Traffic scheduling method and device for multiple data centers, network equipment, readable storage medium and program product

Abstract

The invention relates to a traffic scheduling method and device for multiple data centers, network equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring traffic time sequence data and path topology data of each data center pair; obtaining a comprehensive correlation coefficient of each two data center pairs according to the correlation between the traffic time sequence data and the path topology data of each two data center pairs; determining high-association center pairs in the plurality of data center pairs according to the comprehensive association coefficient, and monitoring the real-time traffic load of each high-association center pair; and under the condition that the real-time traffic load of the high-association center pair group meets a scheduling condition, performing traffic scheduling on each data center pair in the high-association center pair group. By adopting the method, the interruption probability of the network service can be reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of network, in particular to a multi-data center traffic scheduling method and device, network equipment, computer readable storage medium and computer program product. BACKGROUND

[0002] With the development of cloud computing and distributed technology, network system services gradually evolve from single data center to multi-data center deployment mode, and can realize cross-data center intercommunication by establishing data center interconnection links. In related technologies, the traffic scheduling of network system usually takes single data center and single link as the decision unit, and relies on the passive response mechanism of real-time alarm. However, this way is difficult to efficiently schedule the multi-data center network system with complex association, and is easy to cause chain congestion and lead to network service interruption. SUMMARY

[0003] Therefore, it is necessary to provide a multi-data center traffic scheduling method, device, network equipment, computer readable storage medium and computer program product to solve the above technical problems.

[0004] In a first aspect, the present application provides a multi-data center traffic scheduling method, comprising:

[0005] obtaining traffic time series data and path topology data of each data center pair;

[0006] According to the correlation of the traffic time series data and the path topology data of each two data center pairs, the comprehensive correlation coefficient of each two data center pairs is obtained;

[0007] According to the comprehensive correlation coefficient, a high correlation center pair group in the plurality of data center pairs is determined, and the real-time traffic load of each high correlation center pair group is monitored;

[0008] In the case that the real-time traffic load of the high correlation center pair group meets the scheduling condition, the traffic scheduling is performed on each data center pair in the high correlation center pair group.

[0009] In one embodiment, the comprehensive correlation coefficient of each two data center pairs is obtained according to the correlation of the traffic time series data and the path topology data of each two data center pairs, comprising: calculating the traffic change correlation coefficient of two data center pairs according to the traffic time series data of two data center pairs; calculating the topology coupling coefficient of two data center pairs according to the path topology data of two data center pairs; and obtaining the comprehensive correlation coefficient of two data center pairs according to the traffic change correlation coefficient and the topology coupling coefficient.

[0010] In one of the embodiments, the calculating the traffic change correlation coefficient of the two data center pairs according to the traffic time series data of the two data center pairs comprises: calculating the time lag correlation coefficient of the traffic time series data of the two data center pairs at a plurality of time offsets respectively; and obtaining the traffic change correlation coefficient of the traffic time series data of the two data center pairs according to the time lag correlation coefficient corresponding to each time offset.

[0011] In one of the embodiments, the calculating the topology coupling coefficient of the two data center pairs according to the path topology data of the two data center pairs comprises: obtaining the node set and the link set of the two data center pairs according to the path topology data of the two data center pairs; obtaining the node coupling coefficient of the two data center pairs according to the set size of the node set and the shared nodes between the node set; obtaining the link coupling coefficient of the two data center pairs according to the set size of the link set and the shared links between the link set; and obtaining the topology coupling coefficient according to the node coupling coefficient and the link coupling coefficient.

[0012] In one of the embodiments, the obtaining the node coupling coefficient of the two data center pairs according to the set size of the node set and the shared nodes between the node set comprises: obtaining the node weight of each shared node according to the betweenness centrality index of each shared node in the network; and obtaining the node coupling coefficient according to the set size of the node set and the node weight of each shared node.

[0013] In one of the embodiments, before the performing traffic scheduling on each data center pair in the high-correlation center pair group in the case that the real-time traffic load of the high-correlation center pair group meets the scheduling condition, the method comprises: obtaining the load excess index of each data center pair according to the real-time traffic load of each data center pair in the high-correlation center pair group; calculating the linkage load index of the high-correlation center pair group according to the load excess index of each data center pair, the center pair weight of each data center pair, and the comprehensive correlation coefficient between the data center pairs; and determining that the real-time traffic load of the high-correlation center pair group meets the scheduling condition in the case that the linkage load index is greater than a linkage load threshold.

[0014] In a second aspect, the present application further provides a traffic scheduling device for multiple data centers, comprising:

[0015] a data acquisition module, configured to acquire traffic time series data and path topology data of each data center pair;

[0016] The correlation analysis module is configured to obtain a comprehensive correlation coefficient of each pair of data center pairs according to the correlation of the traffic time series data and the path topology data of each pair of data center pairs.

[0017] The center pair combination module is configured to determine a high-correlation center pair group from the plurality of data center pairs according to the comprehensive correlation coefficients, and monitor real-time traffic loads of each high-correlation center pair group.

[0018] The combination scheduling module is configured to perform traffic scheduling on each data center pair in the high-correlation center pair group when the real-time traffic loads of the high-correlation center pair group meet a scheduling condition.

[0019] In a third aspect, the present application further provides a network device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the following steps when executing the computer program:

[0020] The traffic time series data and the path topology data of each data center pair are obtained.

[0021] A comprehensive correlation coefficient of each pair of data center pairs is obtained according to the correlation of the traffic time series data and the path topology data of each pair of data center pairs.

[0022] A high-correlation center pair group is determined from the plurality of data center pairs according to the comprehensive correlation coefficients, and real-time traffic loads of each high-correlation center pair group are monitored.

[0023] Traffic scheduling is performed on each data center pair in the high-correlation center pair group when the real-time traffic loads of the high-correlation center pair group meet a scheduling condition.

[0024] In a fourth aspect, the present application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the following steps:

[0025] The traffic time series data and the path topology data of each data center pair are obtained.

[0026] A comprehensive correlation coefficient of each pair of data center pairs is obtained according to the correlation of the traffic time series data and the path topology data of each pair of data center pairs.

[0027] A high-correlation center pair group is determined from the plurality of data center pairs according to the comprehensive correlation coefficients, and real-time traffic loads of each high-correlation center pair group are monitored.

[0028] Traffic scheduling is performed on each data center pair in the high-correlation center pair group when the real-time traffic loads of the high-correlation center pair group meet a scheduling condition.

[0029] In a fifth aspect, the present application also provides a computer program product comprising a computer program which, when executed by a processor, implements the following steps:

[0030] obtaining traffic time series data and path topology data of each pair of data centers;

[0031] obtaining a comprehensive correlation coefficient of each pair of data centers according to the correlation of the traffic time series data and the path topology data of each pair of data centers;

[0032] determining a high-correlation center pair group in the plurality of pairs of data centers according to the comprehensive correlation coefficient, and monitoring real-time traffic load of each high-correlation center pair group;

[0033] performing traffic scheduling on each pair of data centers in the high-correlation center pair group when the real-time traffic load of the high-correlation center pair group meets a scheduling condition.

[0034] The traffic scheduling method, device, network equipment, computer readable storage medium and computer program product of the plurality of data centers, first obtain traffic time series data and path topology data of each pair of data centers, and then obtain a comprehensive correlation coefficient of each pair of data centers according to the correlation of the traffic time series data and the path topology data of each pair of data centers. Subsequently, a high-correlation center pair group in the plurality of pairs of data centers is determined according to the comprehensive correlation coefficient, and real-time traffic load of each high-correlation center pair group is monitored. Then, traffic scheduling is performed on each pair of data centers in the high-correlation center pair group when the real-time traffic load of the high-correlation center pair group meets a scheduling condition. According to the scheme, for a network system of a plurality of data centers, a comprehensive correlation coefficient of each pair of data centers is calculated according to the correlation of traffic time series data and path topology data between each pair of data centers. The correlation between the two pairs of data centers can be quantified from the dynamic fluctuation of business traffic and the coupling of physical topology. Therefore, according to the comprehensive correlation coefficient between each pair of data centers, a high-correlation center pair group with traffic coordination characteristics can be effectively located in the network system. Subsequently, by monitoring the real-time traffic load of the high-correlation center pair group and performing traffic scheduling when the real-time traffic load meets a scheduling condition, the traffic scheduling of the plurality of pairs of data centers with traffic coordination characteristics can be cooperatively decided and processed. The limitations of isolated analysis in the traditional method are overcome to achieve global cooperative optimization of complex networks. The efficiency of traffic scheduling can be improved, and the probability of network service interruption can be reduced. BRIEF DESCRIPTION OF DRAWINGS

[0035] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the following will briefly introduce the drawings needed to be used in the description of the embodiments of the present application or the related art. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 An application environment diagram of a traffic scheduling method for multiple data centers in an embodiment;

[0037] Figure 2 A flowchart of a traffic scheduling method for multiple data centers in an embodiment;

[0038] Figure 3 A flowchart of obtaining an integrated correlation coefficient of each pair of data centers in an embodiment;

[0039] Figure 4 A flowchart of obtaining a topological coupling coefficient of a pair of data centers in an embodiment;

[0040] Figure 5 A flowchart of a traffic scheduling method for multiple data centers in another embodiment;

[0041] Figure 6 A structural block diagram of a traffic scheduling device for multiple data centers in an embodiment;

[0042] Figure 7 An internal structure diagram of a network device in an embodiment. DETAILED DESCRIPTION

[0043] In order to make the objects, technical solutions and advantages of the present application clearer, the following will further describe the present application in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0044] It should be noted that the terms "first", "second", and the like used in the present application can be used to describe various objects, but these objects are not limited by these terms. These terms are only used to distinguish the first object from the second object. The terms "include" and "have" and any variations thereof used in the present application are intended to cover non-exclusive inclusion. The term "multiple" used in the present application refers to two and more than two. The term "and / or" used in the present application refers to one of the solutions, or any combination of multiple solutions.

[0045] Specifically, the network service can be deployed in a cloud computing environment, such as Figure 1The diagram illustrates a multi-datacenter network system. A data center (DC) can be a centralized information infrastructure, deploying servers, storage devices, network equipment, and supporting facilities to centrally process, store, manage, and interact with data, thereby supporting the operation of various business applications. Data centers can interconnect using tunnels based on the IPv6 forwarding plane, such as segment routing IPv6 (SRv6) and virtual eXtensible Local Area Networks (VxLAN), forming a logical overlay built upon the physical underlay. The physical network provides the bearer and transmission support for the logical topology, enabling cross-datacenter access for network system services.

[0046] In multi-datacenter network systems, traffic patterns of datacenter pairs with similar service call relationships or related topologies are prone to correlation. For example, mutual service calls between different datacenters can easily lead to traffic interactions with certain time-series characteristics. If the service call dependencies of two datacenter pairs are similar, their dynamic traffic changes are likely to be significantly correlated. Another example is that the interconnection links of different datacenter pairs may be carried by different underlying network paths, but these paths may pass through common router nodes, leading to correlations in traffic forwarding due to topological associations. For instance, taking... Figure 1 Taking the network system shown as an example, assuming data centers A and B deploy backend services, and data center C deploys data storage services, then the traffic time-series changes of data centers (A, C) and (B, C) will have a certain correlation due to their similar service dependencies. For example, Figure 1 The interconnection link between the data center and (A, C) is carried by the underlying network path R1 (Router 1) → R2 (Router 2) → R4 (Router 4), and the interconnection link between the data center and (B, C) is carried by the path R3 (Router 3) → R4 (Router 4). Both underlying paths are routed through R4 (Router 4), and there will be mutual influence at the traffic forwarding level.

[0047] Traditional data center interconnection traffic scheduling methods generally rely on real-time alarm mechanisms and develop strategies to trigger scheduling in units of a single data center or a single link, without considering the dynamic fluctuations of business and easily ignoring the changes in traffic and the relevance of topology connection between different data center pairs. Thus, the traditional method is easy to split the complex associated network system into isolated individuals, which cannot capture the collaborative correlation characteristics formed by topology association and business dependence between different data center pairs, and is difficult to cope with the topology complexity and business dynamics brought by the expansion of data center scale. In the scenes such as e-commerce promotion and financial transactions, which have high requirements for network service quality, it is difficult to predict the congestion diffusion risk of associated links, and it is easy to cause resource allocation imbalance and scheduling response lag, ultimately leading to inefficient scheduling and chain congestion, causing business interruption.

[0048] In one embodiment, as shown in Figure 2 , a multi-data center traffic scheduling method is provided, which can be applied to a network device in a network system as shown in Figure 1 . The network device can be a server, which can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. In this embodiment, the method includes the following steps:

[0049] Step S201, obtaining traffic time series data and path topology data of each data center pair.

[0050] The traffic time series data of the data center pair can be the traffic data sequence of the two data centers in a period of time, and the path topology data of the data center pair can include nodes, links, etc. contained in the underlying path of the interconnection link between the two data centers.

[0051] Exemplarily, a traffic analysis device can be deployed at the outlet of the data center, and the traffic information between the source data center and the destination data center can be recorded according to the sampling time interval, the packet destination address, etc. Sampling conditions, and the collected records at multiple times are subjected to data cleaning, standardization processing, etc. to form the traffic time series data of the data center pair. Exemplarily, the traffic time series data of the data center pair P can be expressed in the following mathematical form:

[0052]

[0053] Wherein, represents the traffic value of the data center pair at the i-th moment.

[0054] Exemplarily, the path topology data of the data center pair P can be obtained according to the data center interconnection link information constructed based on SRv6, VxLAN, etc. tunneling, which can include node set data and link set data of the interconnection path. Exemplarily, the path topology data of the data center pair P can be expressed in the following mathematical form:

[0055]

[0056]

[0057] wherein, represents the node set data of the data center pair P, and each element in the set respectively represents a node contained in the underlying path carrying the interconnection link of the data center pair P; represents the link set data of the data center pair P, and each element in the set respectively represents a link contained in the underlying path carrying the interconnection link of the data center pair P.

[0058] Exemplarily, assuming that the interconnection link of the data center pair P is carried by the R1→R2→R3 path of the underlying network, the node set data of the data center pair can be expressed as , and the link set data can be expressed as .

[0059] In this step, the above data can be obtained for each data center pair in the network system that exists data interaction.

[0060] Step S202, obtaining the comprehensive correlation coefficient of each two data center pairs according to the correlation of the traffic time series data and the path topology data of each two data center pairs.

[0061] Wherein, there can be multiple data center pairs in the network system. In this step, multiple data center pairs can be combined two by two to form multiple data center pair groups, and the comprehensive correlation coefficient of each data center pair group (i.e. each two data center pairs) is calculated.

[0062] Wherein, the comprehensive correlation coefficient of two data center pairs can reflect the correlation degree of traffic changes and the coupling degree of path topology between the data center pairs. Exemplarily, the correlation degree of traffic changes and the coupling degree of path topology between two data center pairs can be quantified according to the traffic time series data and the path topology data of the two data center pairs respectively, and then the comprehensive correlation coefficient of the two data center pairs can be obtained according to the quantification result. The greater the comprehensive correlation coefficient of two data center pairs, the higher the correlation degree of traffic changes and the coupling degree of path topology between the two data center pairs.

[0063] In step S203, a high-correlation center pair group is determined according to the comprehensive correlation coefficient, and real-time traffic load of each high-correlation center pair group is monitored.

[0064] The comprehensive correlation coefficient between each two data center pairs can be used to classify the data center pair groups in the network system. For example, a classification threshold parameter can be set in advance according to the actual network situation and business requirements of the network system, so as to divide the data center pair groups into high-correlation or low-correlation groups. When the comprehensive correlation coefficient of a data center pair group is not less than the classification threshold parameter, the data center pair group is determined as a high-correlation center pair group. When the comprehensive correlation coefficient of a data center pair group is less than the classification threshold parameter, the data center pair group is determined as a low-correlation center pair group.

[0065] The greater the comprehensive correlation coefficient between two data center pairs, the more significant the traffic synchronous fluctuation and the closer the path coupling, and thus the higher the risk of collaborative congestion. When two data center pairs simultaneously generate high traffic load, the overlapped underlying link is prone to be impacted. Therefore, the real-time traffic load of each high-correlation center pair group in the network system can be monitored in this step.

[0066] In step S204, when the real-time traffic load of the high-correlation center pair group meets the scheduling condition, the traffic of each data center pair in the high-correlation center pair group is scheduled.

[0067] The scheduling condition can be set according to historical fault data of the network system, for example, the statistical value of the real-time traffic load of each data center pair in the high-correlation center pair group exceeds a warning threshold. When the real-time traffic load of the high-correlation center pair group meets the scheduling condition, the traffic of each data center pair in the high-correlation center pair group is scheduled. For example, the traffic of two data center pairs in the high-correlation center pair group can be collaboratively scheduled by using optimization methods such as traffic engineering, backup underlying path switching, reserved bandwidth configuration, and the like, so as to alleviate the network congestion of the data center interconnection and the underlying bearing network.

[0068] In the multi-data center traffic scheduling method, for the network system of the multi-data centers, the correlation coefficient of each pair of data centers is calculated according to the correlation of the traffic time series data and the path topology data of each pair of data centers, the correlation between the two data centers can be quantified from the dynamic fluctuation of the business traffic and the coupling of the physical topology, and thus the high correlation center pair group with traffic coordination characteristics can be effectively located in the network system according to the correlation coefficient of each pair of data centers. Then, by monitoring the real-time traffic load of the high correlation center pair group and scheduling the traffic under the condition that the scheduling condition is met, the traffic scheduling of the multiple data center pairs with traffic coordination characteristics can be cooperatively decided and linked, the limitations of isolated analysis in the traditional way are broken through to realize the global cooperative optimization of the complex network, the efficiency of the traffic scheduling can be improved, the chain congestion can be avoided, and the interruption probability of the network service can be reduced.

[0069] In one exemplary embodiment, as shown in Figure 3 According to the correlation of the traffic time series data and the path topology data of each pair of data centers, the correlation coefficient of each pair of data centers can include:

[0070] Step S301, according to the traffic time series data of the two data center pairs, the traffic change correlation coefficient of the two data center pairs is calculated.

[0071] According to the traffic time series data of the two data center pairs, the correlation of the traffic fluctuation of the two data center pairs can be quantified to obtain the traffic change correlation coefficient of the two data center pairs. For example, the Pearson correlation analysis can be performed on the traffic time series data of the two data center pairs to obtain the traffic change correlation coefficient between them. The greater the value of the traffic change correlation coefficient, the more relevant the traffic fluctuation of the two data center pairs.

[0072] Step S302, according to the path topology data of the two data center pairs, the topology coupling coefficient of the two data center pairs is calculated.

[0073] According to the path topology data of the two data center pairs, the coupling degree of the path topology of the two data center pairs can be quantified to obtain the topology coupling coefficient of the two data center pairs. For example, the proportion of shared nodes and / or links in the path topology data of the two data center pairs can be counted to obtain the topology coupling coefficient of the two data center pairs. The greater the value of the topology coupling coefficient, the higher the coupling degree of the path topology of the two data center pairs.

[0074] Step S303, according to the traffic change correlation coefficient and the topology coupling coefficient, the correlation coefficient of the two data center pairs is obtained.

[0075] The comprehensive correlation coefficient of the data center pair P and the data center pair Q can be calculated according to the traffic change correlation coefficient and the topology coupling coefficient of the two data center pairs.

[0076]

[0077] In the formula, the comprehensive correlation coefficient of the data center pair P and the data center pair Q is The comprehensive correlation weight coefficient can be set according to the emphasis on the topology correlation or the traffic correlation.

[0078] In this embodiment, the traffic change correlation coefficient and the topology coupling coefficient of the two data center pairs are calculated respectively, and then the comprehensive correlation coefficient of the two data center pairs is calculated by combining the two coefficients. The correlation between the two data center pairs can be quantified from two dimensions of traffic fluctuation and physical topology constraint. The comprehensive correlation coefficient obtained can reflect the correlation characteristics of the two data center pairs caused by traffic synchronous fluctuation and topology coupling. The limitation of the related art that only analyzes the traffic or topology correlation in one dimension is broken through, and the accurate identification of the correlation between the two data center pairs is realized, which provides a more comprehensive decision basis for subsequent traffic scheduling.

[0079] In an exemplary embodiment, the traffic change correlation coefficient of the two data center pairs can be calculated according to the traffic time series data of the two data center pairs, which can include: calculating the time lag correlation coefficient of the traffic time series data of the two data center pairs at a plurality of time offsets respectively; and obtaining the traffic change correlation coefficient of the two data center pairs according to the time lag correlation coefficient corresponding to each time offset.

[0080] Specifically, in this embodiment, the traffic change correlation coefficient of the two data center pairs can be calculated based on the Pearson correlation method, in combination with the time series dynamics (time lag effect) and data timeliness (recent traffic is more important for current scheduling) of the traffic time series data of the two data center pairs.

[0081] ​​​Considering the potential "time lag" in traffic correlation between data center pairs, a time lag factor (time offset) can be introduced into the calculation to calculate the time lag correlation coefficient of the traffic time-series data of the two data center pairs under multiple time offsets. Furthermore, in the calculation of each time lag correlation coefficient, corresponding weight values ​​can be assigned to different time points. For example, exponential decay weights can be used to assign differentiated weights to data at different times, giving higher weights to recent data and lower weights to older data. For instance, the formula for calculating the time lag correlation coefficient can be expressed as:

[0082]

[0083] In the formula, For time window, This is the time-series traffic data from the data center to P. [This refers to the time offset of Q in the data center] Traffic time series data below, This represents the traffic data of data center P at time i. This indicates that the data center's response to Q at the i+th time... Real-time traffic data Let be the dynamic weight at time i.

[0084] For example, The calculation formula can be expressed as:

[0085]

[0086] In the formula, i is the i-th time in the time series, i=1 is the earliest time, and i=n is the latest time; The attenuation coefficient is... The larger the value, the higher the weight of recent data. For example, When the value is 0.2, the weight of the latest time (i=n) is 11 times that of 1 hour ago (i=n-12, assuming a sampling interval of 5 minutes).

[0087] After calculating the time-delay correlation coefficients corresponding to each time offset, the maximum value can be selected as the traffic change correlation coefficient between the two data center pairs. For example, the traffic change correlation coefficients for data center pairs P and Q can be expressed as:

[0088]

[0089] Therefore, we can search for the optimal time offset to find the most significant correlation pattern between the traffic time series data of two data center pairs within a certain time window.

[0090] In this embodiment, the time sequence dynamics and timeliness of the traffic are comprehensively considered to calculate the traffic change correlation coefficient. The traffic time sequence synchronization of the two data center pairs at different time offsets is quantified by using the Pearson correlation coefficient improved based on the dynamic weight of time sequence data, which can highlight the value of recent data by dynamic weighting and reflect the timeliness of traffic data. The time lag correlation coefficient at multiple time offsets is calculated by introducing a time lag factor, which can search for the optimal time offset and find the most significant correlation mode of the traffic time sequence data of the two data center pairs within a certain time window, so as to accurately quantify the traffic change correlation of the two data center pairs.

[0091] In one exemplary embodiment, as shown in FIG. 4, the method for calculating the traffic change correlation of the two data center pairs can comprise the following steps. Figure 4

[0092] Step S401, obtaining the node set and link set of the two data center pairs according to the path topology data of the two data center pairs.

[0093] Specifically, according to the path topology data of the two data center pairs, the node set and link set of the two data center pairs can be obtained. For example, for the data center pair P, its node set and link set can be obtained; for the data center pair Q, its node set and link set can be obtained.

[0094] Step S402, obtaining the node coupling coefficient of the two data center pairs according to the set size of the node set and the shared nodes between the node sets.

[0095] wherein, according to the intersection of the node sets of the two data center pairs, the shared node set of the two data center pairs can be obtained. Exemplarily, the shared node set of the data center pairs P and Q can be represented as , and the elements in the set are the shared nodes of the data center pairs P and Q. For example, the path of the data center pair P passes through nodes , , , and the path of the data center pair Q passes through nodes , , .

[0096] ​The node coupling coefficient of the pair of data centers can be calculated according to the set size of the node sets of the pair of data centers and the shared nodes therebetween. For example, in some embodiments, the node coupling coefficient of the pair of data centers can be calculated by calculating the ratio of the number of shared nodes to the maximum of the set sizes of the node sets of the pair of data centers.

[0097] In an exemplary embodiment, the node coupling coefficient of the pair of data centers can be calculated according to the set size of the node sets and the shared nodes therebetween, including: calculating the node weight of each shared node according to the betweenness centrality of the shared node in the network; and calculating the node coupling coefficient according to the set size of the node sets and the node weight of each shared node.

[0098] Specifically, different nodes in the network system can assume different roles and thus have different influences on the network as a whole. For example, the shared influence of a core router (e.g., a backbone node carrying a large amount of cross-regional traffic) and an edge router (e.g., an access node connecting only a single data center) can be quite different. In this embodiment, the importance of the role assumed by the nodes in the network can be considered, and the node coupling coefficient of the pair of data centers can be calculated by calculating the proportion of the total importance of the shared nodes in the two paths of the pair of data centers in the two paths.

[0099] Exemplarily, the node weight of each shared node can be calculated by calculating the betweenness centrality of the shared node in the network system. The node weight can measure the importance of the shared node in the network. Core nodes in the network system generally have greater node weights and higher shared risks, and edge nodes are the opposite. Subsequently, the node coupling coefficient can be calculated according to the set size of the node sets and the node weight of each shared node. Exemplarily, the node coupling coefficient of the pair of data centers P and Q can be represented as:

[0100]

[0101] In the formula, is one shared node in the pair of data centers P and Q, is the node weight of the shared node, and the value range of the node weight is [0, 1].

[0102] In this embodiment, the node weight of each shared node can be calculated according to the betweenness centrality of the shared node in the network, and the proportion of the total importance of the shared nodes in the paths of the pair of data centers, so that the importance of the role assumed by the shared nodes in the network can be fully considered, and the coupling risk between the nodes of the pair of data centers can be accurately evaluated.

[0103] ​Step S403, obtaining the link coupling coefficient of the two data center pairs according to the set size of the link set and the shared link between the link sets.

[0104] wherein the shared link set of the two data center pairs can be obtained according to the intersection of the link sets of the two data center pairs. Exemplarily, the shared link set of the data center pairs P and Q can be represented as , the elements in the set are the shared links of the data center pairs P and Q. For example, the path of the data center pair P contains the link , , the path of the data center pair Q contains the link , , then .

[0105] wherein the link coupling coefficient of the two data center pairs can be calculated according to the set size of the link sets of the two data center pairs and the shared link between them. Exemplarily, the link coupling coefficient of the data center pairs P and Q can be represented as:

[0106]

[0107] Step S404, obtaining the topology coupling coefficient according to the node coupling coefficient and the link coupling coefficient.

[0108] wherein the topology coupling coefficient between the two data center pairs can be obtained by weighted fusion according to the node coupling coefficient and the link coupling coefficient of the two data center pairs. Exemplarily, the topology coupling coefficient of the data center pairs P and Q can be represented as:

[0109]

[0110] wherein, is the topology coupling coefficient, is the topology value allocation coefficient, and the value range is [0, 1]. The value of can be set according to the importance requirement of “node coupling” or “link coupling” in the actual network situation. For example, in the topology dense scene such as backbone network, the influence of node coupling is greater, and can be set to 0.6. In the link sensitive scene such as access network, the influence of link coupling is greater, and can be set to 0.4.

[0111] ​​In this embodiment, by calculating the proportion of shared nodes in the path and the proportion of shared links in the path of each pair of data centers, the node coupling coefficient and the link coupling coefficient of the two pairs of data centers are obtained, which can quantify the coupling risk between the two pairs of data centers from the node level and the link level, and the topology coupling coefficient obtained by fusing the two coefficients can accurately reflect the fault conduction risk between the two pairs of data centers caused by topology correlation.

[0112] In one exemplary embodiment, in the case that the real-time traffic load of the high-correlation center pair group meets the scheduling condition, before scheduling the traffic of each data center pair in the high-correlation center pair group, the following steps can be included: obtaining the load excess index of each data center pair according to the real-time traffic load of each data center pair in the high-correlation center pair group; calculating the linkage load index of the high-correlation center pair group according to the load excess index of each data center pair, the center pair weight of each data center pair, and the comprehensive correlation coefficient between the data center pairs; and determining that the real-time traffic load of the high-correlation center pair group meets the scheduling condition in the case that the linkage load index is greater than the linkage load threshold.

[0113] Specifically, the scheduling condition of the high-correlation center pair group can be that the linkage load index of the two data center pairs in the center pair group is greater than the linkage load threshold. The linkage load index of the two data center pairs can be calculated according to the load excess index of each data center pair, the center pair weight of each data center pair, and the comprehensive correlation coefficient between the data center pairs.

[0114] Exemplarily, assuming that the high-correlation center pair group includes data center pairs P and Q, the load excess index of each data center pair can be calculated by the following formula:

[0115]

[0116] In the formula, 、 respectively represent the load excess index of the two data center pairs P and Q, 、 respectively represent the real-time traffic load of the two data center pairs P and Q, 、 respectively represent the traffic warning threshold of the two data center pairs P and Q, which can be set according to historical fault data or specific business requirements.

[0117] The load excess index can quantify the proportion of the real-time traffic load of a single data center pair exceeding the threshold. When the real-time traffic load does not exceed the threshold, the load excess index is 0, so that only the part of the real-time traffic exceeding the threshold can be focused on in subsequent calculations, avoiding the interference of normal load on scheduling decisions.

[0118] The linkage load index of the high-correlation center pair group can be calculated according to the load excess index of each data center pair, the center pair weight of each data center pair, and the comprehensive correlation coefficient between the data center pairs by the following formula:

[0119]

[0120] In the formula, is the linkage load index of the high-correlation center pair group; is the center pair weight of the data center pair P, is the center pair weight of the data center pair Q, , which can reflect the business importance of the two data center pairs, and the sum of the two is 1. The greater the value is, the higher the business importance of the data center pair is. is the comprehensive correlation coefficient of the data center pairs P and Q, represents a synergy term.

[0121] When the linkage load index of the high-correlation center pair group is greater than the linkage load threshold , it can be determined that the real-time traffic load of the high-correlation center pair group meets the scheduling condition.

[0122] In this embodiment, the linkage load index of the high-correlation center pair group is calculated by the above-mentioned method, which can comprehensively consider the situation that the real-time traffic load of a single data center pair in the high-correlation center pair group is excessive, and the situation that the real-time traffic loads of two data center pairs are synergistically excessive. When the real-time traffic load of the data center pair P is severely excessive (the value is large), even if the real-time traffic load of the data center pair Q is normal (the value is 0), the linkage load index can still exceed the threshold to trigger scheduling due to the too large value. When the real-time traffic loads of the two data center pairs are slightly excessive (the values are small), but the correlation between the two data center pairs is high (the value is large), the synergy term can amplify the risk and make the value of the linkage load index larger to trigger scheduling.

[0123] ​​​​​​​​In this embodiment, by designing the above-mentioned correlation-driven linkage load index calculation model, the traffic load monitoring process of two data center pairs in the high-risk center pair group can be linked together. Based on the real-time traffic load of the data center pairs and the comprehensive correlation coefficient, the linkage load index of the two data center pairs can be calculated. The risk of an individual can be captured by calculating the load excess degree of a single data center pair (single pair deviation degree), and the linkage risk can be captured by calculating the correlation synergy term of the two data center pairs, thereby achieving effective early warning of traffic load risk.

[0124] In one exemplary embodiment, such as Figure 5 As shown, a traffic scheduling method for multiple data centers is provided, including the following steps:

[0125] Step S501: Multi-dimensional data acquisition and processing of data center interconnection network.

[0126] Specifically, this step can obtain the traffic time-series data and path topology data of each data center pair in the network system. The path topology data can include the node set data and link set data of the interconnection path between data center pairs.

[0127] Step S502: The data center dynamically quantifies and classifies the correlation.

[0128] In this step, the collected traffic time-series data and path topology data of data center pairs can be used as input to quantify the correlation and topological coupling of traffic changes between each pair of data center pairs in the system. This yields the traffic change correlation coefficient and topological coupling coefficient between the two data center pairs. The correlation between the two data center pairs is then comprehensively quantified by weighted summation of these two coefficients, resulting in a comprehensive correlation coefficient for each pair of data center pairs. Subsequently, highly correlated data center pairs can be identified from among multiple data center pairs based on the comprehensive correlation coefficient.

[0129] The correlation coefficient of traffic changes between the two data center pairs can be expressed as:

[0130]

[0131] In the formula, For time window, This is the time-series traffic data from the data center to P. [This refers to the time offset of Q in the data center] Traffic time series data below, This represents the traffic data of data center P at time i. This indicates that the data center's response to Q at the i+th time... Real-time traffic data Let be the dynamic weight at time i. The calculation formula can be expressed as: , is the attenuation coefficient.

[0132] The topology coupling coefficient of the two data center pairs can be expressed as:

[0133]

[0134] In the formula, is the topology coupling coefficient, is the topology value distribution coefficient, and are the node sets of the data center pairs P and Q respectively, is the shared node set of the data center pairs P and Q, is the shared node in , is the node weight of node , which can be obtained by calculating the betweenness centrality index of node , and are the link sets of the data center pairs P and Q respectively, is the shared link set of the data center pairs P and Q.

[0135] The comprehensive correlation coefficient of the two data center pairs can be obtained by weighted summation of the traffic change correlation coefficient and the topology coupling coefficient of the two data center pairs:

[0136]

[0137] In the formula, is the comprehensive correlation coefficient of the data center pair P and the data center pair Q, is the traffic change correlation coefficient, is the topology coupling coefficient, is the comprehensive correlation weight coefficient, which can be set according to the emphasis of topology correlation or traffic correlation, is the hierarchical threshold parameter.

[0138] When the comprehensive correlation coefficient of the two data center pairs is not less than the hierarchical threshold parameter , it can be determined that the two data center pairs are high correlation center pair groups; and when the comprehensive correlation coefficient of the two data center pairs is less than the hierarchical threshold parameter, it can be determined that the two data center pairs are low correlation center pair groups.

[0139] Step S503, data center interconnection network traffic collaborative scheduling strategy configuration.

[0140] The higher the comprehensive correlation degree of the two data center pairs indicates the more significant traffic synchronous fluctuation and the tighter path coupling, and thus the higher the collaborative congestion risk of the two data center pairs. When the two data center pairs simultaneously generate high traffic load, the overlapped underlying link is prone to bearing impact.

[0141] In this step, the network operation system can configure a data center interconnection networking traffic collaborative scheduling strategy according to the calculation results and hierarchical conditions of the comprehensive correlation coefficient of each two data center pairs, realize the upgrade from "isolated monitoring of single pair traffic" to "correlation analysis of multiple pairs of collaborative decision-making", and accurately judge whether joint traffic scheduling is needed. When the two data center pairs are low-correlation center pair groups, there is no need to implement joint scheduling strategy, avoiding resource waste; when the two data center pairs are high-correlation center pair groups, the traffic load monitoring process of the two data center pairs can be bound to each other, and the linkage load index of the two data center pairs can be calculated based on real-time traffic load and comprehensive correlation coefficient.

[0142] Specifically, the linkage load index of the high-correlation center pair group can capture individual risk by calculating the load overage index of a single data center pair, and capture linkage risk by calculating the correlation collaborative term of the two data center pairs, realizing the early warning goal of "both focusing on individual abnormalities and not missing slight collaborative abnormalities under high correlation". By monitoring the maximum traffic load of the high-correlation center pair group, the risk point that is most likely to trigger collaborative fluctuation can be captured in time, which serves as the trigger basis for joint scheduling, ensuring that the collaborative strategy is started when the traffic load of any data center pair in the high-correlation center pair group exceeds the threshold, thereby preventing congestion from spreading and avoiding synchronous congestion of multiple high-correlation links. Exemplarily, the linkage load index of the high-correlation center pair group can be calculated by the following formula:

[0143]

[0144]

[0145] In the formula, is the linkage load index of the high-correlation center pair group; is the center pair weight of the data center pair P, is the center pair weight of the data center pair Q, , can reflect the business importance of the two data center pairs, and the sum of the two is 1. The larger the value is, the higher the business importance of the data center pair is; is the comprehensive correlation coefficient of the data center pairs P and Q, represents the collaborative term, , respectively represent the load overage index of the two data center pairs P and Q, , respectively represent the real-time traffic load of two data center pairs P, Q, , respectively represent the traffic warning thresholds of two data center pairs P, Q, which can be set according to historical fault data or specific business requirements.

[0146] wherein when the linkage load index of the high-correlation center pair group exceeds the linkage load threshold , it can be determined that the real-time traffic load of the high-correlation center pair group meets the scheduling condition. The value of the linkage load threshold can be set according to historical fault data.

[0147] Step S504, data center interconnection network traffic collaborative scheduling monitoring and implementation.

[0148] wherein the network operation and maintenance system can continuously monitor and record the traffic load between data centers, and determine whether to trigger traffic collaborative scheduling according to the traffic collaborative scheduling strategy configured in the foregoing steps. If the linkage load index of the high-correlation center pair group exceeds the linkage load threshold, traffic scheduling can be performed on the two data center pairs in the high-correlation center pair group through optimization means such as traffic engineering, backup underlying path switching, reserved bandwidth configuration, etc., to relieve network congestion of the data center interconnection and underlying bearing network.

[0149] In this embodiment, the correlation of two data center pairs is quantified for traffic scheduling of a multi-data center network system, and a collaborative decision-making mode for data center interconnection traffic scheduling is proposed. In the correlation quantification of the data center pair, the traffic change correlation between the two data center pairs is calculated by considering the traffic timing dynamics and timeliness of the two data center pairs, and the importance sum of the shared nodes between the two data center pairs and the overlapping link association risk are considered to calculate the topology coupling degree between the two data center pairs. Then, the comprehensive correlation coefficient of the two data center pairs is constructed through dynamic weight, and the collaborative quantification of business dynamic fluctuation and physical topology constraint is realized. This model can break through the limitation of analyzing traffic or topology correlation in only one dimension in related technologies, and can effectively identify the correlation relationship of the data center pair, providing more comprehensive decision basis for subsequent scheduling. In the scheduling decision, the real-time traffic load of the high-correlation center pair group is monitored, and a linkage load index calculation model of the high-correlation center pair group is designed, which can integrate the two risk factors of the single data center pair load exceeding degree and the collaborative state of the two data center pairs under high correlation, and combine the traffic load change process of the high-correlation center pair group to form a dynamic traffic collaborative scheduling strategy. This mechanism can dynamically bind the load monitoring process of the high-correlation center pair group to realize traffic scheduling collaborative decision-making, which is conducive to coping with the dynamic changes of network and business state, and controlling the congestion diffusion risk of associated links. Compared with the method of related technologies based on static threshold and single-link scheduling, the method in this embodiment can improve the timeliness and accuracy of traffic regulation, which is conducive to reducing the probability of business interruption.

[0150] It should be understood that, although each step in the flowchart involved in each embodiment as described above is displayed 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 embodiment as described above can include multiple steps or 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. It can be understood that the steps in different embodiments can be freely combined as needed, and various non-contradictory schemes formed by combination are within the scope of protection of the present application.

[0151] Based on the same inventive concept, the embodiments of the present application further provide a multi-data-center traffic scheduling device for implementing the multi-data-center traffic scheduling 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 multi-data-center traffic scheduling device embodiments provided below can refer to the limitations of the multi-data-center traffic scheduling method described above, which will not be repeated here.

[0152] In one exemplary embodiment, as shown in Figure 6 a multi-data-center traffic scheduling device is provided, comprising:

[0153] a data acquisition module 601 configured to acquire traffic time series data and path topology data of each data center pair;

[0154] an association analysis module 602 configured to obtain a comprehensive association coefficient of each data center pair according to the association of the traffic time series data and the path topology data of each data center pair;

[0155] a center pair combination module 603 configured to determine a high-association center pair group from a plurality of data center pairs according to the comprehensive association coefficient, and monitor real-time traffic load of each high-association center pair group;

[0156] a combination scheduling module 604 configured to perform traffic scheduling on each data center pair in the high-association center pair group if the real-time traffic load of the high-association center pair group meets a scheduling condition.

[0157] In one exemplary embodiment, the association analysis module 602 is configured to: calculate a traffic change correlation coefficient of two data center pairs according to the traffic time series data of the two data center pairs; calculate a topological coupling coefficient of the two data center pairs according to the path topology data of the two data center pairs; and obtain a comprehensive association coefficient of the two data center pairs according to the traffic change correlation coefficient and the topological coupling coefficient.

[0158] In one exemplary embodiment, the association analysis module 602 is configured to: calculate a time-lag correlation coefficient of two data center pairs at a plurality of time offsets respectively according to the traffic time series data of the two data center pairs; and obtain the traffic change correlation coefficient of the two data center pairs according to the time-lag correlation coefficient corresponding to each time offset.

[0159] In an exemplary embodiment, the correlation analysis module 602 is configured to: obtain a node set and a link set of each of the pairs of data centers according to the path topology data of the pairs of data centers; obtain a node coupling coefficient of each of the pairs of data centers according to a set size of the node set and a shared node between the node sets; obtain a link coupling coefficient of each of the pairs of data centers according to a set size of the link set and a shared link between the link sets; and obtain the topology coupling coefficient according to the node coupling coefficient and the link coupling coefficient.

[0160] In an exemplary embodiment, the correlation analysis module 602 is configured to: obtain a node weight of each of the shared nodes according to a betweenness centrality index of each of the shared nodes in the network; and obtain the node coupling coefficient according to the set size of the node set and the node weight of each of the shared nodes.

[0161] In an exemplary embodiment, the apparatus further comprises: an excess analysis module configured to obtain a load excess index of each of the pairs of data centers according to the real-time traffic load of each of the pairs of data centers in the group of high-correlation pairs of data centers; a linkage analysis module configured to calculate a linkage load index of the group of high-correlation pairs of data centers according to the load excess index of each of the pairs of data centers, a pair weight of each of the pairs of data centers, and the comprehensive correlation coefficient between the pairs of data centers; and a condition determination module configured to determine that the real-time traffic load of the group of high-correlation pairs of data centers satisfies a scheduling condition when the linkage load index is greater than a linkage load threshold.

[0162] Each of the above modules of the traffic scheduling apparatus for multiple data centers can be implemented in whole or in part by software, hardware, and combinations thereof. Each of the above modules can be embedded in or independent of a processor in a network device in hardware form, or can be stored in a memory in the network device in software form, so as to be called and executed by a processor to perform operations corresponding to each of the above modules.

[0163] In an exemplary embodiment, a network device, which can be a server, is provided, and an internal structure diagram of the network device can be as shown in FIG. 2. Figure 7As shown in the figure. The network device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through the system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the network device is used to provide computing and control capability. The memory of the network device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the network device is used to store the traffic time series data and path topology data of each data center pair. The input / output interface of the network device is used to exchange information between the processor and external devices. The communication interface of the network device is used to communicate with the terminal outside through the network connection. The computer program is executed by the processor to realize a traffic scheduling method for multiple data centers.

[0164] Those skilled in the art can understand that, Figure 7 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 network device to which the scheme of the present application is applied. The specific network device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0165] In one embodiment, a network device is also provided, including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to realize the steps in each of the above method embodiments.

[0166] 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 realize the steps in each of the above method embodiments.

[0167] In one embodiment, a computer program product is provided, including a computer program, and the computer program is executed by a processor to realize the steps in each of the above method embodiments.

[0168] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0169] 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.

[0170] The technical features of the above embodiments can be combined in any manner. To make the description concise, not all possible combinations of the technical features in the above embodiments are described, but as long as the combinations of the technical features do not exist contradictions, they should be considered as the scope of the present application.

[0171] 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 traffic scheduling method for multiple data centers, characterized in that, The method comprises: obtaining traffic time series data and path topology data of each data center pair; obtaining a comprehensive correlation coefficient of each data center pair according to the correlation of the traffic time series data and the path topology data of each two data center pairs; determining a high-correlation center pair group in a plurality of data center pairs according to the comprehensive correlation coefficient, monitoring real-time traffic load of each high-correlation center pair group; in the case that the real-time traffic load of the high-correlation center pair group meets the scheduling condition, performing traffic scheduling on each data center pair in the high-correlation center pair group.

2. The method of claim 1, wherein, The comprehensive correlation coefficient of each data center pair is obtained according to the correlation of the traffic time series data and the path topology data of each two data center pairs, comprising: calculating a traffic change correlation coefficient of two data center pairs according to the traffic time series data of the two data center pairs; calculating a topology coupling coefficient of two data center pairs according to the path topology data of the two data center pairs; obtaining a comprehensive correlation coefficient of two data center pairs according to the traffic change correlation coefficient and the topology coupling coefficient.

3. The method of claim 2, wherein, The traffic change correlation coefficient of two data center pairs is calculated according to the traffic time series data of the two data center pairs, comprising: calculating a time delay correlation coefficient of the traffic time series data of two data center pairs under a plurality of time offsets respectively; obtaining the traffic change correlation coefficient of the traffic time series data of two data center pairs according to the time delay correlation coefficient corresponding to each time offset.

4. The method of claim 2, wherein, The topology coupling coefficient of two data center pairs is calculated according to the path topology data of the two data center pairs, comprising: obtaining a node set and a link set of two data center pairs according to the path topology data of the two data center pairs; obtaining a node coupling coefficient of two data center pairs according to the set size of the node set and the shared nodes between the node set; obtaining a link coupling coefficient of two data center pairs according to the set size of the link set and the shared links between the link set; obtaining the topology coupling coefficient according to the node coupling coefficient and the link coupling coefficient.

5. The method of claim 4, wherein, The node coupling coefficient of two data center pairs is obtained according to the set size of the node set and the shared nodes between the node set, comprising: obtaining a node weight of each shared node according to the betweenness centrality index of each shared node in the network; obtaining the node coupling coefficient according to the set size of the node set and the node weight of each shared node.

6. The method according to any one of claims 1 to 5, characterized in that, Before the real-time traffic load of the high-correlation center pair group meets the scheduling condition, the method further comprises: obtaining a load excess index of each data center pair according to the real-time traffic load of each data center pair in the high-correlation center pair group; According to the load excess indexes of each data center pair, and the center pair weight of each data center pair and the integrated correlation coefficient between the data center pair, a linkage load index of the high-correlation center pair group is calculated; In a case where the linkage load index is greater than a linkage load threshold, it is determined that the real-time traffic load of the high-correlation center pair group meets a scheduling condition.

7. A multi-data center traffic scheduling apparatus, characterized by, The device comprises: a data acquisition module configured to acquire traffic time series data and path topology data of each data center pair; a correlation analysis module configured to obtain an integrated correlation coefficient of each two data center pairs according to the correlation of the traffic time series data and the path topology data of each two data center pairs; a center pair combination module configured to determine a high-correlation center pair group from a plurality of data center pairs according to the integrated correlation coefficient, and monitor real-time traffic load of each high-correlation center pair group; a group scheduling module configured to perform traffic scheduling on each data center pair in the high-correlation center pair group in a case where the real-time traffic load of the high-correlation center pair group meets a scheduling condition. 8.A network device, comprising a memory and a processor, wherein the memory stores a computer program, and the network device is configured to perform the method according to any one of claims 1-7. The processor executes the computer program to implement the steps of the method of 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 of 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 of any one of claims 1 to 6.

Images