Disaster recovery backup method and device, communication device and storage medium

Abstract

The application relates to a disaster recovery backup method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring network topology dataset of a data center where a current node is located; the network topology dataset comprises node state data of each data node and link state data of each network link; constructing a state space and an action space of the current node based on the node state data and the link state data, and determining a target candidate node in the data center based on a data center network model, the state space and the action space; the target candidate node is a path node to a target backup node; and transmitting backup data from the current node to the target candidate node until the target backup node. The method can improve the efficiency of the disaster recovery backup method.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of network technology and security technology, and in particular to a disaster recovery backup method and device, communication equipment, a storage medium and a computer program product. BACKGROUND

[0002] A data center comprises data nodes, and network links for transmitting data are arranged between adjacent data nodes. The data nodes include a data center node and a backup node. In order to cope with sudden disasters, the data to be backed up of the data center node is transmitted to the corresponding target backup node, so that the data center node can continue to work based on the data in the target backup node when a disaster occurs.

[0003] The current disaster recovery backup method finds the next data node in the routing table in the process of transmitting the data to be backed up from the data center node to the target backup node. Alternatively, the data center administrator sets a fixed backup link, and the fixed backup link corresponding to each network link is pre-set with a fixed backup data bandwidth. The data nodes on the fixed backup link transmit the data to be backed up based on the fixed backup data bandwidth.

[0004] However, when the routing table or the fixed backup link is used to determine the data nodes in the intermediate transmission process in the current disaster recovery backup method, since the network link is also used to transmit application data, it will cause network congestion, resulting in slow transmission efficiency of the data to be backed up, and thus reducing the efficiency of the disaster recovery backup method. SUMMARY

[0005] Therefore, it is necessary to provide a disaster recovery backup method, device, computer equipment, computer readable storage medium and computer program product in view of the above technical problems.

[0006] In a first aspect, the present application provides a disaster recovery backup method, comprising:

[0007] obtaining a network topology data set of a data center where a current node is located; the network topology data set comprises node state data of each data node and link state data of each network link;

[0008] constructing a state space and an action space of the current node based on each node state data and each link state data, and determining a target candidate node in the data center based on a data center network model, the state space and the action space; the target candidate node is a node on the way to the target backup node;

[0009] transmitting the data to be backed up from the current node to the target candidate node until the target backup node.

[0010] In one of the embodiments, the constructing the state space and the action space of the current node based on the node state data and the link state data comprises:

[0011] determining the total state data set of the data center based on the node state data and the link state data;

[0012] constructing the state space of the current node based on the total state data set, the network link and the preset data average proportion;

[0013] screening the action space of the current node in the data center according to the state space, the preset data amount difference parameter and the network link normal service proportion.

[0014] In one of the embodiments, the data node comprises the data center node and the backup node, the link state data comprises the usage state and the transmission bandwidth of the network link, the node state data comprises the source data amount and the storage capacity of the data node, and the determining the total state data set of the data center based on the node state data and the link state data comprises:

[0015] determining the residual storage capacity of the data center according to the storage capacity and the source data amount of the backup node;

[0016] determining the backup residual data amount of the data center according to the source data amount of the data center node and the source data amount of the backup node;

[0017] determining the total residual bandwidth of the data center based on the transmission bandwidth of the network link in the idle state;

[0018] constructing the total state data set of the data center based on the total residual bandwidth, the backup residual data amount and the residual storage capacity of the data center.

[0019] In one of the embodiments, the screening the action space of the current node in the data center according to the state space, the preset data amount difference parameter and the network link normal service proportion comprises:

[0020] constructing the constraint condition set according to the state space, the preset data amount difference parameter and the network link normal service proportion;

[0021] screening the next hop node of the current node in the data node of the data center based on the constraint condition set to obtain the candidate node set, and determining the candidate node set as the action space of the current node.

[0022] In one of the embodiments, the determining of the target candidate node in the data center based on the data center network model, the state space and the action space comprises:

[0023] The state space and the action space are predicted based on the data center network model to obtain an initial reward value;

[0024] The initial reward value is optimized according to a heuristic optimization algorithm to obtain a reward value;

[0025] The target candidate node is determined in the action space according to the reward value and a probability algorithm.

[0026] In one of the embodiments, the determining of the target candidate node in the action space according to the reward value and the probability algorithm comprises:

[0027] For each of the candidate nodes, a candidate link between the candidate node and the current node is determined;

[0028] The reward value, the node state data of the candidate node and the link state data of the candidate link are subjected to a probability operation according to a probability algorithm to obtain a probability of the candidate node;

[0029] The candidate node with the highest probability is determined as the target candidate node from the candidate nodes.

[0030] In a second aspect, the present application further provides a disaster recovery backup device, comprising:

[0031] An acquisition module is configured to acquire a network topology dataset of a data center where a current node is located; the network topology dataset comprises node state data of each data node and link state data of each network link;

[0032] A determination module is configured to construct a state space and an action space of the current node based on the node state data and the link state data, and to determine a target candidate node in the data center based on a data center network model, the state space and the action space; the target candidate node is a way node to a target backup node;

[0033] A transmission module is configured to transmit backup data from the current node to the target candidate node until the target backup node.

[0034] In a third aspect, the present application further provides a computer device, comprising a memory and a processor; the memory stores a computer program; and the processor implements the following steps when executing the computer program:

[0035] obtaining a network topology dataset of a data center where a current node is located; the network topology dataset comprises node state data of each data node and link state data of each network link;

[0036] constructing a state space and an action space of the current node based on each of the node state data and each of the link state data, and determining a target candidate node in the data center based on a data center network model, the state space and the action space; the target candidate node is an en route node to a target backup node;

[0037] transmitting the data to be backed up from the current node to the target candidate node until the target backup node.

[0038] In a fourth aspect, the present application further provides a computer readable storage medium, having a computer program stored thereon, the computer program being executed by a processor to implement the following steps:

[0039] obtaining a network topology dataset of a data center where a current node is located; the network topology dataset comprises node state data of each data node and link state data of each network link;

[0040] constructing a state space and an action space of the current node based on each of the node state data and each of the link state data, and determining a target candidate node in the data center based on a data center network model, the state space and the action space; the target candidate node is an en route node to a target backup node;

[0041] transmitting the data to be backed up from the current node to the target candidate node until the target backup node.

[0042] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, the computer program being executed by a processor to implement the following steps:

[0043] obtaining a network topology dataset of a data center where a current node is located; the network topology dataset comprises node state data of each data node and link state data of each network link;

[0044] constructing a state space and an action space of the current node based on each of the node state data and each of the link state data, and determining a target candidate node in the data center based on a data center network model, the state space and the action space; the target candidate node is an en route node to a target backup node;

[0045] transmitting the data to be backed up from the current node to the target candidate node until the target backup node.

[0046] The disaster recovery backup method, device, communication equipment, storage medium and computer program product obtain network topology data set of a data center where a current node is located; the network topology data set contains node state data of each data node and link state data of each network link; a state space and an action space of the current node are constructed based on each node state data and each link state data, and a target candidate node is determined in the data center based on a data center network model, the state space and the action space; the target candidate node is a way node to a target backup node; backup data is transmitted from the current node to the target candidate node until the target backup node. By using the method, a network link with an idle link state and a data node with a good node state are selected as the next node of the current node, i.e., the target candidate node, in each data node in the data center based on each node state data, link state data and the data center network model. The backup data to be transmitted is transmitted from the current node to the target candidate node with a good state through the idle network link, the transmission efficiency of the backup data is improved, and the efficiency of the disaster recovery backup method is improved. BRIEF DESCRIPTION OF DRAWINGS

[0047] Figure 1 An application environment diagram of a disaster recovery backup method in an embodiment;

[0048] Figure 2 A flowchart of a disaster recovery backup method in an embodiment;

[0049] Figure 3 A flowchart of constructing a state space and an action space in an embodiment;

[0050] Figure 4 A flowchart of determining a total state data set in an embodiment;

[0051] Figure 5 A flowchart of screening an action space in an embodiment;

[0052] Figure 6 A flowchart of determining a target candidate node in an embodiment;

[0053] Figure 7 An architecture diagram of a data center network model in an embodiment;

[0054] Figure 8 A flowchart of determining a target candidate node based on probability in an embodiment;

[0055] Figure 9 A flowchart of training a reinforcement learning model in an embodiment;

[0056] Figure 10Here is a network topology diagram of a sample network topology model in one embodiment;

[0057] Figure 11 This is a structural block diagram of a disaster recovery backup device in one embodiment;

[0058] Figure 12 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0060] Figure 1 This is a schematic diagram illustrating an application scenario of a disaster recovery backup method provided in an embodiment of this application. For example... Figure 1 As shown, the disaster recovery backup method is applied to data center 100. Data center 100 includes data nodes 110 and terminals 120. Network links for data transmission are set between adjacent data nodes 110. Each network link is configured with a maximum fixed bandwidth. Each data node 110 is divided into three types of data nodes: data center node 111, backup node 112, and intermediate node 113. Data center node 111 is a data node with backup requirements and provides application services. Backup node 112 is a data node that stores backup data. The remaining data nodes are intermediate nodes 113. Terminal 120 is used to execute the disaster recovery backup method, transmitting the backup data to be backed up from data center node 111 to the target backup node. Terminal 120 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, or servers. Servers can be independent physical servers, server clusters or distributed systems composed of multiple physical servers, or cloud servers providing cloud computing services.

[0061] Traditional disaster recovery backup methods employ two backup approaches. The first approach involves identifying the data center node requiring backup as the current node and retrieving its corresponding routing table. This routing table contains the addresses of each next-hop node (alternate node) for the current node. Then, among the alternate nodes, the one with the highest routing priority is selected as the target alternate node, and the data to be backed up is transferred from the current node to the target alternate node. The target alternate node is then updated to the current node, and it is determined whether the current node is indeed the target backup node. If the current node is not the target backup node, the process of retrieving its routing table is repeated until the current node is identified as the target backup node.

[0062] The second backup mode is to determine a data center node with a backup requirement as a starting point and a target backup node as an ending point. Then, a fixed backup link corresponding to the starting point and the ending point is determined, and the data to be backed up is transmitted from the data center node to the target backup node based on the fixed backup link. Since a fixed backup data bandwidth is set in each network link in the data center, the data nodes on the fixed backup link transmit the data to be backed up based on the backup data bandwidth.

[0063] In the conventional disaster backup method, when the data nodes in the intermediate transmission process are determined by using the routing table or the fixed backup link, the network link is also used to transmit application data, which causes congestion of the network link, reduces the transmission efficiency of the data to be backed up, and further reduces the efficiency of the disaster backup method. Moreover, the remaining available data nodes and network links are in an idle state, which reduces the resource utilization of the data center.

[0064] Based on the above conventional technology, the embodiments of the present application provide a disaster backup method. Based on the node state data, the link state data and the data center network model, a network link with an idle state and a data node with a good state are selected as the next node of the current node, i.e., the target candidate node. The data to be backed up is transmitted from the current node to the target candidate node through the idle network link, which improves the transmission efficiency of the data to be backed up, and further improves the efficiency of the disaster backup method. Moreover, the data to be backed up is transmitted through the idle network link and the data node, which improves the resource utilization of the data center.

[0065] It should be noted that the beneficial effects or technical problems solved by the embodiments of the present application are not limited to one, but also other implicit or related problems. For details, please refer to the description of the following embodiments.

[0066] The technical solutions of the present application and how the technical solutions solve the above technical problems will be described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments. The embodiments of the present application will be described below with reference to the accompanying drawings.

[0067] In one embodiment, as shown in Figure 2 , a disaster backup method is provided. The method is applied to the terminal 120 (hereinafter referred to as terminal) in Figure 1 for example, and includes the following steps:

[0068] In step 202, the network topology data set of the data center where the current node is located is obtained.

[0069] The network topology dataset includes node state data of each data node and link state data of each network link.

[0070] In implementation, the terminal determines the data center where the current node is located, and obtains the network topology graph of the data center, the node state data of each data node and the link state data of each network link, to obtain the network topology dataset of the data center.

[0071] Specifically, when a data center node in the data center needs to transmit the data to be backed up to the target backup node, the terminal determines the data center node as the current node. Then, the terminal obtains the node state data of each data node and the link state data of each network link in the data center where the current node is located, to obtain the network topology dataset of the data center.

[0072] In an exemplary embodiment, the terminal obtains the source data amount and the storage capacity of each data node in the data center, and constructs the node state data of the data node according to the source data amount and the storage capacity. The storage capacity is the data amount that can be stored by the data node, and the source data amount is the data amount of the data stored in the data node. Meanwhile, the terminal obtains the usage state and the transmission bandwidth of each network link, and constructs the link state data of the network link according to the usage state and the transmission bandwidth. The usage state represents whether the network link is in data transmission, i.e., whether the network link is in use. The transmission bandwidth is the maximum bandwidth of the network link. The terminal constructs the network topology dataset of the data center according to the node state data of each data node and the link state data of each network link.

[0073] For example, the terminal obtains the network topology graph of the data center where the current node is located . V represents each data node in the data center, and E represents the connection relationship between the data nodes, i.e., the network link. represents the data center node with backup requirement, represents the number of data center nodes in the data center. represents the backup node in the data center. represents the number of backup nodes in the data center. The terminal obtains the source data amount and the storage capacity of each data node in the data center, to obtain the node state data of each data node. The source data amount of the data center node , the storage capacity of the backup node . Meanwhile, the terminal obtains the transmission bandwidth and the usage state of each network link in the data center, to obtain the link state data of each network link. The transmission bandwidth of the network link is . represents the data center node to the backup node . The first The set P is a set of all backup transmission links. Each network link is denoted by . Then denotes the total bandwidth used by all transmission paths on the link , where the binary variable is used to indicate whether the backup link is used or not, and also indicates the usage state of the network link.

[0074] Optionally, the node state data further includes the inflow and outflow of the data node, and the link state data further includes the residual bandwidth of the network link, and the embodiments of the present application do not limit the data contained in the node state data and the link state data.

[0075] In step 204, the state space of the current node and the action space of the current node are constructed based on the node state data and the link state data, and the target candidate node is determined in the data center based on the data center network model, the state space and the action space.

[0076] The target candidate node is a path node to the target backup node. The target backup node is the end point of the transmission of the data to be backed up.

[0077] In implementation, the terminal constructs the state space of the current node and the action space of the current node based on the node state data and the link state data, with the goal of maximizing the transmission flow. Then, the terminal screens the target candidate node in each data node of the data center based on the data center network model, the state space and the action space.

[0078] In step 206, the data to be backed up is transmitted from the current node to the target candidate node until the target backup node.

[0079] In implementation, the terminal transmits the backup data from the current node to the target candidate node, and updates the target candidate node as the current node. Then, the terminal judges whether the current node is the target backup node. If the current node is not the target backup node, the terminal performs the step of obtaining the network topology data set of the data center where the current node is located. If the current node is the target backup node, the terminal stops executing the disaster recovery backup method. At this time, the target backup node receives the data to be backed up and saves the data to be backed up, thereby realizing the disaster recovery backup of the data.

[0080] In the disaster backup method, based on the node state data, the link state data and the data center network model, a network link with an idle link state and a data node with a good node state are selected as the next node of the current node, i.e., the target candidate node. The to-be-backup data is transmitted from the current node to the target candidate node through the idle network link, thereby improving the transmission efficiency of the to-be-backup data, and further improving the efficiency of the disaster backup method.

[0081] In one exemplary embodiment, as shown in FIG. 2, the specific process of constructing the state space and the action space of the current node based on the node state data and the link state data in step 202 includes steps 302 to 306. Wherein: Figure 3

[0082] Step 302, determining the total state data set of the data center based on the node state data and the link state data.

[0083] Wherein, the node state data of each data node includes the storage capacity and the source data amount of the data node.

[0084] In implementation, the terminal determines the backup residual data amount and the residual storage capacity of the data center based on the storage capacity and the source data amount. Then, the terminal determines the total residual bandwidth of the data center based on the link state data. The terminal constructs the total state data set of the data center based on the total residual bandwidth, the backup residual data amount and the residual storage capacity.

[0085] Specifically, the terminal determines the backup residual data amount of the data center based on the storage capacity and the source data amount of each backup node. Then, the terminal determines the backup residual data amount of the data center according to the source data amount of the data center node and the source data amount of the backup node. Then, the terminal determines the total residual bandwidth of the data center based on the usage state and the transmission bandwidth of each network link in the link state data. The terminal combines the total residual bandwidth, the backup residual data amount and the residual storage capacity to obtain the total state data set of the data center. Through the total residual bandwidth, the backup residual data amount and the residual storage capacity, the node state of each data node and the link state of the network link can be determined.

[0086] Step 304, constructing the state space of the current node based on the total state data set, each network link and the preset data average ratio.

[0087] In implementation, the terminal is pre-provided with the data average ratio. The data average ratio is the average ratio of the data amount stored in the target backup node to the total backup data amount. The terminal constructs the state space of the current node based on the total state data set, each network link and the preset data average ratio, with the goal of maximizing the transmission data.

[0088] ​In one exemplary embodiment, the total state dataset includes the total remaining bandwidth. Backup remaining data volume and remaining storage capacity The average proportion of data preset in the terminal is... The terminal identifies the data center node from which data needs to be transmitted as the target data center node and determines the backup transmission paths from the target data center node to the target backup node. Represents data center node To the backup node The There are backup transmission paths. Set P is the set of all backup transmission links. The bandwidth of each network link is represented by... This indicates. Then... This indicates all backup transmission paths. In the link The total bandwidth used. Among them, binary variables. Used to indicate whether a backup link is used. The total remaining bandwidth is So, network links The corresponding maximum flow is shown in the following formula (1):

[0089]

[0090] In the above formula (1), Indicates from data center node To backup node Traffic on the network link. For data center nodes To backup node The minimum network bandwidth required during the process.

[0091] Therefore, in order to minimize backup time, the objective function throughout the backup process is to maximize this maximum flow. The objective function is shown in formula (2) below:

[0092]

[0093] In the above formula (2), Indicates from data center node To backup node Traffic on the network link. For data center nodes To backup node The sum of the bandwidths of each network link on the backup transmission link.

[0094] The terminal constructs the aforementioned network links into a network link set. The terminal constructs a state space of the current node according to the network link set, the total residual bandwidth , the backup residual data volume , the residual storage capacity , and the data average proportion . .

[0095] In step 306, an action space of the current node is screened in the data center according to the state space, the preset data volume difference parameter, and the network link normal service proportion.

[0096] In implementation, the terminal constructs a screening condition set according to the state space, the preset data volume difference parameter, and the network link normal service proportion, and screens each candidate node in each data node of the data center based on the screening condition set to obtain the action space of the current node.

[0097] In this embodiment, the total state data set is determined through the node state data and the link state data, and the state of the data center is determined. The state space is constructed through the total state data set, and the next hop node of the current node is screened based on the state space to obtain the action space, and the next hop node of the current node is preliminarily determined, which facilitates subsequent determination of the target candidate node in the action space. Moreover, the state space is constructed in the form of transmission flow, which improves the transmission efficiency of the data to be backed up and improves the efficiency of the disaster recovery backup method.

[0098] In one exemplary embodiment, as shown in Figure 4 , each data node includes each data center node and each backup node; the link state data includes the use state and the transmission bandwidth of the network link, and the node state data includes the source data volume and the storage capacity of the data node, and the specific processing process of step 302 includes steps 402 to 408. Wherein:

[0099] In step 402, the residual storage capacity of the data center is determined according to the storage capacity and the source data volume of each backup node.

[0100] In implementation, the terminal performs difference processing on the storage capacity and the source data volume of each backup node to obtain the residual storage capacity of the backup node. Then, the terminal performs addition processing on the residual storage capacity of each backup node to obtain the residual storage capacity of the data center.

[0101] Specifically, the terminal screens each backup node in each data node. Then, the terminal performs difference processing on the storage capacity and the source data volume of each backup node to obtain the residual storage capacity of the backup node. Then, the terminal performs addition operation on the residual storage capacity of each backup node to obtain the residual storage capacity of the data center. The residual storage capacity of the data center represents the data volume of the backup data that can be stored in the data center.

[0102] Step 404, determining the backup remaining data amount of the data center according to the source data amount of each data center node and the source data amount of each backup node.

[0103] In implementation, the terminal performs addition processing on the source data amount of each data center node to obtain the total backup data amount of the data center, and performs addition processing on the source data amount of each backup node to obtain the backed-up data amount. Then, the terminal performs difference processing on the total backup data amount and the backed-up data amount to obtain the backup remaining data amount of the data center.

[0104] Step 406, determining the total remaining bandwidth of the data center based on the transmission bandwidth of each network link in the idle state.

[0105] In implementation, the terminal screens the idle network links in the idle state from the network links, and obtains the total remaining bandwidth of the data center according to the addition operation on the states of the idle network links.

[0106] Specifically, the usage state of the network link includes the idle state and the non-idle state. The idle state represents that the network link is in the idle state and is not in the process of transmitting data. The non-idle state represents that the network link is in the process of transmitting data. The terminal determines whether the usage state of each network link is in the idle state. If the usage state of the network link is in the idle state, the terminal determines that the network link is an idle network link. Then, the terminal performs addition operation on the states of the idle network links to obtain the total remaining bandwidth of the data center.

[0107] In an exemplary embodiment, the transmission bandwidth of each network link is represented by . Wherein, , V represents each data node. represents the first backup transmission path from the data center node to the backup node , and the set P is the set of all backup transmission links. Then , represents the total bandwidth used on the network link by all backup transmission paths. Wherein, the binary variable is used to indicate whether the backup transmission link is used. If the backup transmission link is used, all network links on the backup transmission link are in the non-idle state. If the backup transmission link is not used, each network link on the backup transmission link is in the idle state. Then the total remaining bandwidth is .

[0108] ​​​At step 408, the terminal constructs a total state dataset of the data center based on the total residual bandwidth of the data center, the residual data volume of the backup and the residual storage capacity.

[0109] In implementation, the terminal combines the total residual bandwidth of the data center, the residual data volume of the backup and the residual storage capacity to construct the total state dataset of the data center.

[0110] In the embodiment, the total residual bandwidth of the data center, the residual data volume of the backup and the residual storage capacity are determined by the node state data and the link state data, which explicitly shows the overall state of the data center and facilitates the subsequent construction of the state space and the action space.

[0111] In one exemplary embodiment, as shown in Figure 5 the specific processing procedure of step 306 includes steps 502 to 504. In which:

[0112] At step 502, the terminal constructs a constraint condition set according to the state space, the preset data volume difference parameter and the network link normal service ratio.

[0113] In implementation, the terminal is pre-configured with the data volume difference parameter. The data volume difference parameter is a threshold value of the difference between the data volumes received by each backup node. The network link normal ratio is the bandwidth ratio used for transmitting normal service data. The terminal constructs a first screening condition with the target of non-negative transmission flow. Then, the terminal constructs a second screening condition with the target of transmission bandwidth between two nodes not being greater than each flow, and a third screening condition with the target of non-negative bandwidth allocation. The terminal constructs a fourth screening condition based on the network link normal service ratio, and a fifth screening condition based on the data volume difference parameter and the data average ratio in the state space. The terminal constructs a sixth screening condition according to the data receiving volume and the storage capacity of each backup node, and a seventh screening condition according to the input flow and the output flow of each data node. The terminal constructs the constraint condition set according to the above screening conditions.

[0114] Specifically, the terminal constructs a first constraint condition set with non-negative terminal transmission traffic, and constructs a second screening condition based on the target that the transmission traffic between each network link cannot be greater than the transmission bandwidth of the network link. Then, the terminal constructs a third screening condition with the target that the bandwidth allocation is non-negative. The terminal constructs a fourth screening condition with the target that the proportion of the bandwidth of the backup data transmitted on each network link and the proportion of the normal traffic of the network link is less than or equal to 1. Then, the terminal constructs a fifth screening condition with the target that the difference between the received data amount of each backup node is less than the data amount difference parameter. The terminal constructs a sixth screening condition with the target that the data receiving amount of each backup node is less than or equal to the storage capacity of the data node. At the same time, the terminal constructs a seventh screening condition with the target that the input traffic of each data node is equal to the output traffic. The terminal constructs a constraint condition set according to the first screening condition to the seventh screening condition. The constraint condition set is shown in the following formula group (3):

[0115] (3)

[0116] In the above formula group (3), indicates a constraint condition. The constraint condition set includes seven constraint conditions, which are a first constraint condition (3.1), a second constraint condition (3.2), a third constraint condition (3.3), a fourth constraint condition (3.4), a fifth constraint condition (3.5), a sixth constraint condition (3.6), and a seventh constraint condition (3.7). In the above formula (3.1), indicates the traffic transmitted by the nth network link. In the above formula (3.2), indicates the traffic value transmitted by the network link, indicates the transmission bandwidth corresponding to the network link. In the above formula (3.3), is the transmission bandwidth allocated to the network link. In the above formula (3.4), indicates the backup link bandwidth used for transmitting the backup data on the network link. indicates the transmission bandwidth of the network link, is the normal traffic proportion of the network link. In the above formula (3.5), indicates the average proportion of data in the state space. indicates each backup node in the data center. indicates the backup node. indicates each data center node in the data center, indicates the data center node. indicates the backup transmission link from the data center node to the backup node . indicates the backup transmission link from the data center node to the backup node transmission flow. is a backup node of the data center. is a data volume difference parameter. In the above formula (3.6), represents each backup node in the data center. represents a backup node. represents each data center node in the data center, represents a data center node. represents a backup transmission link from a data center node to a backup node . represents a transmission flow from a data center node to a backup node . In the above formula (3.7), is a storage capacity of a backup node . In the above formula (3.7), represents an input flow from a data node to a data node in the data center, represents an output flow from a data node to a data node in the data center. If the input flow is greater than the output flow, the data node is a backup node. If the output flow is greater than the input flow, the data node is a data center node. If the output flow is equal to the input flow, the data node is an intermediate node. Step 504, based on the constraint condition set, filtering the next hop node of the current node in each data node of the data center to obtain a candidate node set, and determining the candidate node set as the action space of the current node.

[0117] In implementation, the terminal filters the next hop node existing network link with the current node in each data node of the data center to obtain an initial candidate node set. The terminal filters each candidate node meeting the constraint condition set in the initial candidate node set to obtain a candidate node set. Then, the terminal determines the candidate node set as the action space of the current node.

[0118] In implementation, the terminal filters the next hop node existing network link with the current node in each data node of the data center to obtain an initial candidate node set. The terminal filters each candidate node meeting the constraint condition set in the initial candidate node set to obtain a candidate node set. Then, the terminal determines the candidate node set as the action space of the current node.

[0119] ​Specifically, based on the network topology map of the data nodes, the terminal filters the next-hop nodes in the data center that have a network link with the current node, obtaining an initial set of candidate nodes. Then, for each initial candidate node in the initial candidate node set, the terminal determines the initial candidate network link between the initial candidate node and the current node. Based on the node state data of the initial candidate nodes and the link state data of the initial candidate network links, the terminal determines whether the initial candidate nodes satisfy the constraint set. If the initial candidate node satisfies the constraint set, the terminal determines that initial candidate node as a candidate node. The terminal constructs a candidate node set based on each candidate node.

[0120] In this embodiment, a set of constraints is constructed by limiting capacity, bandwidth, and load balancing. Based on the set of constraints, the next-hop node of the current node is selected in each data node, and each candidate node that meets the constraints is obtained. This makes it easier to determine the target candidate node in the candidate node set. The target candidate node can effectively shorten the backup time and achieve balanced storage load.

[0121] In one exemplary embodiment, such as Figure 6 As shown, the specific processing steps for determining the target candidate node in the data center based on the data center network model, state space, and action space in step 204 include steps 602 to 606. Wherein:

[0122] Step 602: Based on the data center network model, perform prediction processing on the state space and action space to obtain the initial reward value.

[0123] The data center network model is obtained by training a reinforcement learning model based on a sample network topology dataset.

[0124] In implementation, a reinforcement learning model is pre-configured in the terminal. This reinforcement learning model is Deep Q-Learning, or DQL for short, a deep learning-based reinforcement learning algorithm. The terminal trains the reinforcement learning model based on a sample network topology dataset to obtain a data center network model. The terminal inputs the state space and action space into the data center network model, and uses the data center network model to perform prediction processing on the state space and action space to obtain the initial reward value.

[0125] In an optional embodiment, if the current node is not a data center node, the terminal contains the reward value corresponding to the previous data node. The terminal inputs the state space, action space, and reward value corresponding to the previous data node into the data center network model, and performs prediction processing on the state space and action space of the reward value corresponding to the previous data node through the data center network model to obtain the initial reward value.

[0126] In one exemplary embodiment, Figure 7Figure 1 is a schematic diagram of a data center network model according to an embodiment. Figure 7 As shown in Figure 1, the terminal determines a state space and an action space based on a network topology dataset representing a data center network environment. The terminal inputs the state space and the action space into a data center network model (Q-net), and performs prediction processing on the state space and the action space by the Q-net to obtain an initial reward value. The terminal performs optimization processing on the initial reward value based on a heuristic optimization algorithm to obtain a reward value.

[0127] At step 604, the terminal performs optimization processing on the initial reward value based on a heuristic optimization algorithm to obtain a reward value.

[0128] The heuristic optimization algorithm is a particle swarm optimization (PSO) algorithm.

[0129] In implementation, the terminal performs optimization processing on the initial reward value based on a particle swarm optimization algorithm to obtain a reward value.

[0130] Specifically, the initial reward value is . The terminal determines the initial reward value as a position of each particle of the PSO algorithm, and initializes a velocity set of the particle as . Then, the terminal updates the velocity of the particle based on a velocity update algorithm to obtain an updated particle velocity. Then, the terminal updates the position of the particle according to a position update algorithm to obtain an updated particle position, and determines the updated particle position as the reward value. The velocity update algorithm is shown in the following formula (4):

[0131] (4)

[0132] In the above formula (4), is a velocity set of the particle, is an inertia weight, and are learning factors, and are random numbers. The inertia weight, the learning factors, and the random numbers are all parameters in the particle swarm optimization algorithm. is an individual best position of the particle. is the updated particle velocity.

[0133] The position update algorithm is shown in the following formula (5):

[0134] (5)

[0135] In the above formula (5), is the velocity of the particle. is the position of the particle. is the updated particle position.

[0136] Step 606: Determine the target candidate node in the action space based on the reward value and probability algorithm.

[0137] The action space is a set of candidate nodes. The set of candidate nodes contains all candidate nodes.

[0138] In practice, the terminal calculates the probability of each candidate node based on the reward value and probability algorithm, and determines the candidate node with the highest probability as the target candidate node.

[0139] Specifically, for each candidate node in the action space, the terminal determines the corresponding candidate link and its link status data. Then, based on a probabilistic algorithm, the terminal performs probability calculations on the reward value, the candidate node's node status data, and the candidate link's link status data to obtain the probability of the candidate node. Finally, based on these probabilities, the terminal determines the target candidate node in the action space.

[0140] In this embodiment, by combining heuristic optimization algorithms with a data center network model, the impact of reward sparsity and limited exploration capabilities of the data center network model is mitigated, thereby improving the model's exploration ability, avoiding getting trapped in local optima, and enhancing its performance. Simultaneously, based on the data center network model, predictive processing is performed on the state space and action space to determine target candidate nodes, solving the high-dimensional data center network disaster recovery and backup problem, improving the transmission efficiency of the data to be backed up, and thus enhancing the efficiency of the disaster recovery and backup method.

[0141] In one exemplary embodiment, such as Figure 8 As shown, the action space includes each candidate node, and the specific processing procedure of step 606 includes steps 802 to 806. Wherein:

[0142] Step 802: For each candidate node, determine the alternative link between the candidate node and the current node.

[0143] In practice, for each candidate node, the terminal determines the candidate link connecting the candidate node and the current node in each network link, and determines the link status data corresponding to the candidate link.

[0144] Step 804: Perform probability calculations on the reward value, the node status data of the candidate node, and the link status data of the candidate link according to the probability algorithm to obtain the probability of the candidate node.

[0145] The probabilistic algorithm is based on the Russian roulette algorithm and heuristic factors. Node state data includes the source data volume and storage capacity of candidate nodes. Link state data includes the remaining bandwidth of candidate links.

[0146] In implementation, the terminal performs a difference calculation on the storage capacity and source data volume of each candidate node to obtain the remaining capacity of the candidate node. Then, the terminal performs an addition operation on the remaining capacity of each candidate node to obtain the sum of the remaining capacities. The terminal then performs a probability calculation on the reward value, the remaining capacity of the candidate node, the sum of the remaining capacities, and the remaining bandwidth according to a probability algorithm to obtain the probability of selecting a candidate node.

[0147] The probability algorithm is shown in formula (6) below:

[0148] (6)

[0149] In the above formula (6), As alternative nodes The probability of. These are preset control parameters. This indicates the remaining capacity of the candidate node. This is the sum of the remaining capacity of each candidate node. This represents the reward value for each candidate node. This represents the sum of the remaining bandwidth of each alternative link. This is the reward value for the candidate node. This represents the remaining bandwidth of the candidate node.

[0150] Step 806: Determine the candidate node with the highest probability among all candidate nodes as the target candidate node.

[0151] In practice, the terminal determines the highest probability among all probabilities and identifies the candidate node corresponding to the highest probability as the target candidate node.

[0152] In this embodiment, a probabilistic algorithm is used to determine the probability of each candidate node based on the reward value, node status data, and link status data. This clarifies the probability of each candidate node being selected, and the candidate node with the highest probability is determined as the target candidate node. This yields target candidate nodes with good node status and idle network links. Furthermore, the data to be backed up is transmitted from the current node to the target candidate node with good status via an idle network link, improving the transmission efficiency of the data to be backed up and thus enhancing the efficiency of the disaster recovery backup method.

[0153] In one exemplary embodiment, such as Figure 9 As shown, before the target candidate node is determined in the data center based on the data center network model, the state space, and the action space in step 204, the specific processing procedure of this disaster recovery backup method includes steps 902 to 904. Wherein:

[0154] Step 902: Initialize the sample network topology model and the sample network topology dataset of the sample network topology model.

[0155] In implementation, the terminal generates a sample network topology model based on a random network topology generation model. Then, the terminal initializes the sample network topology dataset of the sample network topology model.

[0156] Specifically, the random network topology generation model is the Waxman-Salama model (a model used to generate random network topologies). The terminal generates a sample network topology model based on the Waxman-Salama model and generates the network topology graph G(V,E) of the sample network topology model on the Python (a coding language) platform. The sample network topology model is a network topology model of a sample data center. V represents each sample data node in the sample data center, and E represents the connection relationship between the sample data nodes. The terminal initializes the sample node state data of each data node and the sample link state data of each sample network link, and constructs the sample network topology dataset of the sample network topology model based on the sample node state data and the sample link state data.

[0157] In one exemplary embodiment, Figure 10 This is a network topology diagram of a sample network topology model in one embodiment. For example... Figure 10 As shown, the sample data center contains 30 sample data nodes. The terminal randomly selects from these 30 sample data nodes. As a sample data center node The remaining nodes serve as sample backup nodes, while the other nodes are sample intermediate nodes. This indicates the sample data center nodes that require backup, and the corresponding backup data volume for each sample data center node is [missing data]. . This represents the sample backup node. The terminal initializes the source data volume for each sample data node. Storage capacity Transmission bandwidth between sample data nodes Terminal determination For sample data center nodes To the sample backup node The There are several transmission paths, and set P is the set of all backup transmission links. The terminals are the network links of each sample. The allocated bandwidth is used This indicates. Then... , indicating all transmission paths In the link Total bandwidth used, of which binary variables Used to indicate backup links Whether the backup link was used. The remaining bandwidth is... .

[0158] At step 904, the preset reinforcement learning model is trained based on the sample network topology dataset, and a data center network model is obtained.

[0159] In implementation, the reinforcement learning model is preset in the terminal. The terminal randomly determines each data center node as a sample current node in the sample network topology model, and constructs a sample state space and a sample action space for each sample current node based on the sample network topology dataset. The terminal trains the reinforcement learning model based on the sample state space and the sample action space of each sample current node until the reinforcement learning model meets the preset training completion condition. The terminal determines the reinforcement learning model as the data center network model. The specific processing process of constructing the sample state space and the sample action space is the same as that of constructing the state space and the sample space, please refer to the description of steps 204 or 302-306 above.

[0160] Specifically, the terminal inputs the sample state space and the sample action space into the reinforcement learning model for each sample current node, obtains a first initial output result, and optimizes the initial output result through the PSO algorithm to obtain a first output result. Then, the terminal determines a sample target candidate node of the sample current node based on the output result, and updates the sample target candidate node as the sample current node. The terminal constructs a sample state space and a sample action space for the sample current node based on the sample network topology dataset, and inputs the sample state space and the sample action space into the reinforcement learning model to obtain a second initial output result. Then, the terminal optimizes the first initial output result through the PSO algorithm to obtain a second output result. The terminal operates the first output result and the second output result through a loss function algorithm to obtain a loss value. The loss function algorithm is shown in the following formula set (7):

[0161] (7)

[0162] In the above formula set (7), is the first output result, is the second output result. is the loss value.

[0163] The terminal determines whether the loss value reaches a preset loss threshold. If the loss value reaches the loss threshold, the terminal determines that the reinforcement learning model meets a preset training completion condition, and determines that the reinforcement learning model is the data center network model. If the loss value does not reach the loss threshold, the terminal determines the second initial output result as the first initial output result, and determines the second output result as the first output result. Then, the terminal determines that the reinforcement learning model does not meet the training completion condition, and continues to perform the step of constructing the sample state space and the sample action space of each sample current node based on the sample network topology dataset until the loss value reaches the preset loss threshold.

[0164] In this embodiment, the preset reinforcement learning model is trained based on the sample network topology dataset to obtain the data center network model, thereby improving the accuracy of the data center network model, and further improving the accuracy of the target candidate node based on the data center network model.

[0165] It should be understood that, although each step in the flowchart involved in each of the above-described embodiments is shown in sequence according to the arrow, these steps are not necessarily executed in sequence according to the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other sequences. Moreover, at least part of the steps in the flowchart involved in each of the above-described embodiments can include multiple steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence of these steps or stages is not necessarily sequential, but can be executed alternately or alternately with at least part of other steps or steps or stages in other steps.

[0166] Based on the same inventive concept, the embodiments of the present application also provide a disaster backup device for implementing the disaster backup method described above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more disaster backup device embodiments provided below can refer to the limitations of the disaster backup method described above, which will not be repeated here.

[0167] In one exemplary embodiment, as shown in Figure 11 A disaster backup device 1100 is provided, which includes an acquisition module 1101, a determination module 1102, and a transmission module 1103, wherein:

[0168] The acquisition module 1101 is configured to acquire a network topology dataset of a data center where the current node is located. The network topology dataset includes node state data of each data node and link state data of each network link.

[0169] The determining module 1102 is configured to construct a state space and an action space of the current node based on the node state data and the link state data, and determine a target candidate node in the data center based on the data center network model, the state space and the action space. The target candidate node is a way node to the target backup node.

[0170] The transmitting module 1103 is configured to transmit the data to be backed up from the current node to the target candidate node until the target backup node.

[0171] In an example embodiment, the determining module 1102 includes a first constructing submodule and a first determining submodule. The first constructing submodule includes:

[0172] The second determining submodule is configured to determine a total state data set of the data center based on the node state data and the link state data.

[0173] The second constructing submodule is configured to construct the state space of the current node based on the total state data set, the network links and a preset data average proportion.

[0174] The first screening submodule is configured to screen the action space of the current node in the data center according to the state space, a preset data quantity difference parameter and a normal traffic proportion of the network links.

[0175] In an example embodiment, the data nodes include data center nodes and backup nodes, the link state data includes usage states and transmission bandwidths of the network links, the node state data includes source data quantities and storage capacities of the data nodes, and the second determining submodule includes:

[0176] The third determining submodule is configured to determine a remaining storage capacity of the data center based on the storage capacities and the source data quantities of the backup nodes.

[0177] The fourth determining submodule is configured to determine a backup remaining data quantity of the data center based on the source data quantities of the data center nodes and the source data quantities of the backup nodes.

[0178] The fifth determining submodule is configured to determine a total remaining bandwidth of the data center based on the transmission bandwidths of the network links in an idle state.

[0179] The third constructing submodule is configured to construct the total state data set of the data center based on the total remaining bandwidth, the backup remaining data quantity and the remaining storage capacity of the data center.

[0180] In an example embodiment, the first screening submodule includes:

[0181] The fourth constructing submodule is configured to construct a constraint condition set according to the state space, the preset data quantity difference parameter and the normal traffic proportion of the network links.

[0182] The second screening submodule is configured to screen the next hop node of the current node in each data node of the data center based on the constraint condition set, to obtain a candidate node set, and determine the candidate node set as the action space of the current node.

[0183] In an example embodiment, the first determining submodule includes:

[0184] The first processing submodule is configured to perform prediction processing on the state space and the action space based on the data center network model, to obtain an initial reward value.

[0185] The second processing submodule is configured to perform optimization processing on the initial reward value according to a heuristic optimization algorithm, to obtain a reward value.

[0186] The sixth determining submodule is configured to determine a target candidate node in the action space according to the reward value and a probability algorithm.

[0187] In an example embodiment, the sixth determining submodule includes:

[0188] The seventh determining submodule is configured to determine, for each candidate node, a candidate link between the candidate node and the current node.

[0189] The first operation submodule is configured to perform probability operation on the reward value, the node state data of the candidate node, and the link state data of the candidate link according to the probability algorithm, to obtain a probability of the candidate node.

[0190] The eighth determining submodule is configured to determine, among the candidate nodes, a candidate node with the highest probability as the target candidate node.

[0191] Each of the above modules in the disaster backup device can be realized by software, hardware, or a combination thereof, in whole or in part. Each of the above modules can be embedded in or independent of a processor in a computer device in hardware form, or can be stored in a memory in the computer device in software form, so as to be called and executed by a processor to perform the operations corresponding to each of the above modules.

[0192] In an example embodiment, a computer device is provided, which can be a terminal, and an internal structure diagram of the computer device can be as shown in Figure 12The computer device shown in the figure includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capability. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner. The wireless manner can be realized through WIFI, mobile cellular network, near field communication (NFC) or other technologies. The computer program is executed by the processor to implement a disaster recovery method. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball or touchpad arranged on the shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0193] Those skilled in the art can understand that, Figure 12 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0194] In one embodiment, a computer device is also provided, including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps in each of the above method embodiments.

[0195] In one embodiment, a computer readable storage medium is provided, which stores a computer program, and the computer program is executed by a processor to implement the steps in each of the above method embodiments.

[0196] In one embodiment, a computer program product is provided, including a computer program, and the computer program is executed by a processor to implement the steps in each of the above method embodiments.

[0197] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiment methods. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. The non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. The volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, the RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., without being limited thereto.

[0198] The technical features of the above embodiments can be combined arbitrarily. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combinations of the technical features do not exist contradictory, they should be considered as the scope of the present application.

[0199] The above-described embodiments are merely illustrative of several embodiments of the present application, and the description is relatively specific and detailed, but should not be understood as a limitation on the scope of the patent. It should be noted that for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the scope of the present application. Therefore, the scope of protection of the present application should be subject to the appended claims.

Claims

1. A method of disaster backup, characterized by, The method comprises: obtaining a network topology dataset of a data center where a current node is located; the network topology dataset comprises node state data of each data node and link state data of each network link; each data node comprises each data center node and each backup node; the link state data comprises a usage state and a transmission bandwidth of the network link, and the node state data comprises a source data volume and a storage capacity of the data node; constructing a state space and an action space of the current node based on each node state data and each link state data, and determining a target candidate node in the data center based on a data center network model, the state space and the action space; the target candidate node is a way node to a target backup node; transmitting backup data from the current node to the target candidate node until the target backup node; the constructing of the state space and the action space of the current node based on each node state data and each link state data comprises: determining a total state dataset of the data center based on each node state data and each link state data; constructing a state space of the current node based on the total state dataset, each network link and a preset data average ratio; the data average ratio is an average ratio of a data volume stored by the target backup node to a total backup data volume; screening an action space of the current node in the data center according to the state space, a preset data volume difference parameter and a network link normal business ratio; the data volume difference parameter is a difference threshold value between data volumes received by each backup node; the network link normal ratio is a bandwidth ratio used for transmitting normal business data; the determining of the total state dataset of the data center based on each node state data and each link state data comprises: determining a backup residual data volume and a residual storage capacity of the data center based on each storage capacity and each source data volume; determining a total residual bandwidth of the data center based on each link state data, and constructing a total state dataset of the data center based on the total residual bandwidth, the backup residual data volume and the residual storage capacity.

2. The method of claim 1, wherein, the determining of the total state dataset of the data center based on each node state data and each link state data comprises: determining a residual storage capacity of the data center according to a storage capacity and a source data volume of each backup node; determining a backup residual data volume of the data center according to a source data volume of each data center node and a source data volume of each backup node; determining a total residual bandwidth of the data center based on a transmission bandwidth of each network link in an idle state; constructing a total state dataset of the data center based on the total residual bandwidth, the backup residual data volume and the residual storage capacity of the data center.

3. The method of claim 1, wherein, the screening of the action space of the current node in the data center according to the state space, the preset data volume difference parameter and the network link normal business ratio comprises: construct a constraint condition set according to the state space, a preset data quantity difference parameter, and a normal service proportion of the network link; filter a next hop node of the current node in each data node of the data center based on the constraint condition set to obtain a candidate node set, and determine the candidate node set as an action space of the current node.

4. The method of claim 3, wherein, The filtering of the next hop node of the current node in each data node of the data center based on the constraint condition set to obtain the candidate node set comprises: filtering the next hop node of the current node in each data node of the data center to obtain an initial candidate node set; filtering each candidate node in the initial candidate node set that meets the constraint condition set to obtain the candidate node set.

5. The method of claim 1, wherein, The determination of the target candidate node in the data center based on the data center network model, the state space, and the action space comprises: performing prediction processing on the state space and the action space based on the data center network model to obtain an initial reward value; performing optimization processing on the initial reward value according to a heuristic optimization algorithm to obtain a reward value; determining the target candidate node in the action space according to the reward value and a probability algorithm.

6. The method of claim 5, wherein, The action space contains each candidate node, and the determination of the target candidate node in the action space according to the reward value and the probability algorithm comprises: determining, for each candidate node, a candidate link between the candidate node and the current node; performing probability operation on the reward value, node state data of the candidate node, and link state data of the candidate link according to a probability algorithm to obtain a probability of the candidate node; determining, among each candidate node, a candidate node with the highest probability as the target candidate node.

7. A disaster backup apparatus characterized by comprising: The apparatus comprises: an acquisition module configured to acquire a network topology data set of a data center where a current node is located; the network topology data set contains node state data of each data node and link state data of each network link; each data node contains each data center node and each backup node; the link state data contains a use state and a transmission bandwidth of the network link, and the node state data contains source data quantity and storage capacity of the data node; a determination module configured to construct a state space and an action space of the current node based on each node state data and each link state data, and determine a target candidate node in the data center based on a data center network model, the state space, and the action space; the target candidate node is a way node to a target backup node; a transmission module configured to transmit backup data from the current node to the target candidate node until the target backup node; the determination module comprises a first construction submodule, and the first construction submodule comprises: a second determination submodule configured to determine a total state data set of the data center based on each node state data and each link state data. a second constructing submodule configured to construct a state space of the current node based on the total state data set, the network links, and a preset data average proportion, the data average proportion being an average ratio of an amount of data stored by a target backup node to a total amount of backup data; a first screening submodule configured to screen an action space of the current node in the data center according to the state space, a preset data amount difference parameter, and a network link normal traffic proportion, the data amount difference parameter being a difference threshold value between amounts of data received by each backup node, and the network link normal proportion being a bandwidth proportion used for transmitting normal traffic data; the second determining submodule is specifically configured to determine a backup residual data amount and a residual storage capacity of the data center based on the storage capacities and the source data amounts, determine a total residual bandwidth of the data center based on the link state data, and construct a total state data set of the data center based on the total residual bandwidth, the backup residual data amount, and the residual storage capacity.

8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that, The processor executes the computer program to implement the steps of the method in any one of claims 1 to 6.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method in any one of claims 1 to 6.

10. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method in any one of claims 1 to 6.

Images