user experience in existing technologies. Specifically, in terms of storage cost control, this embodiment accurately identifies data access patterns and dynamically adjusts storage levels. Compared with traditional time-driven strategies, this embodiment can promptly transfer cold data to lower-cost storage categories while preventing hot data from affecting access performance due to premature migration, thus achieving optimal allocation of storage costs. In terms of system performance improvement, this embodiment relies on a hierarchical strategy based on data popularity and a predictive management mechanism to ensure that hot data is always stored in high-performance storage categories, thereby significantly improving object access response speed and overall system throughput, while reducing unnecessary storage migration operations, reducing system resource consumption, and further improving operating efficiency. In terms of system stability assurance, this embodiment uses an adaptive learning mechanism and predictive management to ensure storage... The strategy can dynamically adapt to business changes, avoiding performance degradation caused by rigid strategies. Fine-grained monitoring and feedback control further enhance the system's self-adjustment capabilities, ensuring long-term stable and reliable operation. Regarding scalability and compatibility, this embodiment can be seamlessly integrated into existing object storage systems without requiring large-scale modifications to the underlying architecture, significantly reducing user deployment costs and technical risks. The flexible strategy configuration function can meet the storage management needs of enterprises of different sizes and various application scenarios. In terms of user experience optimization, the intelligent storage management provided in this embodiment enables the system to automatically identify and respond to various access patterns. Users can obtain more stable, efficient, and economical storage services without manual intervention in adjusting storage strategies, significantly improving user satisfaction and experience.

[0020] For example, in the medical field, the automatic hierarchical storage method for objects provided in this embodiment can automatically classify and store medical objects such as patients' electronic medical records and imaging data according to their access frequency. For instance, for patients with acute illnesses or those who have recently visited the hospital frequently, their relevant medical data has a high access frequency, so this data can be automatically stored in a high-performance storage tier, ensuring that doctors can quickly retrieve it when querying, thus gaining valuable time for timely treatment of patients. For data of patients with chronic illnesses or those who have not had follow-up visits for a long time, the access frequency is relatively low, so it can be migrated to a lower-cost storage tier, reducing storage costs while ensuring data accessibility. Simultaneously, by using time series prediction models to predict patients' follow-up visit times and examination items, potentially accessed data can be migrated to appropriate storage tiers in advance, further improving data access efficiency and providing strong support for the continuity and efficiency of medical services. In the financial industry, taking a securities trading system as an example, the automatic hierarchical storage method for objects provided in this embodiment can also play a significant role. Securities trading generates a large amount of trading data, including order information, transaction records, and market data. The access frequency of trading data varies greatly for different stocks and different time periods. The method described in this embodiment allows for real-time acquisition of historical access data for these transaction data, extracting access popularity values. For data generated during peak trading hours for popular stocks, due to their extremely high access popularity, it can be immediately stored in high-speed, low-latency storage devices. This ensures the trading system can process this data quickly, avoiding transaction failures or delays caused by data access latency and guaranteeing smooth financial transactions. Conversely, for less popular stocks or historical trading data with lower access popularity, it can be stored in low-cost, high-capacity storage devices. Furthermore, time-series forecasting models can be used to predict the trading activity of different stocks over a future period, allowing for advance adjustments to data storage levels to cope with potential trading peaks. This improves the stability and reliability of the financial trading system, saves financial institutions significant storage costs, and enhances customer service quality. In addition, the automatic hierarchical storage method provided in this embodiment is also suitable for the e-commerce sector, given the massive amounts of product information and user access data. E-commerce platforms offer a wide variety of products, each with varying popularity and sales cycles, resulting in vastly different access popularity for their related data. For example, during promotional events, product detail pages and review data for popular items will be frequently accessed by a large number of users. By acquiring historical access records of this data and extracting high access frequency values, a storage-level migration decision can be triggered. This allows the data of these popular products to be quickly migrated to a high-performance storage tier, ensuring fast page loading for users browsing products and improving the shopping experience. Conversely, data on products that have been slow-moving for a long time or are highly seasonal, with lower access frequency, can be stored in lower-cost storage.At the same time, time series forecasting models are used to predict product sales trends and user access patterns, and the storage level of product data is adjusted in advance to ensure that relevant data can quickly respond to a large number of user access requests when the peak sales season arrives, avoiding user churn due to data access problems, and promoting the business development and sales growth of the e-commerce platform.

[0021] In particular, the automatic hierarchical storage method for objects provided in this embodiment is especially suitable for the S3 (Simple Storage Service) object storage system. As a widely used cloud storage solution, this system also faces challenges in storage efficiency, cost control, and user experience when storing massive amounts of data. This embodiment, through its automatic hierarchical storage mechanism integrated into the S3 object storage system, can intelligently adjust the storage tier based on the actual access frequency of the data. This not only optimizes the allocation of storage resources but also significantly improves the efficiency and response speed of data access.

[0022] In one embodiment, such as Figure 2 As shown, step S101 includes steps S201 to S203.

[0023] Step S201: Perform multi-dimensional feature extraction processing on the historical access data of the object to obtain the corresponding multi-dimensional features; wherein, the multi-dimensional features include access frequency, access time, access continuity and popularity score. Step S202: Construct a multi-dimensional access behavior analysis model based on the multi-dimensional features; Step S203: Use sliding window and exponential weighted average techniques to assign weights to the multi-dimensional features in the multi-dimensional access behavior analysis model, and use the weighted popularity score as the access popularity value.

[0024] This embodiment extracts and analyzes multi-dimensional features from historical access data of objects to more comprehensively and accurately reflect the access popularity of data. Specifically, access frequency reflects the number of times data is accessed within a certain time range and is one of the important indicators for measuring data popularity; access time reflects the specific time period during which data is accessed, and the access popularity of data may vary significantly at different times. For example, during specific time periods on weekdays, the access popularity of certain business-related data may be higher; access continuity describes the persistence of data access, and data with more consecutive accesses often has higher popularity; popularity score can comprehensively quantify the data popularity to distinguish different objects into different levels such as "high popularity," "moderate popularity," and "low popularity." In addition, in practical applications, more features can be extracted, such as pattern stability and access source distribution, to further improve the multi-dimensional feature system.

[0025] By constructing a multi-dimensional access behavior analysis model, these characteristics can be integrated and analyzed to more accurately grasp data access patterns. For example, whether access exhibits periodicity (e.g., daily scheduled access), suddenness (e.g., instantaneous peaks caused by marketing campaigns), or persistence (e.g., long-term active services). Simultaneously, employing sliding window and exponentially weighted averaging techniques to assign weights to multi-dimensional features allows for dynamic adjustment of the weights of each feature based on real-time data access patterns. This makes the calculated access popularity values ​​more closely reflect reality and effectively eliminates noise interference, avoiding misjudgments caused by abnormal access at a single point in time, thus providing a reliable basis for subsequent storage-level migration decisions.

[0026] In one embodiment, such as Figure 3 As shown, step S102 includes steps S301 to S303.

[0027] Step S301: Compare the access popularity value with a preset popularity threshold; Step S302: If the access popularity value is greater than or equal to the preset popularity threshold, then a storage level migration decision is triggered. Step S303: If the access popularity value is less than the preset popularity threshold, it is determined that the storage level migration decision will not be triggered.

[0028] This embodiment compares access frequency values ​​with a preset frequency threshold to easily and effectively determine whether a storage tier migration decision needs to be triggered. The preset frequency threshold can be flexibly set based on actual business needs, storage system performance, and cost. For example, during peak business periods or in scenarios with high performance requirements, the preset frequency threshold can be set relatively high to ensure that only data with truly high access frequency will trigger migration, thereby guaranteeing that data in the high-performance storage tier is genuinely frequently accessed and improving overall system performance. Conversely, in scenarios with relatively stable business or higher cost sensitivity, the preset frequency threshold can be appropriately lowered to include more data with a certain level of access frequency in the migration scope, further optimizing storage costs.

[0029] When the access popularity value is determined to be greater than or equal to the preset popularity threshold, a storage level migration decision is automatically triggered, such as migrating from Standard Storage to Intelligent Tiering or Infrequent Access Storage, so as to store the data in a storage level that better matches its current access pattern and achieve efficient utilization of storage resources. When the access popularity value is less than the preset popularity threshold, the current storage level of the data remains unchanged to avoid performance overhead and increased costs caused by unnecessary migration operations.

[0030] In one embodiment, such as Figure 4 As shown, step S103 includes steps S401 to S403.

[0031] Step S401: Extract the corresponding time series features based on the object's historical access data; Step S402: Construct a time series prediction model based on the time series features; Step S403: Use the time series prediction model to predict the access demand within a specified time period to obtain the access demand prediction result.

[0032] This embodiment first extracts time-series features from historical access data of an object. Time-series features can capture the patterns and trends of data access over time. For example, access to certain data may exhibit obvious periodicity, such as higher access volume at specific times of the day; or it may be seasonal, such as a significant increase in access volume for certain products during specific seasons. By extracting these time-series features, we can gain a more accurate understanding of data access patterns. Then, a time-series prediction model is built based on the extracted time-series features. This model can employ different algorithmic architectures, such as the Autoregressive Integral Moving Average (ARIMA) model, Long Short-Term Memory (LSTM) network, or Prophet (a time-series prediction algorithm), Transformer-based models, etc. These models can predict access demand for a specified period in the future based on the time-series features in historical data. For example, by analyzing data access patterns over the past few months, we can predict the daily access volume for the next week to identify access peaks or troughs in advance, proactively adjust storage tiers, and avoid frequent migrations and performance fluctuations caused by sudden access surges.

[0033] Subsequently, the established time-series forecasting model is used to predict access demand within a specified time period, yielding the access demand forecast results. These forecasts provide crucial information for subsequent storage tier migration decisions. For example, if a significant increase in access to certain data is predicted in the future, this data can be migrated to a high-performance storage tier in advance to ensure rapid response during peak access periods and avoid performance issues caused by data access latency. Furthermore, accurate access demand forecasting can prevent excessive data migration to high-performance storage tiers, thereby reducing storage costs.

[0034] In one embodiment, step S104 includes: Obtain system load status information, and determine the execution time of incremental migration operation based on the load status information using an intelligent scheduling algorithm; Based on the execution time, incremental migration operations are performed using idle system resources.

[0035] In this embodiment, by acquiring system load status information during migration, the system resource utilization can be monitored in real time, avoiding incremental migration operations during periods of high system load and thus minimizing the impact on normal business processing. The intelligent scheduling algorithm can accurately determine the execution time of incremental migration operations based on system load status information, such as selecting nighttime periods with lower system load or relatively idle business periods for migration. Based on the determined execution time, incremental migration operations are performed using idle system resources. This fully utilizes idle system resources, improving resource utilization without placing additional burden on the system. This intelligent scheduling method based on system load status ensures the smooth execution of incremental migration operations while guaranteeing stable system operation, effectively avoiding system performance degradation caused by migration operations.

[0036] In one embodiment, step S104 further includes: In response to the user's migration command or through the intelligent scheduling algorithm, a hierarchical strategy is selected from the preset hierarchical strategy template; wherein, the hierarchical strategy includes a high-activity-fast response strategy, a low-frequency-low-cost strategy, and a periodic-on-demand warm-up strategy.

[0037] The objects are migrated in a hierarchical manner according to the described hierarchical strategy.

[0038] This embodiment provides different hierarchical strategy options based on a preset hierarchical strategy template to adapt to storage management needs under different business scenarios and data characteristics. For example, it offers various strategies such as the high-activity-fast response strategy, the low-frequency-low-cost strategy, and the periodic-on-demand warm-up strategy. The high-activity-fast response strategy is primarily for data objects with extremely high access frequency and stringent response time requirements. For instance, in a real-time transaction system, order processing data needs to be accessed and processed at extremely high speeds. This strategy ensures that this data is always stored in the optimal storage tier, such as a high-speed solid-state drive array, thereby guaranteeing the efficient and stable operation of the transaction system. The low-frequency-low-cost strategy is suitable for data with low access frequency and high cost sensitivity. For example, some historical archived data may not be accessed for a long time but needs to be retained for later retrieval. This strategy allows these data to be migrated to low-cost, high-capacity storage devices, such as tape libraries or low-speed hard drives, significantly reducing storage costs while maintaining data retrievability. The periodic-on-demand warm-up strategy is for data with obvious periodic access characteristics. Taking seasonal product data from e-commerce platforms as an example, the access volume of these products increases significantly before the arrival of a specific season. Using this strategy, product data that is about to enter its peak access period can be identified in advance based on time series prediction models, and then preheated and migrated to a high-performance storage tier to ensure a rapid response when the peak access period arrives, avoiding user churn due to data loading delays.

[0039] Once a user's migration command is received or a hierarchical strategy is selected from a preset hierarchical strategy template via an intelligent scheduling algorithm, the objects can be migrated hierarchically according to the selected strategy. During the migration process, the migration path and resource allocation can be rationally planned based on factors such as the current storage location of the data, the target storage level, and the size of the data, ensuring the efficiency and stability of the migration process.

[0040] Simultaneously, a full-link monitoring system can be established to collect key indicators such as access latency, throughput, migration success rate, and resource consumption of objects at each level in real time. This allows for real-time monitoring and recording of the migration process, enabling timely handling and recovery in case of anomalies, ensuring data integrity and availability. For example, if the migration failure rate exceeds a threshold or latency increases by more than 30%, an alarm is immediately triggered and a compensation mechanism is initiated, such as pausing migration, rolling back strategies, or switching to a backup solution, to ensure service stability. This enables closed-loop management of the entire process from data awareness, intelligent decision-making, automated execution to dynamic feedback, significantly improving the intelligence level and operational efficiency of the storage system.

[0041] In one embodiment, the automatic hierarchical storage method for objects further includes: Utilize online machine learning mechanisms to dynamically evaluate objects after graded transfer; The heat threshold is adaptively optimized based on the results of dynamic evaluation.

[0042] This embodiment introduces an online machine learning mechanism to continuously and dynamically evaluate objects after hierarchical migration. The online machine learning mechanism can update model parameters in real time as data is continuously generated and changed, without requiring the collection of large amounts of data for full training. This allows the system to promptly capture new changes in object access patterns. For example, some data that was previously not frequently accessed may suddenly become popular due to changes in the market environment or business promotions. The online machine learning mechanism can quickly detect this change and accurately evaluate it.

[0043] Based on the results of dynamic evaluation, the preset heat threshold can be adaptively optimized. The heat threshold is a key indicator for determining whether to trigger a storage tier migration decision. If the heat threshold is set unreasonably, it may lead to frequent invalid migrations or missed important migration opportunities. Through adaptive optimization, the heat threshold can be automatically adjusted according to actual data access patterns and business needs. For example, if the system finds that after migrating according to the current heat threshold, some data with low access frequency still exists in the high-performance storage tier, while some data with rapidly increasing access frequency in the low-performance storage tier has not been migrated in time, the heat threshold can be appropriately lowered. This allows more data with a certain level of access frequency to be migrated to the appropriate storage tier, improving the utilization efficiency of storage resources. Conversely, if data access is frequent in the high-performance storage tier and system resources are strained, the heat threshold can be appropriately increased to reduce unnecessary migration operations and ensure stable system operation. This adaptive optimization mechanism enables the storage system to automatically adjust its strategy according to actual conditions, forming a "perception-decision-execution-feedback" closed loop. This gives the system continuous learning and self-evolution capabilities, adapting to constantly changing business models.

[0044] Figure 5 This is a schematic block diagram of an automatic hierarchical storage device 500 provided in an embodiment of the present invention. The device 500 includes: The data acquisition unit 501 is used to acquire historical access data of an object and extract access popularity value based on the historical access data of the object. Triggering judgment unit 502 is used to determine whether to trigger a storage level migration decision based on the access popularity value; The demand forecasting unit 503 is used to forecast access demand using a time series forecasting model when a storage level migration decision is triggered. The hierarchical migration unit 504 is used to perform incremental migration operations based on the results of access demand prediction through an intelligent scheduling algorithm, and to perform hierarchical migration of objects according to a preset hierarchical strategy template.

[0045] This embodiment first acquires historical access data for the object and extracts the access popularity value. Then, based on the access popularity value, it determines whether a storage-level migration decision should be triggered. If triggered, a time series prediction model is used to predict access demand, and based on the prediction results, an incremental migration operation is executed through an intelligent scheduling algorithm. At the same time, a tiered migration is performed according to a preset tiered strategy template.

[0046] The automatic hierarchical storage method for objects provided in this embodiment can effectively solve the problems of low storage efficiency, difficulty in cost control, and poor user experience in existing technologies. Specifically, in terms of storage cost control, this embodiment accurately identifies data access patterns and dynamically adjusts storage levels. Compared with traditional time-driven strategies, this embodiment can promptly transfer cold data to lower-cost storage categories while preventing hot data from affecting access performance due to premature migration, thus achieving optimal allocation of storage costs. In terms of system performance improvement, this embodiment relies on a hierarchical strategy based on data popularity and a predictive management mechanism to ensure that hot data is always stored in high-performance storage categories, thereby significantly improving object access response speed and overall system throughput, while reducing unnecessary storage migration operations, reducing system resource consumption, and further improving operating efficiency. In terms of system stability assurance, this embodiment uses an adaptive learning mechanism and predictive management to ensure storage... The strategy can dynamically adapt to business changes, avoiding performance degradation caused by rigid strategies. Fine-grained monitoring and feedback control further enhance the system's self-adjustment capabilities, ensuring long-term stable and reliable operation. Regarding scalability and compatibility, this embodiment can be seamlessly integrated into existing object storage systems without requiring large-scale modifications to the underlying architecture, significantly reducing user deployment costs and technical risks. The flexible strategy configuration function can meet the storage management needs of enterprises of different sizes and various application scenarios. In terms of user experience optimization, the intelligent storage management provided in this embodiment enables the system to automatically identify and respond to various access patterns. Users can obtain more stable, efficient, and economical storage services without manual intervention in adjusting storage strategies, significantly improving user satisfaction and experience.

[0047] For example, in the medical field, the automatic hierarchical storage method for objects provided in this embodiment can automatically classify and store medical objects such as patients' electronic medical records and imaging data according to their access frequency. For instance, for patients with acute illnesses or those who have recently visited the hospital frequently, their relevant medical data has a high access frequency, so this data can be automatically stored in a high-performance storage tier, ensuring that doctors can quickly retrieve it when querying, thus gaining valuable time for timely treatment of patients. For data of patients with chronic illnesses or those who have not had follow-up visits for a long time, the access frequency is relatively low, so it can be migrated to a lower-cost storage tier, reducing storage costs while ensuring data accessibility. Simultaneously, by using time series prediction models to predict patients' follow-up visit times and examination items, potentially accessed data can be migrated to appropriate storage tiers in advance, further improving data access efficiency and providing strong support for the continuity and efficiency of medical services. In the financial industry, taking a securities trading system as an example, the automatic hierarchical storage method for objects provided in this embodiment can also play a significant role. Securities trading generates a large amount of trading data, including order information, transaction records, and market data. The access frequency of trading data varies greatly for different stocks and different time periods. The method described in this embodiment allows for real-time acquisition of historical access data for these transaction data, extracting access popularity values. For data generated during peak trading hours for popular stocks, due to their extremely high access popularity, it can be immediately stored in high-speed, low-latency storage devices. This ensures the trading system can process this data quickly, avoiding transaction failures or delays caused by data access latency and guaranteeing smooth financial transactions. Conversely, for less popular stocks or historical trading data with lower access popularity, it can be stored in low-cost, high-capacity storage devices. Furthermore, time-series forecasting models can be used to predict the trading activity of different stocks over a future period, allowing for advance adjustments to data storage levels to cope with potential trading peaks. This improves the stability and reliability of the financial trading system, saves financial institutions significant storage costs, and enhances customer service quality. In addition, the automatic hierarchical storage method provided in this embodiment is also suitable for the e-commerce sector, given the massive amounts of product information and user access data. E-commerce platforms offer a wide variety of products, each with varying popularity and sales cycles, resulting in vastly different access popularity for their related data. For example, during promotional events, product detail pages and review data for popular items will be frequently accessed by a large number of users. By acquiring historical access records of this data and extracting high access frequency values, a storage-level migration decision can be triggered. This allows the data of these popular products to be quickly migrated to a high-performance storage tier, ensuring fast page loading for users browsing products and improving the shopping experience. Conversely, data on products that have been slow-moving for a long time or are highly seasonal, with lower access frequency, can be stored in lower-cost storage.At the same time, time series forecasting models are used to predict product sales trends and user access patterns, and the storage level of product data is adjusted in advance to ensure that relevant data can quickly respond to a large number of user access requests when the peak sales season arrives, avoiding user churn due to data access problems, and promoting the business development and sales growth of the e-commerce platform.

[0048] In particular, the automatic hierarchical storage method for objects provided in this embodiment is especially suitable for the S3 (Simple Storage Service) object storage system. As a widely used cloud storage solution, this system also faces challenges in storage efficiency, cost control, and user experience when storing massive amounts of data. This embodiment, through its automatic hierarchical storage mechanism integrated into the S3 object storage system, can intelligently adjust the storage tier based on the actual access frequency of the data. This not only optimizes the allocation of storage resources but also significantly improves the efficiency and response speed of data access.

[0049] In one embodiment, such as Figure 6 As shown, the data acquisition unit 501 includes: The feature extraction unit 601 is used to perform multi-dimensional feature extraction processing on the historical access data of the object to obtain corresponding multi-dimensional features; wherein, the multi-dimensional features include access frequency, access time, access continuity and popularity score. The first construction unit 602 is used to construct a multi-dimensional access behavior analysis model based on the multi-dimensional features; The weight allocation unit 603 is used to allocate weights to the multi-dimensional features in the multi-dimensional access behavior analysis model using a sliding window and exponential weighted average technique, and to use the weighted popularity score as the access popularity value.

[0050] This embodiment extracts and analyzes multi-dimensional features from historical access data of objects to more comprehensively and accurately reflect the access popularity of data. Specifically, access frequency reflects the number of times data is accessed within a certain time range and is one of the important indicators for measuring data popularity; access time reflects the specific time period during which data is accessed, and the access popularity of data may vary significantly at different times. For example, during specific time periods on weekdays, the access popularity of certain business-related data may be higher; access continuity describes the persistence of data access, and data with more consecutive accesses often has higher popularity; popularity score can comprehensively quantify the data popularity to distinguish different objects into different levels such as "high popularity," "moderate popularity," and "low popularity." In addition, in practical applications, more features can be extracted, such as pattern stability and access source distribution, to further improve the multi-dimensional feature system.

[0051] By constructing a multi-dimensional access behavior analysis model, these characteristics can be integrated and analyzed to more accurately grasp data access patterns. For example, whether access exhibits periodicity (e.g., daily scheduled access), suddenness (e.g., instantaneous peaks caused by marketing campaigns), or persistence (e.g., long-term active services). Simultaneously, employing sliding window and exponentially weighted averaging techniques to assign weights to multi-dimensional features allows for dynamic adjustment of the weights of each feature based on real-time data access patterns. This makes the calculated access popularity values ​​more closely reflect reality and effectively eliminates noise interference, avoiding misjudgments caused by abnormal access at a single point in time, thus providing a reliable basis for subsequent storage-level migration decisions.

[0052] In one embodiment, such as Figure 7 As shown, the trigger determination unit 502 includes: The popularity comparison unit 701 is used to compare the access popularity value with a preset popularity threshold. The first determination unit 702 is used to determine whether to trigger a storage level migration decision if the access popularity value is greater than or equal to the preset popularity threshold. The second determination unit 703 is used to determine that storage level migration decision is not triggered if the access popularity value is less than the preset popularity threshold.

[0053] This embodiment compares access frequency values ​​with a preset frequency threshold to easily and effectively determine whether a storage tier migration decision needs to be triggered. The preset frequency threshold can be flexibly set based on actual business needs, storage system performance, and cost. For example, during peak business periods or in scenarios with high performance requirements, the preset frequency threshold can be set relatively high to ensure that only data with truly high access frequency will trigger migration, thereby guaranteeing that data in the high-performance storage tier is genuinely frequently accessed and improving overall system performance. Conversely, in scenarios with relatively stable business or higher cost sensitivity, the preset frequency threshold can be appropriately lowered to include more data with a certain level of access frequency in the migration scope, further optimizing storage costs.

[0054] When the access popularity value is determined to be greater than or equal to the preset popularity threshold, a storage level migration decision is automatically triggered, such as migrating from Standard Storage to Intelligent Tiering or Infrequent Access Storage, so as to store the data in a storage level that better matches its current access pattern and achieve efficient utilization of storage resources. When the access popularity value is less than the preset popularity threshold, the current storage level of the data remains unchanged to avoid performance overhead and increased costs caused by unnecessary migration operations.

[0055] In one embodiment, such as Figure 8 As shown, the demand forecasting unit 503 includes: The sequence extraction unit 801 is used to extract corresponding time series features based on the historical access data of the object; The second construction unit 802 is used to construct a time series prediction model based on the time series features; The specified prediction unit 803 is used to predict access demand within a specified time period using the time series prediction model, and obtain access demand prediction results.

[0056] This embodiment first extracts time-series features from historical access data of an object. Time-series features can capture the patterns and trends of data access over time. For example, access to certain data may exhibit obvious periodicity, such as higher access volume at specific times of the day; or it may be seasonal, such as a significant increase in access volume for certain products during specific seasons. By extracting these time-series features, we can gain a more accurate understanding of data access patterns. Then, a time-series prediction model is built based on the extracted time-series features. This model can employ different algorithmic architectures, such as the Autoregressive Integral Moving Average (ARIMA) model, Long Short-Term Memory (LSTM) network, or Prophet (a time-series prediction algorithm), Transformer-based models, etc. These models can predict access demand for a specified period in the future based on the time-series features in historical data. For example, by analyzing data access patterns over the past few months, we can predict the daily access volume for the next week to identify access peaks or troughs in advance, proactively adjust storage tiers, and avoid frequent migrations and performance fluctuations caused by sudden access surges.

[0057] Subsequently, the established time-series forecasting model is used to predict access demand within a specified time period, yielding the access demand forecast results. These forecasts provide crucial information for subsequent storage tier migration decisions. For example, if a significant increase in access to certain data is predicted in the future, this data can be migrated to a high-performance storage tier in advance to ensure rapid response during peak access periods and avoid performance issues caused by data access latency. Furthermore, accurate access demand forecasting can prevent excessive data migration to high-performance storage tiers, thereby reducing storage costs.

[0058] In one embodiment, the hierarchical migration unit 504 includes: The load acquisition unit is used to acquire system load status information and determine the execution time of incremental migration operation based on the load status information and an intelligent scheduling algorithm. The incremental migration unit is used to perform incremental migration operations based on the execution time and utilizing idle system resources.

[0059] In this embodiment, by acquiring system load status information during migration, the system resource utilization can be monitored in real time, avoiding incremental migration operations during periods of high system load and thus minimizing the impact on normal business processing. The intelligent scheduling algorithm can accurately determine the execution time of incremental migration operations based on system load status information, such as selecting nighttime periods with lower system load or relatively idle business periods for migration. Based on the determined execution time, incremental migration operations are performed using idle system resources. This fully utilizes idle system resources, improving resource utilization without placing additional burden on the system. This intelligent scheduling method based on system load status ensures the smooth execution of incremental migration operations while guaranteeing stable system operation, effectively avoiding system performance degradation caused by migration operations.

[0060] In one embodiment, the hierarchical migration unit 504 further includes: The strategy selection unit is used to select a hierarchical strategy from a preset hierarchical strategy template in response to the user's migration command or through the intelligent scheduling algorithm; wherein the hierarchical strategy includes a high-activity-fast response strategy, a low-frequency-low-cost strategy, and a periodic-on-demand warm-up strategy.

[0061] The strategy migration unit is used to perform hierarchical migration of objects according to the hierarchical strategy.

[0062] This embodiment provides different hierarchical strategy options based on a preset hierarchical strategy template to adapt to storage management needs under different business scenarios and data characteristics. For example, it offers various strategies such as the high-activity-fast response strategy, the low-frequency-low-cost strategy, and the periodic-on-demand warm-up strategy. The high-activity-fast response strategy is primarily for data objects with extremely high access frequency and stringent response time requirements. For instance, in a real-time transaction system, order processing data needs to be accessed and processed at extremely high speeds. This strategy ensures that this data is always stored in the optimal storage tier, such as a high-speed solid-state drive array, thereby guaranteeing the efficient and stable operation of the transaction system. The low-frequency-low-cost strategy is suitable for data with low access frequency and high cost sensitivity. For example, some historical archived data may not be accessed for a long time but needs to be retained for later retrieval. This strategy allows these data to be migrated to low-cost, high-capacity storage devices, such as tape libraries or low-speed hard drives, significantly reducing storage costs while maintaining data retrievability. The periodic-on-demand warm-up strategy is for data with obvious periodic access characteristics. Taking seasonal product data from e-commerce platforms as an example, the access volume of these products increases significantly before the arrival of a specific season. Using this strategy, product data that is about to enter its peak access period can be identified in advance based on time series prediction models, and then preheated and migrated to a high-performance storage tier to ensure a rapid response when the peak access period arrives, avoiding user churn due to data loading delays.

[0063] Once a user's migration command is received or a hierarchical strategy is selected from a preset hierarchical strategy template via an intelligent scheduling algorithm, the objects can be migrated hierarchically according to the selected strategy. During the migration process, the migration path and resource allocation can be rationally planned based on factors such as the current storage location of the data, the target storage level, and the size of the data, ensuring the efficiency and stability of the migration process.

[0064] Simultaneously, a full-link monitoring system can be established to collect key indicators such as access latency, throughput, migration success rate, and resource consumption of objects at each level in real time. This allows for real-time monitoring and recording of the migration process, enabling timely handling and recovery in case of anomalies, ensuring data integrity and availability. For example, if the migration failure rate exceeds a threshold or latency increases by more than 30%, an alarm is immediately triggered and a compensation mechanism is initiated, such as pausing migration, rolling back strategies, or switching to a backup solution, to ensure service stability. This enables closed-loop management of the entire process from data awareness, intelligent decision-making, automated execution to dynamic feedback, significantly improving the intelligence level and operational efficiency of the storage system.

[0065] In one embodiment, the object auto-leveling storage device 500 further includes: The dynamic evaluation unit is used to dynamically evaluate objects after hierarchical transfer using online machine learning mechanisms. An adaptive optimization unit is used to adaptively optimize the heat threshold based on the results of dynamic evaluation.

[0066] This embodiment introduces an online machine learning mechanism to continuously and dynamically evaluate objects after hierarchical migration. The online machine learning mechanism can update model parameters in real time as data is continuously generated and changed, without requiring the collection of large amounts of data for full training. This allows the system to promptly capture new changes in object access patterns. For example, some data that was previously not frequently accessed may suddenly become popular due to changes in the market environment or business promotions. The online machine learning mechanism can quickly detect this change and accurately evaluate it.

[0067] Based on the results of dynamic evaluation, the preset heat threshold can be adaptively optimized. The heat threshold is a key indicator for determining whether to trigger a storage tier migration decision. If the heat threshold is set unreasonably, it may lead to frequent invalid migrations or missed important migration opportunities. Through adaptive optimization, the heat threshold can be automatically adjusted according to actual data access patterns and business needs. For example, if the system finds that after migrating according to the current heat threshold, some data with low access frequency still exists in the high-performance storage tier, while some data with rapidly increasing access frequency in the low-performance storage tier has not been migrated in time, the heat threshold can be appropriately lowered. This allows more data with a certain level of access frequency to be migrated to the appropriate storage tier, improving the utilization efficiency of storage resources. Conversely, if data access is frequent in the high-performance storage tier and system resources are strained, the heat threshold can be appropriately increased to reduce unnecessary migration operations and ensure stable system operation. This adaptive optimization mechanism enables the storage system to automatically adjust its strategy according to actual conditions, forming a "perception-decision-execution-feedback" closed loop. This gives the system continuous learning and self-evolution capabilities, adapting to constantly changing business models.

[0068] Please see Figure 9 , Figure 9 This is a schematic block diagram of a computer device provided in an embodiment of the present invention. The computer device is equipped with both wireless and wired communication capabilities.

[0069] The computer device includes a processor 902, a memory, and a network interface 905 connected via a system bus 901. The memory may include a non-volatile storage medium 903 and internal memory 904.

[0070] The non-volatile storage medium 903 can store an operating system 9031 and a computer program 9032. When the computer program 9032 is executed, it causes the processor 902 to execute an automatic hierarchical storage method for objects.

[0071] The processor 902 provides computing and control capabilities to support the operation of the entire computer device.

[0072] The internal memory 904 provides an environment for the execution of the computer program 9032 in the non-volatile storage medium 903. When the computer program 9032 is executed by the processor 902, the processor 902 can execute an automatic hierarchical storage method for objects.

[0073] This network interface 905 is used for network communication with other devices. Those skilled in the art will understand that... Figure 9 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the computer device to which the present invention is applied. A specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0074] The processor 902 is used to run a computer program 9032 stored in the memory to implement any embodiment of the above-described automatic hierarchical storage method for objects.

[0075] It should be understood that, in this embodiment of the invention, the processor 902 may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0076] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps: Obtain historical access data for an object, and extract access popularity values ​​based on the historical access data. Determine whether to trigger a storage-level migration decision based on the access popularity value; When a storage-level migration decision is triggered, a time-series forecasting model is used to predict access demand. Based on the results of access demand prediction, incremental migration operations are performed through intelligent scheduling algorithms, and objects are migrated in a hierarchical manner according to a preset hierarchical strategy template.

[0077] This invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed, can perform the steps provided in the above embodiments. The storage medium may include various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0078] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor: Obtain historical access data for an object, and extract access popularity values ​​based on the historical access data. Determine whether to trigger a storage-level migration decision based on the access popularity value; When a storage-level migration decision is triggered, a time-series forecasting model is used to predict access demand. Based on the results of access demand prediction, incremental migration operations are performed through intelligent scheduling algorithms, and objects are migrated in a hierarchical manner according to a preset hierarchical strategy template.

[0079] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0080] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0081] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0082] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. An object auto-classification storage method, characterized by, include: Obtain historical access data for an object, and extract access popularity values ​​based on the historical access data. Determine whether to trigger a storage-level migration decision based on the access popularity value; When a storage-level migration decision is triggered, a time-series forecasting model is used to predict access demand. Based on the results of access demand prediction, incremental migration operations are performed through intelligent scheduling algorithms, and objects are migrated in a hierarchical manner according to a preset hierarchical strategy template.

2. The object automated storage and retrieval method of claim 1, wherein, The step of obtaining historical access data for an object and extracting access popularity values ​​based on the historical access data includes: Multi-dimensional feature extraction processing is performed on the historical access data of the object to obtain corresponding multi-dimensional features; wherein, the multi-dimensional features include access frequency, access time, access continuity and popularity score; A multi-dimensional access behavior analysis model is constructed based on the aforementioned multi-dimensional features; The sliding window and exponential weighted average techniques are used to assign weights to the multi-dimensional features in the multi-dimensional access behavior analysis model, and the weighted popularity score is used as the access popularity value.

3. The method of claim 1, wherein, The step of determining whether to trigger a storage-level migration decision based on the access popularity value includes: The access popularity value is compared with a preset popularity threshold; If the access popularity value is greater than or equal to the preset popularity threshold, then a storage level migration decision is triggered. If the access popularity value is less than the preset popularity threshold, it is determined that the storage level migration decision will not be triggered.

4. The method of claim 1, wherein, When a storage level migration decision is triggered, the method of using a time series forecasting model to predict access demand includes: Extract the corresponding time series features based on the object's historical access data; A time series prediction model is constructed based on the aforementioned time series features; The time series prediction model is used to predict access demand within a specified time period, and the access demand prediction results are obtained.

5. The method of claim 4, wherein, Based on the results of access demand prediction, incremental migration operations are performed through an intelligent scheduling algorithm, and objects are migrated in a hierarchical manner according to a preset hierarchical strategy template, including: Obtain system load status information, and determine the execution time of incremental migration operation based on the load status information using an intelligent scheduling algorithm; Based on the execution time, incremental migration operations are performed using idle system resources.

6. The method of claim 4, wherein, The process of performing incremental migration operations based on access demand prediction results using an intelligent scheduling algorithm, and migrating objects in a hierarchical manner according to a preset hierarchical strategy template, also includes: In response to the user's migration command or through the intelligent scheduling algorithm, a hierarchical strategy is selected from the preset hierarchical strategy template; wherein, the hierarchical strategy includes a high-activity-fast response strategy, a low-frequency-low-cost strategy, and a periodic-on-demand warm-up strategy; The objects are migrated in a hierarchical manner according to the described hierarchical strategy.

7. The method of claim 3, wherein, Also includes: Utilize online machine learning mechanisms to dynamically evaluate objects after graded transfer; The heat threshold is adaptively optimized based on the results of dynamic evaluation.

8. An object auto-classification storage device, characterized by, include: The data acquisition unit is used to acquire historical access data of an object and extract access popularity values ​​based on the historical access data of the object. A triggering judgment unit is used to determine whether to trigger a storage level migration decision based on the access popularity value. The demand forecasting unit is used to forecast access demand using a time series forecasting model when a storage-level migration decision is triggered. The hierarchical migration unit is used to perform incremental migration operations based on the results of access demand prediction through an intelligent scheduling algorithm, and to migrate objects hierarchically according to a preset hierarchical strategy template.

9. A computer device, comprising: The device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the automatic hierarchical storage method for objects as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the automatic hierarchical storage method for objects as described in any one of claims 1 to 7.

Images