Storage resource adjustment method and device, equipment, storage medium and program product

Abstract

The invention discloses a storage resource adjustment method and device, equipment, a storage medium and a program product, and relates to the technical field of big data. The method comprises the following steps: acquiring first storage information of a distributed storage cluster and second storage information of a virtual machine node, wherein the distributed storage cluster is used for providing storage resources for the virtual machine node; performing feature processing on the first storage information and the second storage information to obtain N feature items; the N feature items are related to the KVM; inputting the target input information into a pre-trained distribution water level prediction model to obtain a distribution water level prediction result corresponding to the to-be-predicted target time period; and under the condition that the distribution water level prediction result is lower than the current distribution water level, performing simplification operation on the storage data in the virtual machine node based on the distribution water level prediction result. According to the embodiment of the invention, reasonable storage and management of the data in the virtual machine can be more fully and effectively realized, so that the purpose of accurately reducing the utilization rate of storage resources is achieved.

Description

Technical Field

[0001] The present application belongs to the field of big data technology, and in particular, relates to a storage resource adjustment method, device, equipment, storage medium and program product. Background Art

[0002] With the advent of the big data era, the amount of data has exploded. To meet the needs of storage resource adjustment, distributed storage systems have been widely used. However, at present, the traditional KVM (Kernel-based Virtual Machine) virtual machine will allocate a certain amount of storage space for each KVM when allocating distributed storage. In order to improve the utilization rate of storage resources, distributed storage systems usually adopt automatic thin configuration. Based on distributed hash technology, after setting automatic thin configuration, when allocating storage resources, only the storage space required for storing data is delivered, and the storage space that has been allocated but not used is not delivered. Users always see the configured capacity, but the actual physical disk space is dynamically adjusted with the used space. After long-term use, the storage usage continues to increase. In order to ensure the overall availability of storage and avoid the use of redundant storage due to hard disk failure, maintenance personnel will set a storage usage threshold based on experience. When the usage reaches the threshold, an alarm is triggered. Under the above traditional solution, since the watermark is set by the operation and maintenance personnel based on experience, it is often easy to have unreasonable situations, resulting in excessive allocation of storage resources, causing the traditional distributed storage system to have the problem of high allocation and low utilization, resulting in low utilization of storage resources and high storage costs.

[0003] Based on this, the industry is still in urgent need of a distributed storage resource adjustment solution based on the KVM environment to fully and effectively realize the reasonable storage and management of data in virtual machines, so as to achieve the purpose of accurately reducing the utilization rate of storage resources. Summary of the invention

[0004] The embodiments of the present application provide a storage resource adjustment method, apparatus, device, storage medium and program product, which can more fully and effectively realize the reasonable storage and management of data in a virtual machine, thereby achieving the purpose of accurately reducing the utilization rate of storage resources.

[0005] In a first aspect, an embodiment of the present application provides a storage resource adjustment method, the storage resource adjustment method comprising:

[0006] Obtaining first storage information of a distributed storage cluster and second storage information of a virtual machine node at a current moment, where the distributed storage cluster is used to provide storage resources for the virtual machine node;

[0007] Based on the first storage information and the second storage information, N feature items are obtained by processing, where N is a positive integer; the N feature items are related to a kernel-based virtual machine KVM;

[0008] Input the target input information into the pre-trained allocation water level prediction model to obtain the allocation water level prediction result corresponding to the future target time period to be predicted, where the target input information is obtained based on N feature items and time information corresponding to the future target time period;

[0009] When the allocation waterline prediction result is lower than the current allocation waterline, a slimming operation is performed on the storage data in the virtual machine node, and / or the virtual machine node is migrated.

[0010] In some possible implementations, before inputting the target input information into a pre-trained allocation water level prediction model to obtain an allocation water level prediction result corresponding to a future target time period to be predicted, the storage resource adjustment method further includes:

[0011] Obtaining first historical storage information of the distributed storage cluster and second historical storage information of the virtual machine node at multiple historical moments;

[0012] Based on the first historical storage information and the second storage information, generate a plurality of data samples corresponding to a plurality of historical moments one by one;

[0013] Based on multiple data samples, the bidirectional long short-term memory network Bi-LSTM time series model is used for regression training to obtain the distribution water level prediction model.

[0014] In some possible implementations, the N feature items include: at least one of KVM utilization, the number of read and write operations per second (IOPS) of stored data, the number of nodes, the memory size, the hard disk size, the CPU utilization, the memory utilization, the network bandwidth utilization, and the storage utilization.

[0015] In some possible implementations, performing a thinning operation on storage data in a virtual machine node includes:

[0016] Deleting invalid data bits in the storage data blocks under the virtual machine nodes to generate streamlined storage data blocks;

[0017] After performing a thinning operation on the storage data in the virtual machine node, the storage resource adjustment method further includes:

[0018] The address information of the streamlined storage data block is cached in the high-speed cache of the distributed storage cluster for data reading processing.

[0019] In some possible implementations, after deleting invalid data bits in a storage data block under a virtual machine node to generate a streamlined storage data block, migrating the virtual machine node includes:

[0020] The streamlined storage data blocks are stored in the new storage space;

[0021] Migrate the VM nodes to the new storage space and delete the original storage space after the migration is complete.

[0022] In some possible implementations, before inputting the target input information into a pre-trained allocation water level prediction model to obtain an allocation water level prediction result corresponding to a future target time period to be predicted, the storage resource adjustment method further includes:

[0023] Convert the attribute values ​​corresponding to the N feature items and the time information into a first attribute value set, and perform normalization and data alignment processing on the attribute values ​​in the attribute value set to obtain a second attribute value set;

[0024] The second attribute value set is converted into a target vector group by performing one-hot encoding, and the target vector group is determined as target input information.

[0025] Based on the same inventive concept, in a second aspect, an embodiment of the present application provides a storage resource adjustment device, the storage resource adjustment device comprising:

[0026] A first acquisition module is used to acquire first storage information of a distributed storage cluster and second storage information of a virtual machine node at a current moment, where the distributed storage cluster is used to provide storage resources for the virtual machine node;

[0027] A first processing module is used to process and obtain N feature items based on the first storage information and the second storage information, where N is a positive integer; the N feature items are related to a kernel-based virtual machine KVM;

[0028] The first prediction module is used to input the target input information into the pre-trained distribution water level prediction model to obtain the distribution water level prediction result corresponding to the future target time period to be predicted, and the target input information is obtained based on N feature items and time information corresponding to the future target time period;

[0029] The first streamlining module is used to perform a streamlining operation on the storage data in the virtual machine node and / or migrate the virtual machine node when the allocation waterline prediction result is lower than the current allocation waterline.

[0030] In a third aspect, an embodiment of the present application provides a storage resource adjustment device, the storage resource adjustment device comprising:

[0031] a processor and a memory storing computer program instructions;

[0032] When the processor executes the computer program instructions, the storage resource adjustment method provided in any one of the above-mentioned embodiments of the present application is implemented.

[0033] In a fourth aspect, an embodiment of the present application provides a computer storage medium, on which computer readable storage medium is stored computer program instructions, and when the computer program instructions are executed by a processor, a storage resource adjustment method as provided in any one of the above-mentioned embodiments of the present application is implemented.

[0034] In a fifth aspect, an embodiment of the present application provides a computer program product. When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes a storage resource adjustment method provided in any one of the above-mentioned embodiments of the present application.

[0035] The embodiments of the present application provide a storage resource adjustment method, device, equipment, storage medium and program product, which obtains the first storage information of the distributed storage cluster and the second storage information of the virtual machine node at the current moment, and processes and obtains N feature items related to KVM based on the first storage information and the second storage information. In this way, the target input information obtained by processing the time information corresponding to the future target time period based on the N feature items and the future target time period is input into the pre-trained allocation water level prediction model to obtain the allocation water level prediction result corresponding to the future target time period to be predicted. Finally, when the allocation water level prediction result is lower than the current allocation water level, the storage data in the virtual machine node is streamlined, and / or the virtual machine node is migrated.

[0036] From the above description, it can be seen that a storage resource adjustment method, device, equipment, storage medium and program product of an embodiment of the present application obtains the first storage information of the distributed storage cluster and the second storage information of the virtual machine node, and performs data feature processing on the above information to retain the feature information related to KVM, and then makes an intelligent and accurate prediction of the allocation waterline of the future target time period based on the feature information and the pre-trained allocation waterline prediction model. In this way, the data can be dynamically streamlined in a timely manner according to the current storage usage and the KVM resources can be reallocated based on the migration method, so that the reasonable storage and management of the data in the virtual machine can be more fully and effectively realized, thereby achieving the purpose of accurately reducing the utilization rate of storage resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solution of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0038] Figure 1 This is a schematic diagram of a traditional KVM virtual machine storage allocation method provided in an embodiment of the present application;

[0039] Figure 2 It is a flowchart of a storage resource adjustment method provided in an embodiment of the present application;

[0040] Figure 3 This is a diagram of the overall architecture of a storage resource adjustment method provided in one embodiment of the present application;

[0041] Figure 4 It is an overall flow chart of a storage resource adjustment method provided in an embodiment of the present application;

[0042] Figure 5 It is a schematic diagram of a monitoring data list of a storage resource adjustment method provided in an embodiment of the present application;

[0043] Figure 6 It is a schematic diagram of a distributed queue process of a storage resource adjustment method provided in an embodiment of the present application;

[0044] Figure 7 It is a schematic diagram of the Bi-LSTM model construction structure of the storage resource adjustment method provided in one embodiment of the present application;

[0045] Figure 8 It is a schematic diagram of a dynamic streamlining process of a storage resource adjustment method provided in an embodiment of the present application;

[0046] Fig. 9 This is a schematic diagram of a data dynamic reduction processing method of a storage resource adjustment method provided in an embodiment of the present application;

[0047] Fig.10 is a schematic diagram of an address data structure of a storage resource adjustment method provided in an embodiment of the present application;

[0048] Fig.11 It is a schematic diagram of a KVM storage migration architecture of a storage resource adjustment method provided in an embodiment of the present application;

[0049] Fig.12 is a structural diagram of a storage resource adjustment device provided in an embodiment of the present application;

[0050] Fig.13 It is a structural diagram of a storage resource adjustment device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0051] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present application by illustrating the examples of the present application.

[0052] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "include..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0053] As described in the background technology section, at present, when allocating distributed storage, traditional KVM virtual machines will allocate a certain amount of storage space for each KVM. In order to improve the utilization rate of storage resources, distributed storage systems usually adopt automatic thin configuration. Based on distributed Hash technology, after setting automatic thin configuration, when allocating storage resources, only the storage space required for storing data is delivered, and the storage space that has been allocated but not used is not delivered. What users see is always the configuration capacity, but the actual physical disk space occupied is dynamically adjusted with the used space. After long-term use, the storage usage continues to increase. In order to ensure the overall availability of storage and avoid the use of redundant storage due to hard disk failure, maintenance personnel will set a storage usage threshold based on experience (for example, the storage usage reaches 75%). When the usage reaches this threshold, an alarm is triggered, and some necessary measures are taken to reduce the storage usage to avoid storage capacity overload, and this threshold is the water level.

[0054] For example Figure 1As shown in the figure, assume that in the initial use, the distributed storage system allocates 40GB (Gigabyte) of storage space to each of the two KVM virtual machines, and the two KVM virtual machines use 15GB of storage space each; then in the later use, due to the different storage resource usage of different KVM virtual machines, one of the two KVM virtual machines has already filled up 40GB, while the storage resource usage of the other KVM virtual machine is still 15GB, thus forming a situation of high allocation and low usage. Figure 1 It can be seen that the existing technical solution has unreasonable allocation waterlines because the allocation waterline is set by the operation and maintenance personnel based on experience, resulting in over-allocation of storage resources. In addition, since the current storage automatic reduction method does not have a dynamic storage capacity adjustment function, there is a certain waste of data storage, resulting in a low storage utilization rate.

[0055] In view of the above, in order to solve the above problems, the present invention uses a distributed storage dynamic streamlining device in a KVM environment, analyzes the operation and configuration data of KVM storage based on machine learning technology, intelligently predicts the storage allocation waterline and dynamically streamlines it, replacing the traditional manual allocation method, and dynamically adjusts the storage capacity based on the migration method, which can effectively solve the problems of low storage resource utilization and high storage costs caused by high allocation and low usage.

[0056] In order to solve the above-mentioned prior art problems, the embodiments of the present application provide a storage resource adjustment method, device, equipment, storage medium and program product. It should be noted that the embodiments provided in the present application are not intended to limit the scope of the present application.

[0057] The following first introduces the storage resource adjustment method provided in the embodiment of the present application.

[0058] Figure 2 The flowchart of the storage resource adjustment method provided by an embodiment of the present application is shown. The storage resource adjustment method is applied to an electronic device, which may include a server or a user terminal. Figure 2 As shown, the storage resource adjustment method includes the following steps:

[0059] S210, obtaining first storage information of a distributed storage cluster and second storage information of a virtual machine node at a current moment, where the distributed storage cluster is used to provide storage resources for the virtual machine node;

[0060] S220, based on the first storage information and the second storage information, obtain N feature items by processing, where N is a positive integer; the N feature items are related to a kernel-based virtual machine KVM;

[0061] S230, inputting the target input information into a pre-trained allocation water level prediction model to obtain an allocation water level prediction result corresponding to the future target time period to be predicted, wherein the target input information is obtained based on N feature items and time information corresponding to the future target time period;

[0062] S240: When the allocation waterline prediction result is lower than the current allocation waterline, a slimming operation is performed on the storage data in the virtual machine node, and / or the virtual machine node is migrated.

[0063] The embodiment of the present application provides a storage resource adjustment method, which obtains the first storage information of the distributed storage cluster and the second storage information of the virtual machine node at the current moment, and processes N feature items related to KVM based on the first storage information and the second storage information. In this way, the target input information obtained by processing the time information corresponding to the future target time period based on the N feature items is input into the pre-trained allocation water level prediction model to obtain the allocation water level prediction result corresponding to the future target time period to be predicted. Finally, when the allocation water level prediction result is lower than the current allocation water level, the storage data in the virtual machine node is streamlined, and / or the virtual machine node is migrated.

[0064] From the above description, it can be seen that a storage resource adjustment method of an embodiment of the present application obtains the first storage information of the distributed storage cluster and the second storage information of the virtual machine node, and performs data feature processing on the above information to retain the feature information related to KVM, and then makes an intelligent and accurate prediction of the allocation waterline of the future target time period based on the feature information and the pre-trained allocation waterline prediction model. In this way, the data can be dynamically streamlined in a timely manner according to the current storage usage, and the KVM resources can be reallocated based on the migration method, so that the reasonable storage and management of the data in the virtual machine can be more fully and effectively realized, thereby achieving the purpose of accurately reducing the utilization rate of storage resources.

[0065] The specific implementation of the above steps 210 to 240 is described in detail below.

[0066] In S210, when specifically implemented, the first storage information of the distributed storage cluster and the second storage information of the virtual machine node at the current moment are obtained. The distributed storage cluster is used to provide storage resources for the virtual machine node. Considering the actual data acquisition scenario, in this embodiment, the current moment may also refer to the latest collection moment of the storage information, and this application does not impose strict restrictions on this.

[0067] More specifically, the above-mentioned distributed storage cluster provides storage resources, data access, data protection and performance optimization services for virtual machine nodes, while the virtual machine nodes interact with the cluster through specific interfaces or protocols to manage and use their storage resources.

[0068] The first storage information corresponds to the storage information of the distributed storage cluster, and the first storage information mainly refers to the utilization of the processor, memory, hard disk, network card and other hardware of each server node in the distributed storage cluster. The second storage information corresponds to the storage information of the virtual machine node in the KVM environment, and the second storage information may include the storage utilization of the virtual machine node and its data reading status, etc., which is not strictly limited in this embodiment.

[0069] According to some embodiments of the present application, optionally, considering the high concurrency of received data, in order to ensure normal data reception and security, a distributed processing task queue can be used to distribute tasks. Specifically, before obtaining the first storage information of the distributed storage cluster and the second storage information of the virtual machine node at the current moment, the storage resource adjustment method also includes:

[0070] Establish a distributed multi-node message queue;

[0071] Obtaining the first storage information of the distributed storage cluster and the second storage information of the virtual machine node at the current moment includes:

[0072] The first storage information and the second storage information are received through a multi-node message queue.

[0073] In this embodiment, the first storage information and the second storage information are received in order through a multi-node message queue. Further, the message queue can also be used for storage data distribution and transmission in subsequent steps, which is not strictly limited here.

[0074] In this embodiment, by distributing tasks (such as data reception and data distribution, etc.) in the form of distributed processing task queues, the message queue can ensure the order and consistency of request messages, while also being able to fully alleviate system pressure, which is beneficial to improving system availability and stability.

[0075] In S220, in a specific implementation, based on the first storage information and the second storage information, N feature items are processed and obtained, where N is a positive integer. The N feature items are related to the kernel-based virtual machine KVM.

[0076] According to some embodiments of the present application, optionally, based on the first storage information and the second storage information, N feature items are obtained by processing, including:

[0077] Performing data cleaning and denoising processing on the first storage information and the second storage information to obtain pre-processed storage information;

[0078] Perform feature engineering on the preprocessed storage information and screen out N feature items.

[0079] In this embodiment, after obtaining the first storage information and the second storage information, invalid, erroneous or duplicate data that may affect the data quality are identified and removed. The cleaning process also includes filling missing values, processing abnormal values, and format normalization to ensure that the data is consistent throughout the entire processing flow. This meticulous cleaning process helps prevent erroneous analysis and inaccurate conclusions caused by dirty data.

[0080] After data cleaning and denoising are performed on the first storage information and the second data information, feature engineering processing can be performed on the pre-processed storage information to screen out N feature items. The feature engineering processing can include, for example, feature processing and feature selection.

[0081] When performing feature processing, the mean method can be used to fill in the missing values ​​in some important features. At the same time, combined with the business, considering that the original feature information such as network throughput contains redundant information and abnormal values, the dimensionality reduction operation can be performed on this type of original feature information to improve the accuracy of subsequent predictions and filter out noise. When performing feature selection, useless feature items can be deleted, and feature items related to KVM, such as network traffic, fluctuation range, storage data read and write times and transmission volume, node memory and hard disk utilization, etc., can be retained to determine whether the distributed storage cluster and virtual machine node operation status are normal.

[0082] More specifically, in combination with the actual KVM scenario, the above N feature items may specifically include: KVM utilization, the number of read and write operations performed per second of stored data IOPS, number of nodes, memory size, hard disk size, CPU (Central Processing Unit / Processor) utilization, memory utilization, network bandwidth utilization, and at least one of storage utilization.

[0083] In S230, during the specific implementation, the target input information is input into a pre-trained allocation water level prediction model to obtain an allocation water level prediction result corresponding to the future target time period to be predicted, and the target input information is obtained based on N feature items and time information corresponding to the future target time period.

[0084] According to some embodiments of the present application, optionally, in order to more reasonably construct the above-mentioned allocation water level prediction model for predicting the allocation water level, thereby effectively improving the reliability of the prediction result, before the target input information is input into the pre-trained allocation water level prediction model to obtain the allocation water level prediction result corresponding to the future target time period to be predicted, the storage resource adjustment method also includes:

[0085] Obtaining first historical storage information of the distributed storage cluster and second historical storage information of the virtual machine node at multiple historical moments;

[0086] Based on the first historical storage information and the second storage information, generate a plurality of data samples corresponding to a plurality of historical moments one by one;

[0087] Based on multiple data samples, the bidirectional long short-term memory network Bi-LSTM time series model is used for regression training to obtain the distribution water level prediction model.

[0088] In this embodiment, considering the need to predict the future storage allocation water level line in the future, the temporal model Bi-LSTM in deep learning is used to train the regression model so that the temporal model can learn more about the data features.

[0089] It is understandable that in some other feasible implementations, other time series models or training based on neural network algorithms can be used to obtain the above-mentioned distribution water level prediction model, and it is not limited to using the deep learning model Bi-LSTM time series model for regression training.

[0090] According to some embodiments of the present application, optionally, before inputting the target input information into a pre-trained allocation water level prediction model to obtain an allocation water level prediction result corresponding to a future target time period to be predicted, the storage resource adjustment method further includes:

[0091] Convert the attribute values ​​corresponding to the N feature items and the time information into a first attribute value set, and normalize and align the attribute values ​​in the attribute value set to obtain a second attribute value set;

[0092] The second attribute value set is converted into a target vector group by performing one-hot encoding, and the target vector group is determined as target input information.

[0093] In this embodiment, the above N feature items are converted into sets, and then normalized to align the values. Then, these sets are one-hot encoded and converted into vector groups, which are used as inputs of the above allocation waterline prediction model to fully improve the model prediction efficiency.

[0094] In some other embodiments, other encoding forms may be used to perform numerical conversion on the attribute values ​​of the above-mentioned feature items, such as sub-coding, etc., and this application does not impose strict restrictions on this.

[0095] In S240, during specific implementation, when the allocation waterline prediction result is lower than the current allocation waterline, a slimming operation is performed on the storage data in the virtual machine node, and / or the virtual machine node is migrated.

[0096] More specifically, in order to fully implement the thinning operation on the storage data to achieve the purpose of compressing the storage space, the thinning operation on the storage data in the virtual machine node may include:

[0097] Deleting invalid data bits in the storage data blocks under the virtual machine nodes to generate streamlined storage data blocks;

[0098] After performing a slimming operation on the storage data in the virtual machine node, the storage resource adjustment method may further include:

[0099] The address information of the streamlined storage data block is cached in the high-speed cache of the distributed storage cluster for data reading processing.

[0100] The address data structure of the address information may be composed of a "tag" field, a "block index" field, and a "bit offset" field. Specifically, the tag field is used to mainly record the identification of the stored file as a globally unique identification of the file in storage. The block index field is used to record the position index of the simplified data block in the file, so as to facilitate the rapid search for the simplified data block. The bit offset field is used to record the offset of the simplified data bit of the simplified data block, so as to facilitate data reading.

[0101] In this way, the storage data is dynamically streamlined through the above-mentioned allocation water level prediction results. When the allocation water level is lower than the current allocation water level, the streamlined operation is started, the invalid data bits in the storage data block under the virtual machine node are deleted to compress the storage space, and the streamlined storage address is cached in the cache, thereby achieving streamlined compression of the data and dynamic adjustment of storage resources, which is beneficial to ensuring the availability of storage space and effectively improving the utilization rate of storage resources.

[0102] According to some embodiments of the present application, optionally, even if data has been streamlined and compressed, some no longer needed data or redundant data may still be accumulated over time and with the operation of the virtual machine. Based on this, in order to further optimize the utilization of storage resources, after deleting the invalid data bits in the storage data blocks under the virtual machine nodes and generating the streamlined storage data blocks, the virtual machine nodes are migrated, including:

[0103] The streamlined storage data blocks are stored in the new storage space;

[0104] Migrate the VM nodes to the new storage space and delete the original storage space after the migration is complete.

[0105] In this embodiment, by migrating data to new thin storage and further cleaning and compressing data in the process, the utilization of storage resources can be further optimized. In addition, by migrating virtual machines to new thin storage, it is beneficial to fully realize the integration of storage resources, so that storage resources are more effectively utilized.

[0106] In order to facilitate understanding of the storage resource adjustment method provided in the above embodiment, the above method is described below using a specific scenario embodiment. Figure 3 as well as Figure 4 , Figure 3 is a diagram showing the overall architecture of a storage resource adjustment method provided in an embodiment of the present application. Figure 4 It is an overall flow chart of a storage resource adjustment method provided in one embodiment of the present application.

[0107] The purpose of this embodiment is to dynamically and scientifically predict and dynamically adjust the KVM storage usage in the future time period through the distributed storage dynamic thinning device in the KVM environment, so as to reduce the storage allocation rate and improve the storage utilization rate. The architecture diagram of this solution is as follows: Figure 3 As shown, Figure 4 The following is a flow chart of the business resource allocation method. Figure 3 as well as Figure 4 As shown, the specific module composition and implementation functions of the distributed storage dynamic thinning device under the KVM environment are as follows:

[0108] (I) Storage monitoring module

[0109] The storage monitoring module is mainly responsible for obtaining the operating status of the virtual machine node and the distributed storage cluster through the KVM management node and the distributed storage (cluster), and sending data to the data receiving module as needed.

[0110] See also Figure 5 , Figure 5It is a schematic diagram of a monitoring data list of a storage resource adjustment method provided in an embodiment of the present application. In the storage monitoring module, two storage information are mainly monitored: the first storage information and the second storage information. The above-mentioned first storage information corresponds to the storage information of the distributed storage cluster, and the first storage information mainly refers to the utilization rate of the processor, memory, hard disk, network card and other hardware of each server node in the distributed storage cluster. The above-mentioned second storage information corresponds to the storage information of the virtual machine node in the KVM environment, and the second storage information may include the storage utilization rate of the virtual machine node and its data reading status, etc., which is not strictly limited in this embodiment.

[0111] (II) Data receiving module

[0112] The data receiving module is responsible for receiving the data sent by the storage monitoring module. Its function mainly relies on network communication and distributed message queue technology. This data receiving module needs to set up a dedicated communication port. Its main function is to receive data messages from the storage monitoring module. The communication port acts as a data receiving station, responsible for receiving real-time data information from the storage monitoring module. By establishing this port, the effective transmission of data can be ensured.

[0113] Furthermore, in this embodiment, taking into account the high concurrency of received data, in order to ensure normal data reception and security, a distributed multi-node message queue is established so as to distribute tasks in a distributed processing task queue manner.

[0114] Specifically, see Figure 6 , Figure 6 FIG. 1 is a schematic diagram of a distributed queue process of a storage resource adjustment method provided in an embodiment of the present application. Figure 6As shown, first create a task queue. Create a distributed task queue (usually a list) with the received storage data to store different data information of the received storage monitoring module to be processed: data 1, data 2, data 3, etc. Secondly, push the received different data information to the left side of the distributed task queue in order to add the data distribution tasks corresponding to the data information to the distributed task queue in turn. The second is the registration of working nodes. Each working node (execution unit) needs to register itself in order to coordinate the allocation and processing of data distribution tasks. This is achieved by adding the identification information of the working node (such as name, ID (Identifier, unique identifier) ​​etc.) to a set. The working nodes corresponding to the tasks corresponding to different data information can be the same or different. For example, the working nodes corresponding to the data distribution tasks corresponding to data 1 and data 2 are both working nodes 1, and the working nodes corresponding to the data distribution tasks corresponding to data 3 are working nodes 2. Finally, the execution operation of the data distribution tasks for different data information in the distributed task queue. Once there is a data distribution task to be processed, the data distribution task corresponding to the data information (such as data 1) is popped out from the right side of the distributed task queue, and then the data distribution task is assigned to one or more workers corresponding to the working node. Taking data 1 as an example, the data distribution task corresponding to data 1 is popped out from the right side of the distributed task queue, and then the task is assigned to the worker corresponding to worker node 1. At the same time, this embodiment also uses a distributed lock mechanism to ensure that the task will not be processed by multiple workers at the same time.

[0115] The message queue in the data receiving module can ensure the order and consistency of the request message, and can also relieve the system pressure and improve the availability and stability of the system. In addition, the subsequent data receiving module will also continuously take out data messages from the message queue and perform subsequent processing.

[0116] (III) Analysis module

[0117] This analysis module mainly cleans, denoises, and features the stored data to prepare for the subsequent model prediction. It has the following steps:

[0118] 1. Data cleaning. Ensure that the received data is of high quality by cleaning the data. This step includes identifying and removing invalid, erroneous or duplicate data that may affect the data quality. The cleaning process also includes filling missing values, handling outliers, and formatting to ensure that the data is consistent throughout the entire processing flow. This meticulous cleaning process helps prevent incorrect analysis and inaccurate conclusions caused by dirty data. Here, tools such as pandas can be used to remove data with many missing values ​​and duplicate data; rules can be used to remove duplicate data and irregular and abnormal data to ensure the high quality and reliability of the received data, so as to ensure that the subsequent data set is fully reliable.

[0119] 2. Feature processing. Please refer to Table 1, which is a sample table selected from the storage resource adjustment method provided in an embodiment of the present application, and can be used to display some features:

[0120] Table 1

[0121]

[0122]

[0123] Each row in the sample table shown in Table 1 is a sample, and the sample sorting can be set according to the order of data collection. Each column in the sample table is a different feature dimension of the sample. When performing feature processing, the mean method can be used to fill in the missing values ​​in some important features. For example, the first 5 and last 5 digits of the missing value column of the sample can be added and summed, then divided by 10, and the mean is taken as the filling number of the missing value. In another example, the memory utilization column in Table 1, the third row is vacant, but the first 2 rows and the last 2 rows have values, so the mean method is used to add the first 2 rows and the last 2 rows of the 4 data and divide it by 4, and use it as the filling value of the missing value.

[0124] In addition, considering the business scenarios, network throughput contains redundant information and outliers, which will cause noise in subsequent model training. Therefore, in the analysis module, the PCA (Principal Component Analysis) dimensionality reduction operation is used for the network throughput feature column. First, each sample value of the network throughput item is standardized, for example, the Z-Score standardization method is used to standardize the sample values, and then its covariance matrix and eigenvalues ​​are calculated to achieve dimensionality reduction of the network throughput.

[0125] 3. Feature selection. Useless feature items are deleted, and feature items related to KVM are retained. As shown in Table 1 above, feature items such as network traffic, fluctuation amplitude, number of read and write times and transmission volume of stored data, node memory and hard disk utilization are selected to determine whether the operating status of KVM and nodes is normal.

[0126] (IV) Prediction Module

[0127] The prediction module is mainly used to train the model of the sample data analyzed and processed in the above analysis module, so as to predict the data in the future time period and provide a basis for scientific and dynamic allocation of storage space. It is mainly divided into the following steps:

[0128] 1. Model construction and training. Based on the need to predict the future storage allocation water level, this embodiment integrates business development and uses the time series model Bi-LSTM in deep learning to train the regression model. This is because it is based on a neural network structure and can learn more about data features. Secondly, it is a time series model, based on the learning mode of features at different times such as t_0, t_1, t_2...t_n, and its feasibility in regression tasks. Therefore, the Bi-LSTM time series model is selected to perform the allocation water level prediction task. Figure 7 As shown, Figure 7 It is a schematic diagram of the Bi-LSTM model construction structure of the storage resource adjustment method provided in one embodiment of the present application.

[0129] The multiple feature values ​​(X0, X1, X2 and X3) of the KVM storage samples are converted into sets, normalized, and the values ​​are aligned. These sets are then one-hot encoded (for example, encoded as 001000000... etc.) and converted into vector groups as the input of the model. After feature learning by the neurons of the two layers of LSTM in the forward and backward directions, they are fully connected (FC) to convert the values ​​into a predictable form, and finally the output values ​​are predicted.

[0130] Before training the water level prediction model, the entire sample set can be cross-validated in three folds, specifically by dividing the original data into a training set and a test set, and then training and testing the model multiple times to avoid overfitting and underfitting, so as to confirm the sample set delineation format for the optimal model effect. In this embodiment, the sample set is finally divided in an 8:1:1 format, that is, the ratio of the training, test, and validation data sets is 8:1:1. During the training process, the model is tuned by hyperparameters such as learning rate and step size.

[0131] 2. Model application: After the training of the allocation water level prediction model is completed, the accuracy and response speed of the model effect are verified on the validation set. In order to support the prediction business of the allocation water level, the future allocation water level is predicted. Business support can be designed for short-term predictions (such as changes in the allocation water level in the next few hours or days), medium-term predictions (such as the trend of the allocation water level in the next few weeks or months) and long-term predictions (such as the direction of the allocation water level in the next year or longer). The operation and maintenance management personnel only need to enter the time period, and it can predict the allocation water level value within the time period and give recommended operations. The specific operations are as follows:

[0132] (1) Information input. Based on the latest information collected by the analysis module, eight characteristic attribute values ​​are obtained, such as KVM utilization, IOPS, number of nodes, memory size, hard disk size, CPU utilization, memory utilization, network bandwidth utilization, and storage utilization. In addition, the operation and maintenance personnel need to input the time information corresponding to the future target time period to be predicted. These two parts of information are finally used as the input of the allocation waterline prediction model for the prediction of the allocation waterline.

[0133] (2) Calculation within the model. After inputting KVM utilization, IOPS, number of nodes, memory size, hard disk size, CPU utilization, memory utilization, network bandwidth utilization, storage utilization, and time information, the above allocation watermark prediction model will first perform numerical conversion on these attribute values, such as sub-coding and one-hot coding. Then, it will be multiplied with the W weight matrix in the model (the final parameter of model training) to finally obtain the final value.

[0134] (3) Information output. After the prediction calculation of the allocation water level prediction model, the future storage threshold is calculated, that is, the future allocation water level prediction result. In addition, the allocation water level prediction model can also be combined with a special template package to set corresponding processing suggestions according to the allocation water level. In this way, the output result of the allocation water level prediction model can ultimately include both the future allocation water level prediction result and the operation suggestion.

[0135] (V) Streamlining modules

[0136] The slimming module is mainly to dynamically slim the storage according to the prediction results sent by the prediction module. When the allocation waterline is lower than the current allocation waterline, the slimming operation is started to slim the data to ensure storage availability and improve utilization. At the same time, the KVM management node is scheduled to migrate the KVM virtual machine node to the slimmed-down storage for operation. Figure 8 As shown, Figure 8 This is a schematic diagram of a dynamic streamlining process of a storage resource adjustment method provided in an embodiment of the present application:

[0137] 1. Determine whether the latest allocation water level prediction result of the prediction module is lower than the current allocation water level.

[0138] 2. If it is lower than that, it means that there is a certain amount of waste in the current storage and the utilization rate is low. Streamlining operations are needed to avoid storage capacity overload. Otherwise, no streamlining operations are performed.

[0139] 3. The streamline operation first streamlines the data on the current storage. The streamline method is as follows: Fig. 9 As shown, Fig. 9 This is a schematic diagram of a data dynamic reduction processing method of a storage resource adjustment method provided in an embodiment of the present application:

[0140] In this embodiment, the data dynamic reduction processing method mainly deletes the "0000" data bits in the data block to compress the storage space, and records the address of the reduced database by marking. The address will be cached in the high-speed cache of the distributed system to facilitate the subsequent reading and processing of data. The data structure of the address is as follows: Fig.10 As shown, Fig.10 This is a schematic diagram of the address data structure of a storage resource adjustment method provided in an embodiment of the present application:

[0141] The address includes a tag field (t bits in total), an index field (s bits in total), and a block offset field (b bits in total), where s and b are positive integers.

[0142] Tag: Mainly records the ID of the stored file, which serves as the globally unique ID of the file in the storage.

[0143] Block index: records the location index of the streamlined data blocks in the file, making it easy to quickly find the streamlined data blocks.

[0144] Bit offset: records the offset of the reduced data bits in the reduced data block to facilitate data reading.

[0145] 4. After data is streamlined, it will be stored in the new storage space.

[0146] 5. Call the KVM management node to automatically migrate the KVM virtual machine to the streamlined new storage, and delete the original storage space after the migration is completed. Fig.11 As shown, 11 is a schematic diagram of the KVM storage migration architecture of the storage resource adjustment method provided in an embodiment of the present application.

[0147] In this embodiment, even if data has been streamlined and compressed, some unnecessary data or redundant data may still be accumulated over time and during the operation of the virtual machine. Therefore, data is migrated to a new streamlined storage, and further data cleaning and compression are performed during the process, which can further optimize the utilization of storage resources. In addition, by migrating the virtual machine to the new streamlined storage, resource integration can be achieved, so that storage resources can be used more efficiently.

[0148] This embodiment uses data feature processing and applies feature engineering operations in machine learning to retain important feature information in KVM storage information samples to provide information basis for scientific decision-making. In addition, this embodiment uses model prediction, based on historical sample sets, and uses a deep learning model bi-lstm time series model for regression training to predict future storage allocation water level values ​​for short-term, medium-term, and long-term time periods. Compared with the traditional solution of allocating water levels based on manual experience, the scientific nature of the prediction is greatly improved. In addition, this embodiment combines the future water level prediction results to realize dynamic adjustment of storage space in the KVM environment, improves the utilization rate of storage resources, and reduces storage costs, thereby ensuring the availability and efficiency of the system, and has important practical value and social benefits.

[0149] Based on the storage resource adjustment method provided in the above embodiment, out of the same inventive concept, the present application also provides a storage resource adjustment device corresponding to the above storage resource adjustment method. Fig.12 A storage resource adjustment device is introduced in detail.

[0150] Fig.12 A schematic diagram of the structure of a storage resource adjustment device provided in an embodiment of the present application is shown.

[0151] Fig.12 The storage resource adjustment device 1200 shown includes:

[0152] A first acquisition module 1210 is used to acquire first storage information of a distributed storage cluster and second storage information of a virtual machine node at a current moment, where the distributed storage cluster is used to provide storage resources for the virtual machine node;

[0153] A first processing module 1220 is used to process and obtain N feature items based on the first storage information and the second storage information, where N is a positive integer; the N feature items are related to a kernel-based virtual machine KVM;

[0154] The first prediction module 1230 is used to input the target input information into the pre-trained distribution water level prediction model to obtain the distribution water level prediction result corresponding to the future target time period to be predicted, and the target input information is obtained based on N feature items and time information corresponding to the future target time period;

[0155] The first slimming module 1240 is configured to perform a slimming operation on the storage data in the virtual machine node and / or migrate the virtual machine node when the allocation waterline prediction result is lower than the current allocation waterline.

[0156] A storage resource adjustment device provided in an embodiment of the present application, by setting corresponding functional modules, obtains the first storage information of the distributed storage cluster and the second storage information of the virtual machine node at the current moment, and processes and obtains N feature items related to KVM based on the first storage information and the second storage information. In this way, the target input information obtained by processing the time information corresponding to the future target time period based on the N feature items and the future target time period is input into the pre-trained allocation water level prediction model to obtain the allocation water level prediction result corresponding to the future target time period to be predicted. Finally, when the allocation water level prediction result is lower than the current allocation water level, the storage data in the virtual machine node is streamlined, and / or the virtual machine node is migrated.

[0157] From the above description, it can be seen that a storage resource adjustment device of an embodiment of the present application obtains the first storage information of the distributed storage cluster and the second storage information of the virtual machine node, and performs data feature processing on the above information to retain the feature information related to KVM, and then makes an intelligent and accurate prediction of the allocation waterline of the future target time period based on the feature information and the pre-trained allocation waterline prediction model. In this way, the data can be dynamically streamlined in a timely manner according to the current storage usage and the KVM resources can be reallocated based on the migration method, so that the reasonable storage and management of the data in the virtual machine can be more fully and effectively realized, thereby achieving the purpose of accurately reducing the utilization rate of storage resources.

[0158] Based on the storage resource adjustment method provided in the above embodiment, out of the same inventive concept, the present application also provides a storage resource adjustment device corresponding to the above storage resource adjustment method. Fig.13 Provides a detailed introduction to storage resource adjustment devices.

[0159] See below Fig.13 , Fig.13 It is a structural diagram of a storage resource adjustment device provided in one embodiment of the present application.

[0160] The storage resource adjustment device may include a processor 1301 and a memory 1302 storing computer program instructions.

[0161] Specifically, the processor 1301 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0162] The memory 1302 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 1302 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, the memory 1302 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 1302 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 1302 is a non-volatile solid-state memory.

[0163] The memory may include read-only memory (ROM), random access memory (RAM), magnetic disk storage media devices, optical storage media devices, flash memory devices, electrical, optical or other physical / tangible memory storage devices. Thus, typically, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to an aspect of the present disclosure.

[0164] The processor 1301 implements any one of the storage resource adjustment methods in the above embodiments by reading and executing computer program instructions stored in the memory 1302 .

[0165] In one example, the storage resource adjustment device may further include a communication interface 1303 and a bus 1310. Fig.13 As shown, the processor 1301, the memory 1302, and the communication interface 1303 are connected via a bus 1310 and communicate with each other.

[0166] The communication interface 1303 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0167] Bus 1310 includes hardware, software or both, and the parts of storage resource adjustment equipment are coupled to each other. For example, but not limitation, bus may include accelerated graphics port (AGP) or other graphics bus, enhanced industrial standard architecture (EISA) bus, front side bus (FSB), hypertransport (HT) interconnection, industrial standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In suitable cases, bus 1310 may include one or more buses. Although the present application embodiment describes and shows a specific bus, the application considers any suitable bus or interconnection.

[0168] The storage resource adjustment device executes the storage resource adjustment method in the embodiment of the present application, thereby realizing the storage resource adjustment method described in the embodiment of the present application.

[0169] In addition, in combination with the storage resource adjustment method in the above embodiment, the embodiment of the present application can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when the computer program instructions are executed by a processor, any one of the storage resource adjustment methods in the above embodiment is implemented.

[0170] Based on the storage resource adjustment method in the above embodiment, an embodiment of the present application provides a computer program product. When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the storage resource adjustment method provided in any one of the above embodiments of the present application.

[0171] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.

[0172] The functional blocks shown in the above-described block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present application are programs or code segments that are used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0173] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be performed simultaneously.

[0174] Aspects of the present disclosure are described above with reference to the flowchart and / or block diagram of the method, device (system) and computer program product according to the embodiment of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field programmable logic circuit. It can also be understood that each box in the block diagram and / or flowchart and the combination of boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs a specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0175] The above is only a specific implementation of the present application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the protection scope of the present application is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should be included in the protection scope of this application.

Claims

1. A storage resource adjustment method, characterized in that: The method comprises: Acquire first storage information of a distributed storage cluster and second storage information of a virtual machine node at a current moment, wherein the distributed storage cluster is used to provide storage resources for the virtual machine node; Based on the first storage information and the second storage information, N feature items are obtained by processing, where N is a positive integer; the N feature items are related to a kernel-based virtual machine KVM; Inputting target input information into a pre-trained allocation water level prediction model to obtain an allocation water level prediction result corresponding to a future target time period to be predicted, wherein the target input information is obtained based on the N feature items and time information corresponding to the future target time period; When the allocation waterline prediction result is lower than the current allocation waterline, a slimming operation is performed on the storage data in the virtual machine node, and / or the virtual machine node is migrated.

2. The method according to claim 1, characterized in that Before inputting the target input information into the pre-trained allocation water level prediction model to obtain the allocation water level prediction result corresponding to the future target time period to be predicted, the method further includes: Acquire first historical storage information of the distributed storage cluster and second historical storage information of the virtual machine node at multiple historical moments; Based on the first historical storage information and the second storage information, generate a plurality of data samples corresponding to the plurality of historical moments one by one; Based on the multiple data samples, a bidirectional long short-term memory network Bi-LSTM time series model is used for regression training to obtain the distribution water level prediction model.

3. The method according to claim 1, characterized in that The N feature items include: at least one of KVM utilization, the number of read and write operations per second (IOPS) of stored data, the number of nodes, the memory size, the hard disk size, the CPU utilization, the memory utilization, the network bandwidth utilization, and the storage utilization.

4. The method according to any one of claims 1 to 3, characterized in that: The performing a thinning operation on the storage data in the virtual machine node includes: Deleting invalid data bits in the storage data block under the virtual machine node to generate a streamlined storage data block; After performing a thinning operation on the storage data in the virtual machine node, the method further includes: The address information of the streamlined storage data block is cached in the high-speed cache of the distributed storage cluster for data reading processing.

5. The method according to claim 4, characterized in that After deleting the invalid data bits in the storage data block under the virtual machine node to generate a streamlined storage data block, migrating the virtual machine node includes: Storing the streamlined storage data blocks in a new storage space; The virtual machine node is migrated to the new storage space, and the original storage space is deleted after the migration is completed.

6. The method according to any one of claims 1 to 3, characterized in that: Before inputting the target input information into the pre-trained allocation water level prediction model to obtain the allocation water level prediction result corresponding to the future target time period to be predicted, the method further includes: Converting the attribute values ​​corresponding to the N feature items and the time information into a first attribute value set, and performing normalization and data alignment processing on the attribute values ​​in the attribute value set to obtain a second attribute value set; The second attribute value set is converted into a target vector group by performing one-hot encoding, and the target vector group is determined as the target input information.

7. A storage resource adjustment device, characterized in that: The device comprises: A first acquisition module, used to acquire first storage information of a distributed storage cluster and second storage information of a virtual machine node at a current moment, wherein the distributed storage cluster is used to provide storage resources for the virtual machine node; A first processing module, configured to obtain N feature items based on the first storage information and the second storage information, where N is a positive integer; the N feature items are related to a kernel-based virtual machine KVM; A first prediction module is used to input target input information into a pre-trained distribution water level prediction model to obtain a distribution water level prediction result corresponding to a future target time period to be predicted, wherein the target input information is obtained based on the N feature items and time information corresponding to the future target time period; The first streamlining module is configured to perform a streamlining operation on the storage data in the virtual machine node and / or migrate the virtual machine node when the allocation waterline prediction result is lower than the current allocation waterline.

8. A storage resource adjustment device, characterized in that: The device comprises: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, the storage resource adjustment method according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed by a processor, the storage resource adjustment method according to any one of claims 1 to 6 is implemented.

10. A computer program product, characterized in that When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the storage resource adjustment method according to any one of claims 1 to 6.

Images