Resource allocation method and system for dynamic load balancing of elastic optical network of data center

Abstract

The invention relates to a resource allocation method and system for dynamic load balancing of a data center elastic optical network, and belongs to the technical field of communication. Comprising the following steps: initializing data center elastic optical network parameters, and generating a connection request; judging whether the residual computing power resources of the source node and the destination node meet the request requirement or not, if so, calculating K working paths, selecting the shortest path, calculating the number of required spectrum slots, and judging whether the paths have residual spectrum resources or not; if yes, layered traffic grooming evaluation parameters are calculated, and the optimal spectrum layer is selected for grooming traffic; if not, selecting the highest-price modulation format; judging whether bandwidth and optical channel resources on the path are enough, if so, allocating spectrum resources, otherwise, returning to select the next shortest path; judging whether the spectrum resources meet constraint conditions or not, if yes, allocating computing power resources, and requesting successfully; if not, returning to reselect the path; and after success, updating the network transmission path weight. According to the invention, the network resource waste and blocking rate are effectively reduced, and the optical network energy consumption is reduced.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and in particular to a resource allocation method and system for dynamic load balancing in a data center elastic optical network. Background Technology

[0002] With the widespread application of cloud computing technology and internet services, the demand for network bandwidth from emerging applications has exploded, leading to near-saturation of bandwidth resources in communication networks. The increase in data traffic has not only caused spectrum resource shortages but also resulted in a significant increase in network energy consumption. In Elastic Optical Data Center Networks (EODCNs), the rational allocation of spectrum resources is crucial. However, when spectrum resources are allocated irrationally in an optical network, numerous problems arise, such as wasted spectrum resources, increased network congestion rates, and an increase in the number of energy-consuming network components (such as optical transponders and optical regenerators), ultimately leading to increased energy consumption across the entire optical network and unnecessary waste of resources.

[0003] Traditional load balancing (CLB) and traffic routing methods only consider the proportion of spectrum slots occupied. This singular approach limits their effectiveness in improving load balancing, energy efficiency, and resource utilization. Furthermore, traditional routing and spectrum allocation methods fail to efficiently utilize network bandwidth resources, leading to an increase in the number of energy-consuming devices operating in the network, further exacerbating resource waste. In traditional algorithms, the bandwidth demand of a connection request may be routed to optical channels with mismatched time or frequency domains. This not only reduces the proportion of time available for traffic routing but also wastes remaining bandwidth resources, ultimately increasing the number of energy-consuming optical layer devices required for service transmission and thus increasing the overall network energy consumption. To improve resource utilization efficiency and reduce network transmission energy consumption, it is necessary to minimize the number of optical repeaters and optical regenerators used during service transmission.

[0004] In data center elastic optical network transmission, since connection request traffic varies in size, traffic routing methods can effectively reduce network bandwidth waste by routing services from multiple originating nodes onto a single optical path, thereby saving network resources and reducing the occupancy of energy-intensive components. However, traditional traffic routing methods still have significant limitations in improving energy efficiency and resource utilization.

[0005] Traditional load balancing methods employ a one-size-fits-all approach to spectrum occupancy, using thresholds to determine bandwidth usage in elastic optical networks (OORNs). While this method considers the current spectrum load in the OORN and improves the success rate of spectrum allocation after KSP routing by updating virtual link weights, it only considers the occupancy ratio of spectrum slots. It neglects the fact that traffic routing for connection requests depends not only on bandwidth usage but also on the availability of optical channels that meet the routing conditions. Therefore, load balancing in OORNs requires a comprehensive consideration of the remaining spectrum resources within all optical channels on the fiber, rather than just overall bandwidth usage. Traditional load balancing methods lack precision in their spectrum considerations. In situations with high load but ample remaining spectrum resources in optical channels, they may incorrectly update link weights, hindering traffic routing for connection requests. Although traditional load balancing methods can balance traffic to some extent, their optimization effect is often less than ideal in OORNs where traffic routing is required.

[0006] In the elastic optical network of a data center, to cope with the ever-increasing demand for computing resources, it is necessary to connect the suppliers and consumers of computing resources through the optical network to achieve unified management and scheduling of computing resources. However, since the number of servers in a real data center is limited, the computing and storage resources that can be provided are also limited, and therefore the computing resources at the data center nodes are also limited. When a connection request arrives, the computing power requirement must be met first. When each connection request arrives, it is necessary to first check whether the remaining computing power resources can meet the computing power requirements of the connection request. Only when the computing power constraints are met will subsequent steps be taken, namely, establishing a working path through routing selection and allocating spectrum resources. Summary of the Invention

[0007] Therefore, the technical problem to be solved by the present invention is to overcome the problems of unreasonable spectrum resource allocation, low resource utilization efficiency, excessive energy consumption and increased network congestion rate in the prior art.

[0008] In a first aspect, to solve the above-mentioned technical problems, the present invention provides a resource allocation method for dynamic load balancing of elastic optical networks in data centers, comprising: S1. Obtain the network parameters of the data center's elastic optical network; generate a connection request based on the network parameters; wherein the connection request includes a source node and a destination node; determine whether the remaining computing resources at the source node and the destination node are both greater than or equal to the computing resources required by the connection request; if yes, proceed to S2; if no, the connection request fails to establish. S2. Calculate K working paths, select the shortest path from the K working paths, and calculate the number of spectrum slots required for the connection request; S3. Based on the number of spectrum slots, determine whether there are remaining spectrum resources on the transmission path of the connection request that satisfy traffic routing; if yes, proceed to S4; if no, do not perform traffic routing, select the highest-priced modulation format that can be transmitted, and proceed to S5. S4. Calculate the hierarchical flow diversion evaluation parameters and select the spectrum layer with the smallest hierarchical flow diversion evaluation parameters for flow diversion. S5. Determine whether there are sufficient bandwidth and optical channel resources on the transmission path of the connection request. If yes, allocate spectrum resources to the connection request; otherwise, return to S2 and select the shortest path from the remaining working paths. S6. Determine whether the spectrum resources meet the consistency and continuity constraints; if yes, allocate computing resources to the connection request and calculate the number of optical repeaters and optical regenerators required for the connection request, and the connection request is successfully established; if no, return to S2. S7. After the connection request is successfully established, update the spectrum resources and computing power resources of the connection request, and update the network transmission path weight of the connection request.

[0009] In one embodiment of the present invention, step S7, updating the network transmission path weight of the connection request, includes calculating the network transmission path weight, and the expression for calculating the network transmission path weight is: ; in, This represents the network transmission path weight, and β2 represents an adjustable parameter. This represents the total number of optical channels between nodes m and n. The initial weight value represents the distance of the network transmission path. This indicates the number of optical channels whose bandwidth usage exceeds the adjustable parameter, where C represents the set of optical channels. This indicates the number of spectrum slots in the optical fiber that have been occupied by connection requests. This represents the total number of spectral gaps in the optical fiber between two optical switching nodes m and n.

[0010] In one embodiment of the present invention, step S2, the method for calculating the number of spectrum slots required for the connection request, is as follows: calculate the product of the spectral efficiency value required by the modulation format and the bandwidth of each spectrum slot, and obtain the number of spectrum slots required for the connection request based on the product and the bandwidth required by the connection request.

[0011] In one embodiment of the present invention, step S3, which determines whether there are remaining spectrum resources on the transmission path of the connection request that meet the requirements for traffic diversion based on the number of spectrum slots, includes spectrum layering processing.

[0012] In one embodiment of the present invention, the method of spectrum layering is as follows: traverse the occupied optical channels on the working path, obtain the optical channels with remaining bandwidth resources that meet the transmission bandwidth requirements of the connection request, and obtain the spectrum layer.

[0013] In one embodiment of the present invention, step S4, calculating the expression for the stratified flow diversion evaluation parameters, is as follows: ; in, This represents the parameters for evaluating stratified flow distribution. Represents the spectral residual coefficients. Indicates the number of the spectral layer; This represents the time overlap coefficient.

[0014] In one embodiment of the present invention, step S6, calculating the number of optical repeaters and optical regenerators required for the connection request, is as follows: Based on the source and destination nodes in the selected working path, a temporary topology for calculating the number of optical regenerators is rebuilt. Traverse all path selections in the temporary topology, select the path with the lowest total energy consumption, analyze the usage of optical regenerators at all nodes of each path in turn, perform traffic diversion for optical regenerators with remaining spectrum resources, reconfigure optical regenerators at nodes where diversion is not possible, and calculate the number of optical regenerators required for all paths.

[0015] Secondly, to solve the above-mentioned technical problems, the present invention provides a resource allocation system for dynamic load balancing of elastic optical networks in data centers, used to implement the above-mentioned method, including: The network initialization module is used to obtain the network parameters of the data center elastic optical network and initialize the network parameters; A connection request generation module is used to generate connection requests; wherein the connection request includes a source node and a destination node; The computing power resource judgment module is used to determine whether the remaining computing power resources at the source node and the destination node are both greater than or equal to the computing power resources required by the connection request; if yes, then proceed to the working path establishment module; if no, then the connection request establishment fails. The working path creation module is used to calculate K working paths and select the shortest path from the K working paths; The hierarchical traffic routing module is used to calculate the number of spectrum slots required for the connection request based on the shortest path; determine whether there are remaining spectrum resources on the transmission path of the connection request that satisfy traffic routing based on the number of spectrum slots; if so, calculate the hierarchical traffic routing evaluation parameters, select the spectrum layer with the smallest hierarchical traffic routing evaluation parameters for traffic routing; if not, do not perform traffic routing, select the highest-priced modulation format that can be transmitted, and proceed to the decision and warning module. The decision and warning module is used to determine whether there are sufficient bandwidth and optical channel resources on the transmission path of the connection request. If so, spectrum resources are allocated to the connection request; if not, the process returns to the working path establishment module, which selects the shortest path from the remaining working paths. The module also determines whether the spectrum resources meet the consistency and continuity constraints. If so, computing power resources are allocated to the connection request, and the process proceeds to the data processing and computing module; if not, the process returns to the working path establishment module. The data processing and calculation module is used to calculate the number of optical repeaters and optical regenerators required for the connection request; The computing power resource update module is used to update the spectrum resources and computing power resources of the connection request after the connection request is successfully established. The dynamic load balancing module is used to update the network transmission path weight of the connection request.

[0016] In one embodiment of the present invention, a resource release module is further included, which is used to release the spectrum resources occupied by the working path after the connection request is successfully established, and to release the occupied optical transponder and optical regenerator resources.

[0017] In one embodiment of the present invention, the data processing and calculation module is further configured to calculate the energy consumption of each connection request after each connection request is successfully established, and to calculate the evaluation index of the connection request.

[0018] Compared with the prior art, the above-described technical solution of the present invention has the following advantages: (1) The resource allocation method and system for dynamic load balancing of elastic optical networks in data centers described in this invention evaluates the remaining computing resources of source and destination nodes to determine whether they meet the requirements of connection requests, thereby effectively avoiding resource waste and blind establishment of connection requests, and improving the utilization rate of network resources. At the same time, selecting the shortest path as the working path can reduce data transmission latency, improve network response speed, and reduce energy consumption. In addition, this invention calculates the number of spectrum slots required for connection requests to determine whether there are remaining spectrum resources that meet the traffic routing requirements, further optimizing the utilization efficiency of spectrum resources and ensuring efficient network operation. By calculating the hierarchical traffic routing evaluation parameters, the optimal spectrum layer is selected for traffic routing, which not only improves the utilization rate of spectrum resources but also effectively reduces the network congestion rate and improves the overall performance of the network. This invention ensures that connection requests can be successfully established by comprehensively judging the sufficiency of bandwidth and optical channel resources, as well as whether spectrum resources meet the consistency and continuity constraints, thereby ensuring the stability and reliability of the network. After a connection request is successfully established, this invention dynamically updates spectrum resources, computing resources, and network transmission path weights to achieve dynamic adjustment and optimization of network resources, further improving the overall performance and adaptability of the network and enabling it to better cope with ever-changing network demands.

[0019] (2) This invention addresses the problem that traditional traffic diversion methods do not fully consider the time-domain and frequency-domain compatibility during the diversion process. It comprehensively considers the characteristics of the time dimension, the duration of the service flow, and the actual situation of the remaining spectrum resources in the optical channel. Based on this, a diversion spectrum layer is established and hierarchical traffic diversion evaluation parameters are calculated, and a hierarchical traffic diversion method is adopted. Through this method, the number of optical channels required in the network can be effectively reduced, thereby reducing the number of energy-consuming components in operation, and ultimately achieving a significant reduction in network energy consumption. Attached Figure Description

[0020] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings, wherein: Figure 1 This is a flowchart of a resource allocation method for dynamic load balancing in a data center elastic optical network according to a preferred embodiment of the present invention. Figure 2 This is a diagram of the hierarchical traffic routing network structure for dynamic load balancing in a data center elastic optical network according to a preferred embodiment of the present invention. Figure 3 This is a structural diagram of a resource allocation system for dynamic load balancing in a data center elastic optical network according to a preferred embodiment of the present invention. Detailed Implementation

[0021] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0022] Example 1: Reference Figure 1 As shown, this embodiment of the invention provides a resource allocation method for dynamic load balancing in a data center elastic optical network, including but not limited to the following steps: S1. Obtain the network parameters of the data center's elastic optical network; generate a connection request based on the network parameters; the connection request includes a source node and a destination node; determine whether the remaining computing resources at the source node and the destination node are both greater than or equal to the computing resources required for the connection request; if yes, proceed to S2; otherwise, the connection request fails to establish. S2. Calculate K working paths, select the shortest path from the K working paths, and calculate the number of spectrum slots required for the connection request; S3. Based on the number of spectrum slots, determine whether there are remaining spectrum resources on the transmission path of the connection request that satisfy traffic smoothing; if yes, proceed to S4; if no, do not perform traffic smoothing, select the highest-priced modulation format that can be transmitted, and proceed to S5; S4. Calculate the hierarchical flow diversion evaluation parameters and select the spectrum layer with the smallest hierarchical flow diversion evaluation parameters for flow diversion. S5. Determine if there are sufficient bandwidth and optical channel resources on the transmission path of the connection request. If yes, allocate spectrum resources to the connection request; otherwise, return to S3 and select the shortest path from the remaining working paths. S6. Determine whether the spectrum resources meet the consistency and continuity constraints; if yes, allocate computing resources to the connection request and calculate the number of optical transponders and optical regenerators required for the connection request, and the connection request is successfully established; if no, return to S2. S7. After the connection request is successfully established, update the spectrum resources and computing resources of the connection request, and update the network transmission path weight of the connection request.

[0023] This invention provides a resource allocation method for dynamic load balancing in a data center's elastic optical network. By introducing hierarchical traffic routing evaluation parameters, link weights are dynamically updated based on optical channel occupancy, achieving load balancing of services between the optical fiber and the data center. This efficiently utilizes remaining optical channel bandwidth resources, reduces network congestion rates, optimizes spectrum resources, and reduces energy consumption. The method evaluates the remaining computing resources of source and destination nodes to determine if they meet connection request requirements, effectively avoiding resource waste and blind connection request establishment, thus improving network resource utilization. The shortest path is selected as the working path to reduce data transmission latency, improve network response speed, and reduce energy consumption. Furthermore, by calculating the number of spectrum slots required for a connection request, it determines whether there are remaining spectrum resources sufficient for traffic routing, further optimizing spectrum resource usage and ensuring efficient network operation. Hierarchical traffic routing evaluation parameters are calculated, and the optimal spectrum layer is selected for traffic routing, improving spectrum resource utilization, reducing network congestion rates, and enhancing network performance. This method also ensures successful connection request establishment and guarantees network stability and reliability by assessing the sufficiency of bandwidth and optical channel resources, and whether spectrum resources meet consistency and continuity constraints. After a connection request is successfully established, the spectrum resources, computing resources, and network transmission path weights are updated to achieve dynamic adjustment and optimization of network resources, thereby improving the overall network performance and adaptability.

[0024] In a data center's resilient optical network, both computing power and spectrum resources are finite, and their allocation must meet the following constraints: First, the computing power provided by the servers in the data center nodes must be greater than or equal to the computing power required for connection requests; second, the spectrum resources in the fiber optic links are limited, and their allocation must simultaneously satisfy the constraints of spectrum consistency and spectrum continuity. Furthermore, traffic routing must also meet the following constraints: First, the remaining bandwidth in the optical channel must be greater than or equal to the bandwidth required for routing connection request traffic; second, the modulation format of existing connection requests in the optical channel and newly routed connection requests into the network must be the same.

[0025] To effectively improve resource efficiency and network energy efficiency, the number of optical repeaters and optical regenerators should be minimized. Furthermore, switching idle optical power-consuming components to sleep mode is an effective way to reduce network energy consumption. This is because, when there is no data traffic, the power consumption of optical power-consuming components in sleep mode is negligible.

[0026] Furthermore, the operating energy consumption of optical power-consuming components includes independent energy consumption and non-independent energy consumption. Independent energy consumption refers to the energy consumption generated by bandwidth resources, while non-independent energy consumption is the inherent energy consumption of the power-consuming component during operation.

[0027] Specifically, in step S1, the specific steps for initializing the data center elastic optical network and generating a set of connection request sets are as follows: S110. In the data center elastic optical network, initialize the computing resources of the data center servers and initialize the elastic optical network. In the data center elastic optical network G(N, L, E, F), a set of optical switching nodes uses N={n1, n2, n3,…, n…} |N| Let L = {l1, l2, l3, ..., l} represent the total number of optical switching nodes; a set of fiber optic links is represented by L = {l1, l2, l3, ..., l}. |L|} represents the total number of fiber optic links; the set of computing resources at each data center node is represented by E={e1, e2, e3,…, e |E|} represents the total number of data center nodes; the set of available spectrum slots in each fiber is represented by F={f1, f2, f3,…, f |F| Let} represent the total number of spectral slots in the optical fiber between the two optical switching nodes, where |F| represents the total number of spectral slots in the optical fiber. Furthermore, let (m, n) represent the optical fiber link from node m to node n, where m, n ∈ N.

[0028] S120. Generate a set of connection requests CR, where each connection request CR(s, d, ts, td, es, ed, BR)∈CR. Here, s represents the source node, d represents the destination node, ts represents the arrival time of the connection request, td represents the departure time of the connection request, es represents the computing power resources required by the data center server at the source node, ed represents the computing power resources required by the data center server at the destination node, and BR represents the bandwidth requirement of the connection request.

[0029] Furthermore, for each connection request CR(s, d, ts, td, es, ed, BR), it is first determined whether the data center servers at the source node s and the destination node d have sufficient computing resources. The specific criterion is whether the remaining computing resources at the data center nodes are greater than or equal to the computing resources required for the connection request. If the remaining computing resources at the source node s and the destination node d are greater than or equal to es and ed respectively, then proceed to the next step (i.e., step S2); otherwise, the connection request is blocked, indicating that the connection request has failed to establish.

[0030] Specifically, in step S2, a route is selected for the connection request CR(s, d, ts, td, es, ed, BR). A K-shortest path algorithm is used for route selection, calculating K possible route selection schemes. The route with the smallest sequence number is prioritized to determine the working path. If the working path is successfully established, the next step is performed; otherwise, the connection request is blocked.

[0031] Further, in step S2, the number of spectral slots required for the connection request CR(s, d, ts, td, es, ed, BR) is calculated based on the selected working path. The method for calculating the number of spectral slots required for the connection request is as follows: calculate the product of the required spectral efficiency value of the modulation format and the bandwidth of each spectral slot; based on this product and the bandwidth required for the connection request, the number of spectral slots required for the connection request is obtained.

[0032] For example, the number of spectral slots is equal to the bandwidth required by the connection request divided by the spectral efficiency value required by the modulation format and the bandwidth size of each spectral slot.

[0033] Further, the specific steps for calculating the transmission distance of this working path are as follows: First, collect information on each link on the path. This includes creating a link list, recording the node pairs connected by each link, and the physical length of each link. If the link length is directly known, it is recorded; if unknown, it needs to be obtained through measurement. Second, calculate the length of each link. For each link, its physical length needs to be determined, which includes considering the fiber optic laying path and any possible bends or backslidings to ensure accurate link length data. Finally, sum the path lengths by adding the lengths of all links on the path to obtain the total physical length of the entire working path.

[0034] In this embodiment of the invention, the dynamic load balancing method for traffic routing, by accurately assessing the bandwidth occupancy within the optical channel, not only effectively achieves load balancing across the entire flexible optical transmission network but also improves the overall traffic routing effect. This method further reduces network energy consumption by more rationally reducing the number of ports on energy-consuming components, thus achieving energy saving and efficiency improvement.

[0035] To reduce the energy consumption of data center elastic optical networks and conserve spectrum resources, this embodiment of the invention adopts a spectrum stratification (SS) approach for traffic routing.

[0036] Furthermore, in steps S3 to S5, hierarchical traffic routing and spectrum resource allocation are performed. The specific steps are as follows: S310. For the transmission path of the connection request, perform spectrum layering processing on that path. Specifically, layer the idle spectrum resources of each optical fiber according to optical channels with different rates.

[0037] Furthermore, the method for spectrum layering is as follows: traverse the optical channels that are already occupied on the working path (also known as the routing path) and select the optical channels whose remaining bandwidth resources meet the transmission bandwidth requirements of the connection request, thereby forming a spectrum layer.

[0038] S320. Determine if there are any remaining spectrum resources on the transmission path of the connection request that satisfy traffic diversion. If no spectrum layer meets the conditions for diversion, diversion is not performed, spectrum resources are directly allocated, and subsequent steps continue. If no allocable spectrum resources are found under the current route, the routing method is changed, and the process returns to step S2 to select a new working path. If the connection request still cannot be successfully transmitted after trying K different routing methods, the connection request will be considered blocked.

[0039] This invention addresses the problem that traditional traffic routing methods do not adequately consider the time-domain and frequency-domain compatibility during the routing process. It comprehensively considers the characteristics of the time dimension, the duration of service flows, and the actual situation of remaining spectrum resources in optical channels. Based on this, a routing spectrum layer is established and hierarchical traffic routing evaluation parameters are calculated, employing a hierarchical traffic routing method. This method effectively reduces the number of optical channels required in the network, thereby reducing the number of energy-consuming components in operation and ultimately lowering network energy consumption.

[0040] Specifically, in step S4, each spectrum layer is evaluated, and the hierarchical traffic diversion evaluation parameters are calculated. In this embodiment of the invention, the spectrum layer with the smallest evaluation parameters is selected for traffic diversion. This ensures that service traffic is diverted to the spectrum layer with the greatest overlap in the frequency and time domains and the smallest spectrum label, thereby optimizing resource utilization.

[0041] Specifically, in step S5, the bandwidth and optical channel resources on the transmission path of the current connection request are assessed to determine whether to allocate spectrum resources for the current connection request. If there are insufficient bandwidth and optical channel resources, a new working path needs to be selected.

[0042] Furthermore, based on steps S3, S4, and S5, the specific steps of the Stratification Traffic Grooming (STG) method are as follows: Step 1: Divide the bandwidth resources of each optical fiber link along the transmission path according to the optical channel bandwidth, and mark the M optical channels along the transmission path as follows: .

[0043] Step 2: The bandwidth requirement of the connection request is BR. From the M optical channels, select optical channels with sufficient remaining bandwidth resources to handle the new service; that is, the remaining bandwidth resources of the optical channels are greater than or equal to BR. Based on these qualified optical channels, establish a C-layer spectrum set. .in, This represents the number of optical channels required for the i-th spectral layer. The generated spectral layers must satisfy the constraints of spectral continuity and consistency.

[0044] Step 3: In each spectral layer The arrival time of the original connection request in the k frequency slots is denoted as ts. i Departure time is recorded as td i The arrival time of a new connection request is denoted as t. s The departure time is denoted as t. d In the i-th spectral layer, the product of the proportions of the temporal overlap between two connection requests in the new connection request is defined as the temporal overlap coefficient O. i The calculation formula is as follows: (1) Among them O i The larger the value, the greater the overlap in connection request times when traffic is routed together.

[0045] Step 4: In each spectral layer In the original connection request, the remaining bandwidth resource after allocating spectrum resources is Y. i The bandwidth resources that a new connection request expects to channel in are denoted as Q. i In the i-th spectrum layer, the sum of the remaining bandwidth resources in the spectrum layer after new connection requests are channeled is defined as the spectrum surplus coefficient SR. i The calculation formula is as follows: (2) Among them SR i The smaller the value, the less bandwidth resources are available after the traffic is routed together. New connection requests will try to fill the entire spectrum layer's resources, meaning that new connection requests are more compatible with this optical channel in the frequency domain.

[0046] Step 5: Calculate the i-th spectral layer Mid-level flow diversion assessment parameters The calculation formula is as follows: (3) in, This represents the parameters for evaluating stratified flow distribution. Represents the spectral residual coefficients. Indicates the number of the spectral layer; This represents the time overlap coefficient.

[0047] According to the descriptions in formulas (1) and (2) above, the smaller the stratified flow diversion evaluation parameter, the better the flow diversion effect. Therefore, the stratified flow diversion evaluation parameter in each spectral layer is calculated. And select the stratified flow diversion assessment parameters. The spectral layer corresponding to the minimum value is de-stressed.

[0048] Specifically, in step S6, the number of optical repeaters and optical regenerators required for the connection request CR(s, d, ts, td, es, ed, BR) is calculated. For different modulation formats, the required optical regenerators are configured on the selected working path according to the maximum reachable distance between the optical regenerators. Thus, based on the source and destination nodes in the selected working path, a temporary topology for calculating the number of optical regenerators is re-established. On this working path, for any two node pairs, if the transmission distance of the nodes is less than the maximum transmission distance of the optical regenerators in the connection request, a connection link is established between these node pairs, and its weight is configured to be 1 unit length, thereby forming the temporary topology for calculating the number of optical regenerators.

[0049] Furthermore, in this temporary topology, all path selections are traversed, and the path with the lowest total energy consumption is selected. The specific steps are as follows: in these paths, the usage of optical regenerators at all nodes of each path is analyzed in turn; traffic is diverted to optical regenerators with remaining spectrum resources, and new optical regenerators are reconfigured on nodes that cannot be diverted; the number of optical regenerators required for all paths is calculated.

[0050] Specifically, in step S7, computing power resources and spectrum resources are allocated. Specifically, the required computing power resources and spectrum resources are allocated to the connection request CR. This involves rationally allocating the computing power resources of the data center nodes and the spectrum resources on the fiber optic link to the connection request. The specific steps are as follows: S710: Statistically analyze the current network status, including information such as resource usage and network load.

[0051] S720. Update the dynamic load balancing path weights in the network topology. In this embodiment of the invention, the dynamic load balancing method is implemented by real-time analysis of the resource occupancy of optical channels on each transmission path. When the bandwidth resource occupancy rate of an optical channel exceeds a preset threshold, a count is performed, and the counted number of optical channels is used as a key calculation parameter. Combined with link status information such as the initial link weight and the total number of optical channels in the link, a new link weight is calculated.

[0052] Furthermore, after each connection request is successfully established, the specific method for updating the link weights is as follows: traverse all optical fibers on the working route of the connection request, recalculate and update the weights to ensure that the link weights accurately reflect the actual state of the current network, thereby achieving efficient and dynamic load balancing. The expression for the link weights updated during dynamic load balancing is: (4) (5) (6) Where C represents the set of optical channels, This represents the total number of optical channels between nodes m and n. Represents a set of connection requests. This represents a connection request within a set of connection requests. This represents the total number of spectral gaps in the optical fiber between two optical switching nodes m and n. This represents the bandwidth capacity of optical channel i between nodes m and n. This indicates the bandwidth allocated and channeled in optical channel i for connection requests. This is a binary variable. If the optical channel numbered i is occupied by a connection request, The value is 1 if the condition is met, otherwise it is 0. Both α2 and β2 are adjustable parameters (0 < α2 < 1, β2 > 1). This represents the total number of spectral gaps in the optical fiber between two optical switching nodes m and n. This indicates the number of spectrum slots in the optical fiber that have been occupied by connection requests. This indicates the bandwidth occupancy ratio in optical channel i. Used to count the number of optical channels whose bandwidth usage exceeds α2. The weights updated for the links during dynamic load balancing. The initial weight value represents the distance of the network transmission path.

[0053] S730 updates the data center's elastic optical network status, including updating the resource allocation table and network topology. Based on this, it calculates the total energy consumption of the connection request, which helps assess the network's energy efficiency and cost-effectiveness.

[0054] S740. Determine if there are any unprocessed connection requests. If there are still undelivered connection requests, repeat steps S1 to S7 to continue processing the remaining requests. If all connection requests have been processed, calculate a series of network evaluation metrics, including network congestion rate, spectrum utilization, average end-to-end latency, and average energy consumption. These metrics are important for evaluating network performance, guiding network optimization, and planning future resource expansion. Through these evaluations, a better understanding of the network's operating status can be achieved, allowing for corresponding adjustments and improvements.

[0055] In step S7, after a connection request is established, the path weights are adjusted to give the network adaptive and self-learning capabilities: when a path is successfully occupied, its weight increases accordingly. When subsequent connection requests select paths in S2, the algorithm will automatically reduce the selection of paths with lower weights, thus guiding traffic to be distributed to idle paths, achieving global load balancing and avoiding continuous congestion on hot paths. At the same time, the dynamic changes in path weights can reflect the network load status in real time, allowing the network to adapt autonomously to tidal or sudden traffic without manual adjustment of routing strategies. In addition, the distributed distribution of traffic can reduce the impact of single-point failures and enhance network robustness. This weight update mechanism based on real-time connection feedback can combine short-term resource allocation with long-term performance optimization, reduce the long-term average network congestion rate, and improve overall operating efficiency.

[0056] In this embodiment of the invention, the update coefficient of the dynamic load balancing link weight is dynamically calculated based on the real-time network status, accurately reflecting the bandwidth occupancy within the optical channel. When using the KSP algorithm for routing, this method can better balance link length and bandwidth occupancy within the optical channel, preventing the link weight from suddenly increasing due to excessive bandwidth occupancy, thereby preventing a sudden increase in transmission path length. Therefore, compared with traditional load balancing methods, this embodiment of the invention utilizes a dynamic load balancing method that does not cause a surge in end-to-end latency after bandwidth occupancy exceeds a threshold, thus improving network stability and performance. Furthermore, addressing the shortcomings of traditional load balancing methods that only consider the spectrum slot occupancy ratio, this embodiment of the invention considers the bandwidth occupancy within the optical channel and performs dynamic load balancing, resulting in a more balanced load on services between the optical fiber and the data center, improving the resource utilization efficiency of the data center's elastic optical network, and enhancing energy efficiency.

[0057] This invention optimizes resource allocation and balances network load through a dynamic load balancing method, thereby improving network efficiency and reducing energy consumption. The above steps reduce the number of energy-consuming components in operation, achieving energy reduction and optimized resource allocation. Furthermore, this invention proposes a hierarchical traffic routing method to further refine resource management. In specific implementation, for each connection request, this invention first assesses whether the source and destination nodes have sufficient computing resources. After confirming sufficient resources, a multiple shortest path algorithm is used to determine the optimal working path route. This step ensures the efficiency and rationality of path selection. Next, hierarchical traffic routing and spectrum resource allocation are performed. The system traverses each optical channel of the fiber along the transmission path, assesses its remaining spectrum resources, and calculates hierarchical traffic routing evaluation parameters. The goal is to identify the spectrum layer with the smallest evaluation parameters for effective traffic routing. This hierarchical routing strategy helps maximize spectrum resource utilization and ensures efficient traffic transmission. In the spectrum resource allocation phase, this invention allocates the required spectrum resources according to the specific bandwidth requirements of the connection request. This process ensures that the requirements for spectrum consistency and continuity are met, configuring an appropriate number of optical repeater ports and optical regenerator ports for connection requests to support smooth data transmission. Specifically, this invention employs a minimum optical regenerator configuration method when configuring optical regenerators. This method aims to minimize the number of optical regenerators used, thereby reducing network complexity and operating costs while maintaining network performance. In summary, this invention, through a combination of dynamic load balancing and hierarchical traffic routing, achieves optimized resource allocation and effective network load balancing, providing an effective solution for improving network performance and reducing energy consumption.

[0058] To explain in detail the operating mechanism of the method described in the embodiments of the present invention, we will specifically explain how to perform dynamic load balancing, how to effectively save spectrum resources and reduce the use of network power-consuming components through hierarchical traffic routing strategies, and how to achieve optimized configuration of minimum optical regenerators based on the temporary topology of the working path of connection requests. These measures aim to reduce the number of power-consuming components in operation, thereby reducing overall network energy consumption and improving the resource utilization efficiency of optical networks. The following will use... Figure 2 The following is a detailed explanation using the hierarchical traffic routing network structure for dynamic load balancing in a data center elastic optical network as an example.

[0059] exist Figure 2In this system, each optical network node (AF) is connected to a data center node (DC1-DC6). In the data center elastic optical network, the data centers at each data center node provide limited computing resources. Considering that at a certain moment, the computing resources of the data center servers are 65, 33, 2, 81, 91, and 78 units respectively, the corresponding data centers are DC1, DC2, DC3, DC4, DC5, and DC6. Under these conditions, the specific implementation steps are as follows: The first step is to initialize the elastic optical network G(N, L, E, F), including the topology information of the data center elastic optical network, the optical network connection status, the number of network switching nodes, the number of fiber links, the number of available spectrum gaps in each fiber, the bandwidth of each spectrum gap, and the number of computing resources for each data center node.

[0060] The second step is to generate a set of connection requests CR(s, d, ts, td, es, ed, BR), including the source and destination nodes of the connection request, the arrival and departure times of the connection request (in seconds S), the computing resources required by the data center servers at the source and destination nodes, and the bandwidth requirements of the connection request. Here, two connection requests are generated: CR1(DC6, DC3, 2S, 5S, 6, 5, 40 Gbps) and CR2(DC2, DC4, 3S, 4S, 2, 7, 100Gbps). Figure 2 In the above, CR1 (DC6, DC3, 2S, 5S, 6, 5, 40 Gbps) requires 6 computing resources at the source node DC6 and 5 computing resources at the destination node DC3; CR2 (DC2, DC4, 3S, 4S, 2, 7, 100Gbps) requires 2 computing resources at the source node DC2 and 7 computing resources at the destination node DC4.

[0061] Thirdly, for connection request CR1 (DC6, DC3, 2S, 5S, 6, 5, 40 Gbps), the required computing resources exceed the available computing resources of the data center server. Specifically, at the data center node DC3, the required computing resources are greater than the available resources, i.e., 5 > 2, which cannot meet the user's request. Therefore, CR1 (DC6, DC3, 2S, 5S, 6, 5, 40 Gbps) is blocked. For connection request CR2 (DC2, DC4, 3S, 4S, 2, 7, 100 Gbps), the computing resources provided by the data center at the data center node can meet the connection request requirements, i.e., 33 > 2 at DC2 and 81 > 7 at DC4. The process continues to the next step.

[0062] Fourth, for connection request CR2(DC2, DC4, 3S, 4S, 2, 7, 100Gbps), the K shortest path algorithm is used to calculate the working path from the source node DC2 to the destination node DC4, and the length of the selected working path is calculated. The working path selected for CR2(DC2, DC4, 3S, 4S, 2, 7, 100Gbps) is “DC2-B→F→E→D-DC4”, and the length of this working path is 2800km.

[0063] Step 5: For CR2 (DC2, DC4, 3S, 4S, 2, 7, 100Gbps), first determine whether traffic can be routed. If the remaining spectrum resources meet the routed requirements, calculate the required number of power-consuming component ports for both routed and unroute portions. For the routed portion of the traffic in connection request CR2 (DC2, DC4, 3S, 4S, 2, 7, 100Gbps), a spectrum layer needs to be generated and the hierarchical traffic routed evaluation parameters calculated. The traffic in the spectrum layer with the smallest existing hierarchical traffic routed evaluation parameters is routed together with the number of spectrum slots. If the traffic cannot be routed, calculate the required number of power-consuming component ports for the connection request (when routed, the selected power-consuming component ports are those not selected at that moment, and their state changes from dormant to active) and the number of spectrum slots. The number of spectrum slots is equal to the bandwidth required by the connection request divided by the spectral efficiency value required by the modulation format and the bandwidth of each spectrum slot.

[0064] Step 6: For the configuration of the optical regenerator, establish a new topology on the CR2 (DC2, DC4, 3S, 4S, 2, 7, 100Gbps) working path "DC2-B→F→E→D-DC4". Traverse all paths on the topology and configure the optical regenerator according to the maximum transmission path of the optical signal corresponding to different modulation formats. For example, using the QPSK modulation format, its maximum reachable distance is 1800km. If the lengths of BF, FE, and ED are 900km, 800km, and 1100km respectively, then an optical regenerator needs to be set up at node E.

[0065] Step 7: Allocation of spectrum and computing resources. Spectrum resources are allocated using the first-hit algorithm, and computing resources are allocated at the source and destination nodes, ensuring successful connection establishment. The status of computing and spectrum resources is updated in real-time, and the number of successful connections is recorded. The dynamic load balancing path weights in the network topology are updated according to formulas (4) to (6). Furthermore, the energy consumption increase in the entire optical transmission network after the connection request enters the network is calculated, and the total energy consumption is updated.

[0066] The method described in this invention reduces energy consumption and conserves spectrum resources in resilient optical networks (ROCs) for data centers, while overcoming the shortcomings of traditional load balancing and traffic routing methods. By stratifying the idle spectrum resources of each optical fiber according to optical channels at different rates, and considering both spectrum and channel occupancy, the link weights are updated to address the three core issues of routing, load balancing, and spectrum resource allocation in resilient optical networks within a data center context. This aims to reduce network congestion rates, lower network energy consumption, and optimize resource utilization.

[0067] Example 2: Based on the same inventive concept, this embodiment provides a resource allocation system for dynamic load balancing of elastic optical networks in data centers. The principle of solving the problem is similar to the resource allocation method for dynamic load balancing of elastic optical networks in data centers provided in Embodiment 1, and the repetitions will not be repeated.

[0068] Reference Figure 3 As shown, this embodiment provides a resource allocation system for dynamic load balancing in a data center elastic optical network, used to implement the method described in Embodiment 1, including: The network initialization module is used to obtain network parameters of the data center's elastic optical network and initialize the network parameters; The connection request generation module is used to generate connection requests; the connection request includes a source node and a destination node; The computing power resource judgment module is used to determine whether the remaining computing power resources at the source node and the destination node are both greater than or equal to the computing power resources required for the connection request; if so, the module proceeds to the working path establishment module; otherwise, the connection request establishment fails. The working path creation module is used to calculate K working paths and select the shortest path from the K working paths; where K is a positive integer; The hierarchical traffic routing module is used to calculate the number of spectrum slots required for a connection request based on the shortest path; based on the number of spectrum slots, it determines whether there are remaining spectrum resources on the transmission path of the connection request that satisfy traffic routing; if so, it calculates the hierarchical traffic routing evaluation parameters and selects the spectrum layer with the smallest hierarchical traffic routing evaluation parameters for traffic routing; if not, it does not perform traffic routing, selects the highest-priced modulation format that can be transmitted, and proceeds to the decision and early warning module. The decision and warning module is used to determine whether there are sufficient bandwidth and optical channel resources on the transmission path of the connection request. If so, spectrum resources are allocated to the connection request; if not, it returns to the working path establishment module and selects the shortest path from the remaining working paths. It also determines whether the spectrum resources meet the consistency and continuity constraints. If so, computing power resources are allocated to the connection request and the data processing and computing module is entered; if not, it returns to the working path establishment module. The data processing and calculation module is used to calculate the number of optical repeaters and optical regenerators required for the connection request. The computing power resource update module is used to update the spectrum resources and computing power resources of the connection request after the connection request is successfully established. The dynamic load balancing module is used to update the network transmission path weights for connection requests.

[0069] This embodiment provides a resource allocation system for dynamic load balancing in a data center elastic optical network. Through dynamic load balancing, hierarchical traffic routing, and optical regenerator configuration, it reduces resource waste in the network, lowers the network congestion rate, and reduces the energy consumption of the optical network.

[0070] Specifically, the network initialization module is used to configure the network topology information, optical network connection status, number of network optical switching nodes, number of fiber links, number of spectrum slots in each fiber, and number of computing resources provided by the data center server in the data center optical network G(N, L, E, F).

[0071] Specifically, the connection request generation module is used to generate a set of connection requests and configure information such as the number of connection requests, the number of spectrum gaps required for different connection requests, computing power resource requirements, source nodes and destination nodes, and bandwidth requirements.

[0072] Specifically, the computing power resource judgment module is used to first determine whether the data center servers at the source node and destination node of the user request have sufficient computing power resources, and to determine whether the condition "the remaining computing power resources at the data center node are greater than or equal to the computing power resources required for the connection request" is met.

[0073] Specifically, for the working path establishment module, based on the connection request CR(s, d, ts, td, es, ed, BR), the K shortest path algorithm is used to calculate K candidate paths from the source node to the destination node in order to find the optimal path as the working path.

[0074] Specifically, for the hierarchical traffic diversion module, sufficient remaining spectrum resources are sought in optical channels with remaining bandwidth resources for traffic diversion in order to reduce the number of energy-consuming components used.

[0075] Specifically, the decision and warning module is used to perform coordination functions between various modules, as well as decision and warning functions for whether each module has been successfully established, to achieve the goal of energy consumption optimization in the elastic optical network for data centers, while improving the spectrum resource efficiency of the elastic optical network and reducing the blocking rate.

[0076] Specifically, the data processing and calculation module is used to calculate the number of optical repeaters and optical regenerators after each connection request is successfully established, calculate the energy consumption of each connection request, count the total number of connection requests and the number of successfully established requests, and calculate evaluation indicators such as network congestion rate, spectrum utilization rate, and average energy consumption.

[0077] Specifically, the computing resource update module is used to update the computing resources of the data center servers at the source and destination nodes of the connection request after the spectrum resources are successfully allocated.

[0078] Specifically, the dynamic load balancing module is used to calculate the current network status and update the dynamic load balancing path weights in the network topology after each connection request is successfully established, so as to balance the network load and optimize the resource allocation in the optical channel.

[0079] Furthermore, the resource allocation system for dynamic load balancing of the data center elastic optical network also includes a spectrum resource allocation module, a resource release module, and a network status monitoring module.

[0080] Specifically, the spectrum resource allocation module searches for bandwidth resources in the selected working path that meet the number of spectrum slots required by the connection request CR(s, d, ts, td, es, ed, BR). If both spectrum continuity and spectrum consistency constraints are met simultaneously, the connection request is successfully established; otherwise, the connection request fails to establish. This module works in conjunction with the decision and warning modules.

[0081] Specifically, the resource release module is used to first release the spectrum resources occupied by the working path after the connection request is successfully transmitted, and release the occupied hardware resources such as optical transceivers and optical regenerators; then, release the computing power resources of the data center server occupied by the connection request; and finally, clear the information of the working path established by the connection request.

[0082] Specifically, the network status monitoring module is used to perform status monitoring functions for elastic optical network initialization, connection request generation, computing resource judgment, working path establishment, dynamic load balancing, hierarchical traffic routing, spectrum resource allocation, computing resource update, resource release, and data processing and calculation, so as to achieve the goal of minimizing connection request latency when allocating computing resources.

[0083] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0084] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0085] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0086] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0087] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A resource allocation method for dynamic load balancing in a data center elastic optical network, characterized in that, include: S1. Obtain the network parameters of the data center's elastic optical network; generate a connection request based on the network parameters; wherein the connection request includes a source node and a destination node; determine whether the remaining computing resources at the source node and the destination node are both greater than or equal to the computing resources required by the connection request; if yes, proceed to S2; if no, the connection request fails to establish. S2. Calculate K working paths, select the shortest path from the K working paths, and calculate the number of spectrum slots required for the connection request; S3. Based on the number of spectrum slots, determine whether there are remaining spectrum resources on the transmission path of the connection request that satisfy traffic routing; if yes, proceed to S4; if no, do not perform traffic routing, select the highest-priced modulation format that can be transmitted, and proceed to S5. S4. Calculate the hierarchical flow diversion evaluation parameters and select the spectrum layer with the smallest hierarchical flow diversion evaluation parameters for flow diversion. S5. Determine whether there are sufficient bandwidth and optical channel resources on the transmission path of the connection request. If yes, allocate spectrum resources to the connection request; otherwise, return to S2 and select the shortest path from the remaining working paths. S6. Determine whether the spectrum resources meet the consistency and continuity constraints; if yes, allocate computing resources to the connection request and calculate the number of optical repeaters and optical regenerators required for the connection request, and the connection request is successfully established; if no, return to S2. S7. After the connection request is successfully established, update the spectrum resources and computing power resources of the connection request, and update the network transmission path weight of the connection request.

2. The resource allocation method for dynamic load balancing in a data center elastic optical network according to claim 1, characterized in that, In step S7, updating the network transmission path weight of the connection request includes calculating the network transmission path weight. The expression for calculating the network transmission path weight is: ； in, This represents the network transmission path weight, and β2 represents an adjustable parameter. This represents the total number of optical channels between nodes m and n. The initial weight value represents the distance of the network transmission path. This indicates the number of optical channels whose bandwidth usage exceeds the adjustable parameter, where C represents the set of optical channels. This indicates the number of spectrum slots in the optical fiber that have been occupied by connection requests. This represents the total number of spectral gaps in the optical fiber between two optical switching nodes m and n.

3. The resource allocation method for dynamic load balancing in a data center elastic optical network according to claim 1, characterized in that, In step S2, the method for calculating the number of spectral slots required for the connection request is as follows: calculate the product of the spectral efficiency value required by the modulation format and the bandwidth of each spectral slot, and obtain the number of spectral slots required for the connection request based on the product and the bandwidth required by the connection request.

4. The resource allocation method for dynamic load balancing in a data center elastic optical network according to claim 1, characterized in that, The process of determining whether there are remaining spectrum resources to meet traffic diversion requirements on the transmission path of the connection request based on the number of spectrum slots in step S3 includes spectrum layering.

5. The resource allocation method for dynamic load balancing in a data center elastic optical network according to claim 4, characterized in that, The method for spectrum layering is as follows: traverse the occupied optical channels on the working path, obtain the optical channels with remaining bandwidth resources that meet the transmission bandwidth requirements of the connection request, and obtain the spectrum layer.

6. The resource allocation method for dynamic load balancing in a data center elastic optical network according to claim 1, characterized in that, The expression for calculating the stratified flow diversion assessment parameters in step S4 is as follows: ； in, This represents the parameters for evaluating stratified flow distribution. Represents the spectral residual coefficients. Indicates the number of the spectral layer; This represents the time overlap coefficient.

7. The resource allocation method for dynamic load balancing in a data center elastic optical network according to claim 1, characterized in that, In step S6, the step of calculating the number of optical repeaters and optical regenerators required for the connection request is as follows: Based on the source and destination nodes in the selected working path, a temporary topology for calculating the number of optical regenerators is rebuilt. Traverse all path selections in the temporary topology, select the path with the lowest total energy consumption, analyze the usage of optical regenerators at all nodes of each path in turn, perform traffic diversion for optical regenerators with remaining spectrum resources, reconfigure optical regenerators at nodes where diversion is not possible, and calculate the number of optical regenerators required for all paths.

8. A resource allocation system for dynamic load balancing in a data center elastic optical network, used to implement the method described in any one of claims 1 to 7, characterized in that, include: The network initialization module is used to obtain the network parameters of the data center elastic optical network and initialize the network parameters; The connection request generation module is used to generate connection requests; The connection request mentioned above includes a source node and a destination node; The computing power resource judgment module is used to determine whether the remaining computing power resources at the source node and the destination node are both greater than or equal to the computing power resources required by the connection request; if yes, then proceed to the working path establishment module; if no, then the connection request establishment fails. The working path creation module is used to calculate K working paths and select the shortest path from the K working paths; A hierarchical traffic routing module is used to calculate the number of spectrum slots required for the connection request based on the shortest path. Based on the number of spectrum slots, determine whether there are remaining spectrum resources on the transmission path of the connection request that meet the requirements for traffic diversion; If yes, calculate the hierarchical traffic diversion evaluation parameters and select the spectrum layer with the smallest hierarchical traffic diversion evaluation parameters for traffic diversion; if no, do not perform traffic diversion, select the highest-priced modulation format that can be transmitted, and proceed to the decision and early warning module. The decision and warning module is used to determine whether there are sufficient bandwidth and optical channel resources on the transmission path of the connection request. If so, spectrum resources are allocated to the connection request; if not, the process returns to the working path establishment module, which selects the shortest path from the remaining working paths. The module also determines whether the spectrum resources meet the consistency and continuity constraints. If so, computing power resources are allocated to the connection request, and the process proceeds to the data processing and computing module; if not, the process returns to the working path establishment module. The data processing and calculation module is used to calculate the number of optical repeaters and optical regenerators required for the connection request; The computing power resource update module is used to update the spectrum resources and computing power resources of the connection request after the connection request is successfully established. The dynamic load balancing module is used to update the network transmission path weight of the connection request.

9. A resource allocation system for dynamic load balancing in a data center elastic optical network according to claim 8, characterized in that, It also includes a resource release module, which is used to release the spectrum resources occupied by the working path after the connection request is successfully established, and to release the occupied optical transponder and optical regenerator resources.

10. A resource allocation system for dynamic load balancing in a data center elastic optical network according to claim 8, characterized in that, The data processing and calculation module is also used to calculate the energy consumption of each connection request after each connection request is successfully established, and to calculate the evaluation index of the connection request.

Images