Method for migrating data center optical interconnection resources

Abstract

The application relates to a data center optical interconnection resource migration method. The method comprises the following steps: acquiring first state information of each optical interconnection resource in a data center, and determining optical interconnection resources to be migrated based on the first state information; formulating an optical interconnection resource migration scheme according to the first state information corresponding to the optical interconnection resources to be migrated; the optical interconnection resource migration scheme is used to instruct an optical interconnection resource migration device to migrate the optical interconnection resources to be migrated; acquiring second state information of each optical interconnection resource after migration, and verifying the optical interconnection resources to be migrated after migration in the data center based on the second state information. By using the method, the migration scheme of the optical interconnection resources is formulated in an intelligent manner, the efficiency of migrating the optical interconnection resources can be improved, and the stable operation of the data center can be ensured.

Description

Technical Field

[0001] This invention belongs to the field of data center automated management technology, and in particular relates to a method for migrating optical interconnect resources in data centers. Background Technology

[0002] In today's era of rapid digital development, the importance of data centers as the core hubs for information storage, processing, and transmission is self-evident. With the continuous increase in data center workload and the ongoing technological upgrades, optical interconnect resources occupy a crucial position in the data center architecture.

[0003] However, as business needs change, on the one hand, data centers may require architectural adjustments, equipment upgrades, or layout optimizations, which involves the reconfiguration and migration of optical interconnect resources. On the other hand, optical interconnect resources may need to be migrated due to equipment failure, performance degradation, or network optimization requirements.

[0004] Traditional optical interconnect resource migration often relies on human experience for planning and execution, lacking systematic and scientific rigor, which can easily lead to low migration efficiency and high time costs.

[0005] Therefore, there is an urgent need for an intelligent method for migrating optical interconnect resources in data centers to improve the efficiency of migrating optical interconnect resources and ensure the stable operation of data centers. Summary of the Invention

[0006] Therefore, it is necessary to provide a data center optical interconnect resource migration method, a data center optical interconnect resource migration device, and a data center optical interconnect resource management system that can improve migration efficiency in response to the above-mentioned technical problems.

[0007] Firstly, this application provides a method for migrating optical interconnect resources in data centers, including:

[0008] Obtain the first state information of each optical interconnect resource in the data center, and determine the optical interconnect resources that need to be migrated based on the first state information;

[0009] Based on the first state information corresponding to the optical interconnect resources that need to be migrated, an optical interconnect resource migration plan is formulated; the optical interconnect resource migration plan is used to instruct the optical interconnect resource migration device to migrate the optical interconnect resources that need to be migrated.

[0010] Obtain the second state information of each optical interconnect resource after migration, and verify the migrated optical interconnect resources in the data center based on the second state information.

[0011] Secondly, this application also provides a data center optical interconnect resource migration device, comprising:

[0012] The resource monitoring module is used to monitor the status of each optical interconnect resource in the data center in real time, obtain the first status information of each optical interconnect resource in the data center, and determine the optical interconnect resources that need to be migrated based on the first status information.

[0013] The migration planning module is used to formulate a migration plan for optical interconnect resources based on the first state information corresponding to the optical interconnect resources that need to be migrated.

[0014] The migration execution module is used to migrate the optical interconnect resources that need to be migrated according to the optical interconnect resource migration scheme.

[0015] The verification module acquires the second state information of each optical interconnect resource after migration, and verifies the migrated optical interconnect resources in the data center based on the second state information.

[0016] Thirdly, this application also provides a data center optical interconnect resource management system, including: a data center, sensors, servers, and optical interconnect resource migration equipment;

[0017] The sensor connects the data center and the server to monitor the status of each optical interconnect resource in the data center in real time and send the first status information of each optical interconnect resource in the data center to the server.

[0018] The optical interconnect resource migration device connects the data center and the server, and is used to migrate the optical interconnect resources that need to be migrated according to the optical interconnect resource migration plan sent by the server;

[0019] The server is used to perform the method as described in the first aspect.

[0020] The aforementioned data center optical interconnect resource migration method, apparatus, and data center optical interconnect resource management system improve the efficiency of optical interconnect resource migration and ensure the stable operation of the data center by intelligently formulating migration plans for optical interconnect resources. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating the data center optical interconnect resource migration method provided by the present invention.

[0022] Figure 2 This is a schematic diagram illustrating the steps of developing an optical interconnect resource migration scheme in an optional embodiment of the present invention;

[0023] Figure 3 This is a schematic diagram illustrating the steps of training a deep reinforcement learning model in an optional embodiment of the present invention;

[0024] Figure 4 This is a schematic diagram of the structure of a data center optical interconnect resource migration device provided by the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0026] refer to Figure 1 The document presents a flowchart illustrating a data center optical interconnect resource migration method provided in an embodiment of this application. This method, applied to a server, includes the following steps:

[0027] S101. Obtain the first state information of each optical interconnect resource in the data center, and determine the optical interconnect resources that need to be migrated based on the first state information.

[0028] Specifically, the server acquires the initial status information of each optical interconnect resource in the data center. This status information may include the connection status, bandwidth usage, transmission efficiency, connection path, optical signal strength, and optical link bit error rate of the optical interconnect resource. By analyzing this status information, the server determines which optical interconnect resources need to be migrated. For example, if the bandwidth usage of an optical interconnect resource is too high or the transmission efficiency is low, the server may determine that it is an optical interconnect resource that needs to be migrated.

[0029] S102. Based on the first state information corresponding to the optical interconnect resources to be migrated, formulate an optical interconnect resource migration plan; the optical interconnect resource migration plan is used to instruct the optical interconnect resource migration device to migrate the optical interconnect resources to be migrated.

[0030] Specifically, the server formulates an optical interconnect resource migration plan based on the first state information corresponding to the optical interconnect resources to be migrated. This migration plan instructs the optical interconnect resource migration equipment to perform the migration operation on the optical interconnect resources to be migrated. When formulating the plan, the server can consider the current state of the optical interconnect resources to be migrated, the target location for migration, the migration schedule, potential problems during the migration process, and corresponding countermeasures. For example, if the connection of a certain optical interconnect resource to be migrated is relatively complex, the server may schedule the migration during a low-traffic period and notify the relevant equipment in advance to prepare.

[0031] S103. Obtain the second state information of each optical interconnect resource after migration, and verify the optical interconnect resources to be migrated in the data center based on the second state information.

[0032] Specifically, the server obtains the second status information of each migrated optical interconnect resource. This status information is similar to the first status information, but reflects the post-migration situation. Based on the second status information, the server verifies the migrated optical interconnect resources in the data center. Verification may include whether the optical interconnect resources have been successfully migrated to the target location, whether the connection is normal, whether bandwidth usage has improved, and whether transmission efficiency has increased. If problems are found during verification, the server will take corresponding measures to adjust and ensure the normal operation of the optical interconnect resources.

[0033] By using the methods described above, servers can be used to intelligently formulate migration plans for optical interconnect resources, thereby improving the efficiency of migrating optical interconnect resources and ensuring the stable operation of the data center.

[0034] refer to Figure 2 In one optional embodiment, an optical interconnect resource migration scheme is formulated based on the first state information corresponding to the optical interconnect resources to be migrated, including the following steps:

[0035] S201. Based on the first state information corresponding to the optical interconnect resource to be migrated, determine the current connection path of the optical interconnect resource to be migrated in the data center.

[0036] Specifically, the first state information may include information related to the connection path, such as the connected device nodes and port information. By analyzing this information, the server can accurately determine the specific connection path of the optical interconnect resources that need to be migrated in the data center.

[0037] S202. Based on the current connection path of the optical interconnect resources to be migrated in the data center, and taking into account factors such as path length, service impact, and optical signal loss, plan the optimal migration path.

[0038] Specifically, after determining the current connection path, the server plans the optimal migration path based on key considerations such as path length, service impact, and optical signal loss. Regarding path length, a shorter path likely means lower migration costs and less signal transmission latency. As for optical signal loss, the server selects the path that minimizes optical signal loss to ensure efficient and stable data transmission from the migrated optical interconnect resources. Regarding service impact, the server needs to assess the interference of different paths on ongoing services and choose the path with the least impact. For example, different migration paths traverse different network devices and links within the data center, each with varying bandwidth allocations. If the migration path traverses links with limited bandwidth, while the data traffic involved in the migration process is large, it will lead to increased latency in service data transmission, thus amplifying the service impact.

[0039] S203. Determine the migration time window based on the first state information corresponding to the optical interconnect resources to be migrated.

[0040] Specifically, the first state information can include the usage frequency of optical interconnect resources, peak business periods, historical network traffic data, etc. By analyzing this information, the server selects a migration time window during a period of relatively low business volume, relatively low network traffic, and minimal impact on business. For example, if a certain optical interconnect resource is busy with business and has relatively high network traffic during the day, the server will choose to migrate it at night.

[0041] S204. Use the migration time window and migration path as the migration scheme for optical interconnect resources.

[0042] Specifically, the server integrates the determined migration time window and the planned optimal migration path into a comprehensive optical interconnect resource migration plan. This plan will provide clear guidance for the migration of optical interconnect resources, ensuring an efficient and smooth migration process while minimizing the impact on data center operations.

[0043] In one optional embodiment, based on the current connection path of the optical interconnect resources to be migrated in the data center, and taking into account factors such as path length, service impact, and optical signal loss, the optimal migration path is planned, including the following three steps:

[0044] Step 1: Set the parameters of the simulated annealing algorithm, including the initial temperature T0, the cooling coefficient λ, and the maximum number of iterations N. max And establish the objective function E(x) = αL + γBI + βS; where x represents the current connection path; P represents optical signal loss. in P is the input optical power for optical interconnect resources. out The output optical power of optical interconnect resources; This indicates the degree of business impact, where n is the number of business transactions, and ω is the number of transactions. i Let t be the weight of the i-th business. mi Let t be the estimated maximum latency for the i-th service during the migration process. bi α is the baseline delay time for normal service operation; S is the path length; α is the weighting coefficient of optical signal loss in the objective function; γ is the weighting coefficient of service impact in the objective function; β is the weighting coefficient of path length in the objective function.

[0045] Step 2: Perform iterative optimization, including: in each iteration, generating a new connection path x′ based on the current connection path x; calculating the optical signal loss corresponding to the new connection path x′. Business impact And the length S′ of the new connection path, and then calculate the objective function value E(x′)=αL′+γBI′+βS′; according to the state transition probability formula Determine whether to accept the new path x′. If accepted, update the current connection path to the new connection path; update according to the temperature formula T. k+1 =λT k Update temperature T; where T k T is the temperature of the k-th iteration, and k ≥ 0; k+1 It is the temperature of the (k+1)th iteration; λ is the cooling coefficient, and 0 < λ < 1;

[0046] Step 3: Perform termination checks, including checking whether the termination condition is met during each iteration. The termination condition is that the number of iterations reaches the predetermined maximum number of iterations N. max When the termination condition is met, the algorithm terminates and outputs the found optimal new connection path x as the optimal migration path.

[0047] Specifically, the simulated annealing algorithm used here is a stochastic optimization algorithm based on the Monte Carlo iterative solution strategy, used to find the global optimum or near-optimal solution in a large search space. Its core idea is to simulate the solid-state annealing process, accepting solutions worse than the current solution with a certain probability during the search to avoid getting trapped in local optima. The basic formulas mainly involve state transition probabilities and the objective function.

[0048] First, let the current state (which can be understood as the current migration path in path optimization) be x, the new state (the new migration path) be x′, and the objective function be E(x) (in optical interconnect resource migration path optimization, this can be a function that includes factors such as optical signal loss L, service impact BI, and path length S, such as E(x)=αL+γBI+βS, where α, γ, and β are the weighting coefficients of optical signal loss L, service impact BI, and path length S in the objective function, respectively). The temperature parameter is T, and the state transition probability formula is... Cooling coefficient λ and maximum number of iterations N max .in, P in P is the input optical power for optical interconnect resources. out The output optical power of optical interconnect resources; n is the number of business transactions, ω i Let t be the weight of the i-th business. mi Let t be the estimated maximum latency for the i-th service during the migration process. bi This is the baseline delay time for normal business operations.

[0049] Then, iterative optimization is performed. In each iteration, a new connection path x′ is generated based on the current connection path x; the optical signal loss corresponding to the new connection path x′ is calculated Degree of service impact and the length S′ of the new connection path, and then the objective function value E(x′) = αL′ + γBI′ + βS′ is calculated; according to the state transition probability formula judge whether to accept the new path x′. If accepted, update the current connection path to the new connection path; update the temperature T according to the temperature update formula T k+1 = λT k , update the temperature T; where, T k is the temperature of the k-th iteration, and k ≥ 0; T k+1 is the temperature of the (k + 1)-th iteration; λ is the temperature reduction coefficient, and 0 < λ < 1.

[0050] Finally, check whether the termination condition is satisfied in each iteration process. The termination condition is that the number of iterations reaches the predetermined maximum number of iterations N max ; when the termination condition is satisfied, the algorithm terminates and outputs the found optimal new connection path x as the optimal migration path.

[0051] Among them, the above state transition probability formula means that when E(x′) < E(x), the formula is P(x → x′) = 1. Here, E(x) is the objective function value corresponding to the current state (current connection path), and E(x′) is the objective function value corresponding to the new state (new connection path). When the objective function value of the new connection path is less than that of the current connection path, it means that the new connection path is better in terms of the optimization objective (considering factors such as optical signal loss, degree of service impact, path length, etc.). Therefore, in this case, the probability of accepting the new path is 1, that is, the new connection path will definitely be accepted as the current path for the next iteration. For example, if the objective function value of the current path is 10 and the objective function value of the new path is 8, since the new connection path is better, the algorithm will毫不犹豫地 accept the new connection path.

[0052] When E(x′) ≥ E(x), the formula is At this time, the objective function value of the new connection path is greater than or equal to that of the current connection path, which means that the new connection path may not be as good as the current connection path in terms of the comprehensive performance measured by the objective function. However, to avoid falling into local optimality, the algorithm will not directly reject this new connection path, but decides whether to accept it with a probability. This probability is related to the difference in the objective function value E(x′) - E(x) and the current temperature parameter T. The exponential part The smaller the value of , the greater the probability P(x→x′) of accepting a new connection path. When the temperature T is high, even if the difference in the objective function values ​​is large, the probability of accepting a new connection path is relatively high. This is analogous to the fact that at high temperatures, the energy state of the system is more chaotic, making it more likely to accept states with slightly higher energy (analogous to the objective function values ​​in simulated annealing). As the temperature T decreases, the probability of accepting a worse new path also decreases, and the algorithm gradually converges to a better solution. For example, assuming the current temperature T = 10 and the difference in objective function values ​​E(x′) - E(x) = 2, the calculated acceptance probability... There is a certain probability of accepting this new connection path; when the temperature T=1, for the same difference in objective function values, the probability of acceptance... The probability of acceptance decreases.

[0053] In the optimization of optical interconnect resource connection paths, the state transition probability formula allows the algorithm to be more flexible in exploring new connection paths. Sometimes, although a new connection path may slightly increase optical signal loss, slightly increase service impact, or slightly decrease path length (leading to a larger objective function value), the algorithm still has the opportunity to accept this new path through this probability formula. This mechanism helps the algorithm escape the trap of local optima and find globally optimal or near-optimal connection paths in a wider path space, thereby better balancing multiple optimization objectives such as optical signal loss, service impact, and path length.

[0054] In an optional embodiment, the current connection path of the optical interconnect resource to be migrated in the data center is determined based on the first state information corresponding to the optical interconnect resource to be migrated, including the following steps:

[0055] Obtain the data center topology data from the first state information corresponding to the optical interconnect resources that need to be migrated;

[0056] Based on the data center topology data, determine the current connection path of the optical interconnect resources that need to be migrated within the data center.

[0057] Specifically, the first state information may include a description of the overall network architecture of the data center, such as the connection relationships and hierarchical structure between various devices. The server carefully filters and extracts this information to accurately obtain the data center's topology data. This step provides the basis for determining the specific location and connection status of the optical interconnect resources that need to be migrated.

[0058] After obtaining the data center topology data, the server will gradually track and analyze the connection paths of the optical interconnect resources that need to be migrated based on the device nodes and connection relationships in the topology. For example, by finding the devices connected to the resource and the connection links between these devices, the server can clearly determine the current connection path of the optical interconnect resources to be migrated in the data center, preparing for subsequent migration planning.

[0059] In an optional embodiment, determining the migration time window based on the first state information corresponding to the optical interconnect resources to be migrated includes the following steps:

[0060] Historical network traffic data is obtained from the first state information corresponding to the optical interconnect resources that need to be migrated;

[0061] Based on historical network traffic data, a trained deep reinforcement learning model is used to predict network traffic, resulting in network traffic prediction data.

[0062] Based on network traffic prediction data, the period of minimum network traffic is determined as the migration time window.

[0063] Specifically, the server retrieves historical network traffic data from the first state information corresponding to the optical interconnect resources that need to be migrated. The first state information may include traffic records of the network environment where the optical interconnect resources are located over a past period, including traffic volume at different times and the timing of peak traffic. By extracting this historical network traffic data, the server provides a basis for subsequent traffic prediction and determination of the migration time window.

[0064] The server uses a trained deep reinforcement learning model to predict network traffic based on acquired historical network traffic data, thus obtaining network traffic prediction data. The deep reinforcement learning model can learn from historical data to understand the patterns and trends in network traffic changes. The server inputs historical network traffic data into the model, which then performs calculations and analysis to output network traffic predictions for a future period. This process utilizes advanced machine learning techniques to improve the accuracy and reliability of traffic prediction.

[0065] Based on network traffic prediction data, the server identifies the period of lowest network traffic as the migration time window. By analyzing the network traffic prediction data, the server can pinpoint periods of low network traffic. Migrating optical interconnect resources during these periods minimizes the impact on network services. This is because lower network traffic means fewer ongoing data transmission tasks, thus reducing interference with existing services during the migration process. After careful screening and comparison, the server ultimately determines the period of lowest network traffic as the optimal migration time window to ensure a smooth migration process.

[0066] refer to Figure 3 In an optional embodiment, the training process of the deep reinforcement learning model includes the following steps:

[0067] S301. Obtain historical network traffic data from 30 days ago. The historical network traffic data from 30 days ago is used to train the deep reinforcement learning model, while the historical network traffic data from the past 30 days is used as input to the trained deep reinforcement learning model to obtain network traffic prediction data.

[0068] S302. Preprocess the historical network traffic data from 30 days ago to obtain a network traffic dataset; divide the network traffic dataset into a training set and a validation set.

[0069] S303. Construct a deep reinforcement learning model.

[0070] S304. Input the training set into the deep reinforcement learning model to train the deep reinforcement learning model.

[0071] S305. Input the validation set into the trained deep reinforcement learning model to optimize the parameters of the deep reinforcement learning model, and finally obtain the trained deep reinforcement learning model.

[0072] Specifically, the server retrieves historical network traffic data from 30 days ago, specifying that this data is used to train the deep reinforcement learning model, while historical network traffic data from the past 30 days is used as input to the trained model to obtain network traffic prediction data. This division ensures that model training and prediction use data from different time periods, avoiding overfitting and improving the model's generalization ability.

[0073] Historical network traffic data from 30 days ago was cleaned to remove outliers. The network traffic data is N = {n1, n2, ..., n}. t}, where n t Let represent the network traffic at time t. The mean is calculated. and standard deviation Data points outside the range [μ-k1σ, μ+k1σ] are considered outliers and removed, where k1 is a constant. Normalization is then performed using the formula... The data is placed within the range [0,1]. After preprocessing, a network traffic dataset is obtained, which is then divided into a training set (80%), a validation set (10%), and a test set (10%) in chronological order to ensure the temporal continuity of the data.

[0074] Deep reinforcement learning models can be constructed using DQN (Deep Q-Network). The state space is defined as the network traffic at the current time step and the previous few time steps, i.e. Where n t-i Let represent the network traffic at time ti, and k2 represent the number of historical time points considered. In this model focused on network traffic prediction, the action space can be simplified to not involve specific action selection, focusing primarily on the predicted network traffic output. The reward function is defined as Ri. t =-(Predicted) t -Actual t ) 2 Predicted here t It is the network traffic predicted by the model at time t, Actual t It is the actual network traffic at time t, and the model receives a higher reward when the predicted value is closer to the actual value.

[0075] The server feeds the training set into the deep reinforcement learning model to train it. The network traffic data samples in the training set are defined according to the state space. For each training sample time t, the model input is a vector consisting of network traffic values ​​from k2 previous time points. The model's parameters are set to θ, and the prediction function can be expressed as: in Is the model in state S t The model predicts the network traffic at time t. During training, the model minimizes the loss function. To adjust the parameter θ, here y i This is the actual network traffic value. This represents the model's predicted value, and n is the number of training samples. During training, the model makes predictions based on the input state space data (historical network traffic sequences in the training set) and calculates the reward value for the current prediction using the reward function. Although the loss function is L(θ), the reward function measures the quality of the model's predictions from another perspective. Essentially, during training, the model is trying to find a set of parameters θ that optimizes the prediction results. As close as possible to the actual value y i To maximize reward R t From the perspective of gradient descent, the gradient information of the reward function can provide a reference direction for adjusting the model parameters. Through the backpropagation algorithm, the model parameters are adjusted according to the loss function to make the model's prediction results as close as possible to the actual network traffic, that is, to optimize in the direction of maximizing the reward.

[0076] The validation set data is also organized according to the definition of state space. When the validation set is input into the model, for each sample time t in the validation set, the input received by the model is... By calculating the prediction error on the validation set, such as the mean squared error. Where m is the number of samples in the validation set. It is the actual network traffic value in the verification set. This refers to the model's predictions on the validation set. During the validation phase, when calculating the predictions on the validation set, the reward function can be used to intuitively evaluate the model's prediction quality on that set. For example, the average reward can be calculated for all samples in the validation set. in This is the reward value calculated based on the validation set samples. If the mean squared error of the validation set does not achieve the expected decrease or overfitting occurs, regularization methods can be used to optimize the model. For example, adding an L2 regularization term to the loss function results in a new loss function. Where λ is the regularization coefficient, θ j These are the model parameters. By adjusting the value of λ, the complexity of the model can be controlled, overfitting can be prevented, and the predictive performance of the model on unseen data (validation set) can be improved.

[0077] The test set is input into the trained model, and the model's predictive ability in real-world application scenarios is evaluated by calculating metrics such as the mean squared error (MSE) on the test set. The test set data is also input according to the state space definition. Let the number of test set samples be p, and the mean squared error on the test set is calculated... To evaluate model performance, where These are the actual network traffic values ​​in the test set. This is the model's predicted value on the test set. The average reward can also be calculated for the test set. The model's predictive performance in real-world applications is evaluated by observing the magnitude of the reward value. Simultaneously, combining metrics such as mean squared error on the test set allows for a more comprehensive assessment of whether the model can accurately predict network traffic and whether it can effectively predict based on the objective embodied in the reward function (reducing the gap between predicted and actual values). If the model performs well on the test set, it indicates good generalization ability and can be used for real-world network traffic prediction; if it performs poorly, further adjustments to the model's structure and parameters, or retraining and optimization, are needed.

[0078] In one optional embodiment, determining the optical interconnect resources that need to be migrated based on the first state information includes the following steps:

[0079] Obtain the optical link bit error rate and the optical signal strength corresponding to the connection path in each optical interconnect resource from the first state information;

[0080] Optical interconnect resources with an optical link bit error rate higher than a preset value or optical signal strength outside a preset range are marked.

[0081] The marked optical interconnect resources are treated as optical interconnect resources that need to be migrated.

[0082] Specifically, the server obtains the optical link error rate and the optical signal strength corresponding to the connection path for each optical interconnect resource from the first state information of the optical interconnect resources. The first state information can contain various performance parameters and status indicators of the optical interconnect resources. The server accurately obtains the two key parameters, optical link error rate and optical signal strength, through a specific data extraction method. The optical link error rate reflects the probability of errors occurring in the optical signal during transmission, while the optical signal strength indicates the strength of the optical signal.

[0083] The server marks optical interconnect resources with an optical link bit error rate higher than a preset value or with optical signal strength outside a preset range. The preset value and range are determined based on the data center's performance requirements and stability standards for optical interconnect resources. A high optical link bit error rate indicates a high frequency of errors during optical signal transmission, potentially affecting the accuracy and reliability of data transmission. An optical signal strength outside the preset range may lead to excessive or insufficient signal attenuation, affecting signal transmission quality and distance. The server marks non-compliant optical interconnect resources by comparing the actual optical link bit error rate and optical signal strength with the preset value and range.

[0084] The server identifies the marked optical interconnect resources as those requiring migration. After the marking process, those optical interconnect resources deemed inadequate or unacceptable are determined to be the ones needing migration. The server then organizes and records these marked optical interconnect resources, providing clear targets for subsequent optical interconnect resource migration planning.

[0085] In one optional embodiment, the migration of optical interconnect resources in the data center is verified based on the second state information, including the following steps:

[0086] Step 1: Obtain the second state information corresponding to the optical interconnect resources to be migrated after migration from the second state information; and based on the second state information corresponding to the optical interconnect resources to be migrated after migration, obtain the optical link bit error rate and optical signal strength of the optical interconnect resources to be migrated after migration as verification results;

[0087] Step 2: If the bit error rate in the verification result is higher than the preset value and / or the optical signal strength is not within the preset range, then determine the location of the fault based on the verification result and repair the fault.

[0088] Step 3: Repeat the above steps until the verification result meets the preset conditions; the preset conditions are that the bit error rate in the verification result is lower than the preset value and the optical signal intensity is within the preset range.

[0089] Specifically, the server retrieves the second-state information corresponding to the optical interconnect resources to be migrated from the second-state information. After the optical interconnect resource migration is completed, the server collects new state data to evaluate the migration effect. Then, based on this second-state information, the server obtains the optical link bit error rate and optical signal strength of the migrated optical interconnect resources. These two parameters, as verification results, directly reflect the performance status of the migrated optical interconnect resources. The optical link bit error rate reflects the accuracy of data transmission; the lower the bit error rate, the higher the reliability of data transmission. The optical signal strength determines the signal transmission quality and distance; a suitable optical signal strength can ensure stable signal transmission.

[0090] If the bit error rate in the verification results is higher than the preset value and / or the optical signal strength is outside the preset range, it indicates that the migrated optical interconnect resources still have problems. The server needs to determine the location of the fault based on the verification results. This involves checking the connection paths and device interfaces of the optical interconnect resources one by one, analyzing the anomalies in the bit error rate and optical signal strength, and combining the data center topology and equipment configuration information to determine the specific location of the possible fault. Once the fault location is determined, the server will generate a fault report, which is provided to maintenance personnel so that they can take appropriate measures to repair the fault. For example, if the problem is caused by poor connection, it may be necessary to reconnect or replace the connection cable; if it is a device failure, it may be necessary to replace or repair the relevant equipment.

[0091] The server iterates through the above steps, continuously verifying and repairing the migrated optical interconnect resources until the verification results meet preset conditions. These preset conditions are a bit error rate below a preset value and optical signal strength within a preset range. This process ensures that the optical interconnect resources function correctly after migration, meeting the data center's performance requirements. Through continuous iterative verification and repair, the server can gradually optimize the state of the optical interconnect resources, improving the overall stability and reliability of the data center.

[0092] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0093] refer to Figure 4 This application provides a data center optical interconnect resource migration device 400, which includes:

[0094] The resource monitoring module 401 is used to monitor the status of each optical interconnect resource in the data center in real time, obtain the first status information of each optical interconnect resource in the data center, and determine the optical interconnect resources that need to be migrated based on the first status information.

[0095] The migration planning module 402 is used to formulate a migration plan for optical interconnect resources based on the first state information corresponding to the optical interconnect resources to be migrated.

[0096] The migration execution module 403 is used to migrate the optical interconnect resources that need to be migrated according to the optical interconnect resource migration scheme.

[0097] The verification module 404 obtains the second state information of each optical interconnect resource after migration, and verifies the optical interconnect resources to be migrated in the data center based on the second state information.

[0098] Optionally, the migration planning module 402 includes:

[0099] The connection path acquisition unit is used to determine the current connection path of the optical interconnect resource to be migrated in the data center based on the first state information corresponding to the optical interconnect resource to be migrated.

[0100] The migration path planning unit is used to plan the optimal migration path based on the current connection path of the optical interconnect resources to be migrated in the data center, taking into account factors such as path length, service impact, and optical signal loss.

[0101] The migration time window determination unit is used to determine the migration time window based on the first state information corresponding to the optical interconnect resources to be migrated.

[0102] The migration scheme determination unit is used to determine the migration time window and migration path as the migration scheme for optical interconnect resources.

[0103] Optionally, the migration path planning unit is specifically used to perform the following operations, including:

[0104] Step 1: Set the parameters of the simulated annealing algorithm, including the initial temperature T0, the cooling coefficient λ, and the maximum number of iterations N. max And establish the objective function E(x) = αL + γBI + βS;

[0105] Where x represents the current connection path; P represents optical signal loss. in P is the input optical power for optical interconnect resources. out The output optical power of optical interconnect resources; This indicates the degree of business impact, where n is the number of business transactions, and ω is the number of transactions. i Let t be the weight of the i-th business. mi Let t be the estimated maximum latency for the i-th service during the migration process. bi α is the baseline delay time for normal service operation; S is the path length; α is the weighting coefficient of optical signal loss in the objective function; γ is the weighting coefficient of service impact in the objective function; β is the weighting coefficient of path length in the objective function.

[0106] Step 2: Perform iterative optimization, including:

[0107] In each iteration, a new connection path x′ is generated based on the current connection path x;

[0108] Calculate the optical signal loss corresponding to the new connection path x′ Business impact And the length S′ of the new connection path, and then calculate the objective function value E(x′)=αL′+γBI′+βS′;

[0109] According to the state transition probability formula Determine whether to accept the new path x′. If accepted, update the current connection path to the new connection path.

[0110] According to the temperature update formula T k+1 =λT k Update temperature T;

[0111] Among them, T k T is the temperature of the k-th iteration, and k ≥ 0; k+1 It is the temperature of the (k+1)th iteration; λ is the cooling coefficient, and 0 < λ < 1;

[0112] Step 3: Perform termination checks, including:

[0113] During each iteration, check if the termination condition is met. The termination condition is that the number of iterations reaches the predetermined maximum number of iterations N. max ;

[0114] When the termination condition is met, the algorithm terminates and outputs the found optimal new connection path x as the optimal migration path.

[0115] Optionally, the connection path acquisition unit includes:

[0116] The topology data acquisition component is used to acquire the topology data of the data center from the first state information corresponding to the optical interconnect resources to be migrated.

[0117] The topology data analysis component is used to determine the current connection path of optical interconnect resources that need to be migrated in the data center based on the data center's topology data.

[0118] Optionally, the migration time window determination unit includes:

[0119] The historical network traffic data acquisition component is used to acquire historical network traffic data from the first state information corresponding to the optical interconnect resources to be migrated.

[0120] The network traffic data prediction component is used to predict network traffic based on historical network traffic data using a trained deep reinforcement learning model, thereby obtaining network traffic prediction data.

[0121] The migration time window selection component is used to determine the period of minimum network traffic as the migration time window based on network traffic prediction data.

[0122] Optionally, the network traffic data prediction component is also used to train a deep reinforcement learning model. The training process includes the following steps:

[0123] Historical network traffic data from 30 days ago is obtained from historical network traffic data. The historical network traffic data from 30 days ago is used to train a deep reinforcement learning model, and the historical network traffic data from the past 30 days is used as input to the trained deep reinforcement learning model to obtain network traffic prediction data.

[0124] Preprocess historical network traffic data from 30 days ago to obtain a network traffic dataset; divide the network traffic dataset into a training set and a validation set.

[0125] Build deep reinforcement learning models;

[0126] The training set is fed into the deep reinforcement learning model to train the deep reinforcement learning model;

[0127] The validation set is input into the trained deep reinforcement learning model to optimize the parameters of the deep reinforcement learning model, and finally the trained deep reinforcement learning model is obtained.

[0128] Optionally, the resource monitoring module 401 includes:

[0129] The optical interconnect resource status parameter acquisition component is used to obtain the optical link bit error rate and the optical signal strength corresponding to the connection path in each optical interconnect resource from the first status information.

[0130] The marking component is used to mark optical interconnect resources whose optical link bit error rate is higher than a preset value or whose optical signal strength is outside the preset range.

[0131] The component for determining optical interconnect resources to be migrated is used to identify the marked optical interconnect resources as optical interconnect resources to be migrated.

[0132] Optionally, the verification module 404 is specifically used to perform the following operations, including:

[0133] Step 1: Obtain the second state information corresponding to the optical interconnect resources to be migrated after migration from the second state information; and based on the second state information corresponding to the optical interconnect resources to be migrated after migration, obtain the optical link bit error rate and optical signal strength of the optical interconnect resources to be migrated after migration as verification results;

[0134] Step 2: If the bit error rate in the verification result is higher than the preset value and / or the optical signal strength is not within the preset range, then determine the location of the fault based on the verification result and repair the fault.

[0135] Step 3: Repeat the above steps until the verification result meets the preset conditions; the preset conditions are that the bit error rate in the verification result is lower than the preset value and the optical signal intensity is within the preset range.

[0136] This application also provides a data center optical interconnect resource management system, which includes: a data center, sensors, servers, and optical interconnect resource migration equipment;

[0137] The sensor connects the data center and the server to monitor the status of each optical interconnect resource in the data center in real time and send the first status information of each optical interconnect resource in the data center to the server.

[0138] The optical interconnect resource migration device connects the data center and the server, and is used to migrate the optical interconnect resources that need to be migrated according to the optical interconnect resource migration plan sent by the server;

[0139] The server is used to perform the data center optical interconnect resource migration method as described in any of the above.

[0140] Specifically, a data center consists of a large number of data center servers, storage devices, network equipment, etc. Optical interconnect resources include fiber optic cables, optical modules, optical switches, etc. Sensors can be fiber optic sensors or photodetectors. Servers can be physical servers, virtual servers, or cloud servers. Optical interconnect resource migration equipment can be optical switches, fiber optic plug-in devices, or optical module hot-swappable devices.

[0141] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0142] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A method for migrating optical interconnect resources in a data center, characterized in that, The method includes: Obtain the first state information of each optical interconnect resource in the data center, and determine the optical interconnect resources that need to be migrated based on the first state information; Based on the first state information corresponding to the optical interconnect resources to be migrated, an optical interconnect resource migration plan is formulated; the optical interconnect resource migration plan is used to instruct the optical interconnect resource migration device to migrate the optical interconnect resources to be migrated. Obtain the second state information of each of the migrated optical interconnect resources, and verify the migrated optical interconnect resources in the data center based on the second state information; The step of formulating an optical interconnect resource migration scheme based on the first state information corresponding to the optical interconnect resources to be migrated includes: Based on the first state information corresponding to the optical interconnect resource to be migrated, determine the current connection path of the optical interconnect resource to be migrated in the data center; Based on the current connection path of the optical interconnect resources to be migrated in the data center, and taking into account factors such as path length, service impact, and optical signal loss, the optimal migration path is planned. Based on the first state information corresponding to the optical interconnect resources to be migrated, a migration time window is determined; The migration time window and the migration path are used as the optical interconnect resource migration scheme; The step of determining the optical interconnect resources that need to be migrated based on the first state information includes: Obtain the optical link bit error rate and the optical signal strength corresponding to the connection path in each optical interconnect resource from the first status information; Optical interconnect resources whose optical link bit error rate is higher than a preset value or whose optical signal strength is outside a preset range are marked. The marked optical interconnect resources are referred to as the optical interconnect resources to be migrated.

2. The method according to claim 1, characterized in that, The step of planning the optimal migration path based on the current connection path of the optical interconnect resources to be migrated in the data center, taking into account factors such as path length, service impact, and optical signal loss, includes: Step 1: Set the parameters of the simulated annealing algorithm, including the initial temperature T0, the cooling coefficient λ, and the maximum number of iterations N. max And establish the objective function E(x) = αL + γBI + βS; Where x represents the current connection path; P represents optical signal loss. in P is the input optical power for optical interconnect resources. out The output optical power of optical interconnect resources; This indicates the degree of business impact, where n is the number of business transactions, and ω is the number of transactions. i Let t be the weight of the i-th business. mi Let t be the estimated maximum latency for the i-th service during the migration process. bi The baseline delay time for normal service operation; S is the path length; α is the weighting coefficient of optical signal loss in the objective function; γ is the weighting coefficient of service impact in the objective function; β is the weighting coefficient of path length in the objective function. Step 2: Perform iterative optimization, including: In each iteration, a new connection path x′ is generated based on the current connection path x; Calculate the optical signal loss corresponding to the new connection path x′ Business impact And the length S′ of the new connection path, and then calculate the objective function value E(x′)=αL′+γBI′+βS′; According to the state transition probability formula Determine whether to accept the new path x′. If accepted, update the current connection path to the new connection path. According to the temperature update formula T k+1 =λT k Update temperature T; Among them, T k T is the temperature of the k-th iteration, and k ≥ 0; k+1 It is the temperature of the (k+1)th iteration; λ is the cooling coefficient, and 0 < λ < 1; Step 3: Perform termination checks, including: During each iteration, check whether the termination condition is met. The termination condition is that the number of iterations reaches a predetermined maximum number of iterations N. max ; When the termination condition is met, the algorithm terminates and outputs the found optimal new connection path x as the optimal migration path.

3. The method according to claim 1, characterized in that, Determining the current connection path of the optical interconnect resource to be migrated in the data center based on the first state information corresponding to the optical interconnect resource to be migrated includes: The topology data of the data center is obtained from the first state information corresponding to the optical interconnect resources that need to be migrated; Based on the topology data of the data center, determine the current connection path of the optical interconnect resource to be migrated in the data center.

4. The method according to claim 1, characterized in that, The step of determining the migration time window based on the first state information corresponding to the optical interconnect resources to be migrated includes: Historical network traffic data is obtained from the first state information corresponding to the optical interconnect resources that need to be migrated; Based on the historical network traffic data, a trained deep reinforcement learning model is used to predict network traffic, resulting in network traffic prediction data. Based on the network traffic prediction data, the time period with the lowest network traffic is determined as the migration time window.

5. The method according to claim 4, characterized in that, The training process of the deep reinforcement learning model includes: Historical network traffic data from 30 days ago is obtained from the historical network traffic data; wherein, the historical network traffic data from 30 days ago is used to train the deep reinforcement learning model, and the historical network traffic data from the past 30 days is used to input the trained deep reinforcement learning model to obtain the network traffic prediction data; The historical network traffic data from 30 days ago is preprocessed to obtain a network traffic dataset; the network traffic dataset is then divided into a training set and a validation set. Construct the deep reinforcement learning model; The training set is input into the deep reinforcement learning model to train the deep reinforcement learning model; The validation set is input into the trained deep reinforcement learning model to optimize the parameters of the deep reinforcement learning model, and finally the trained deep reinforcement learning model is obtained.

6. The method according to claim 1, characterized in that, The verification of the migrated optical interconnect resources in the data center based on the second state information includes: Step 1: Obtain the second state information corresponding to the migrated optical interconnect resource from the second state information; and based on the second state information corresponding to the migrated optical interconnect resource, obtain the optical link bit error rate and optical signal strength of the migrated optical interconnect resource as the verification result; Step 2: If the bit error rate in the verification result is higher than the preset value and / or the optical signal strength is not within the preset range, then determine the location of the fault based on the verification result and repair the fault. Step 3: Repeat the above steps until the verification result meets the preset conditions; wherein, the preset conditions are that the bit error rate in the verification result is lower than a preset value and the optical signal intensity is within a preset range.

7. A data center optical interconnect resource migration device, characterized in that, The device includes: The resource monitoring module is used to monitor the status of each optical interconnect resource in the data center in real time, obtain the first status information of each optical interconnect resource in the data center, and determine the optical interconnect resources that need to be migrated based on the first status information. The migration planning module is used to formulate a migration plan for the optical interconnect resources based on the first state information corresponding to the optical interconnect resources to be migrated. The migration execution module is used to migrate the optical interconnect resources to be migrated according to the optical interconnect resource migration scheme. The verification module acquires the second status information of each of the migrated optical interconnect resources, and verifies the migrated optical interconnect resources in the data center based on the second status information.

8. A data center optical interconnect resource management system, characterized in that, The system includes: a data center, sensors, servers, and optical interconnect resource migration equipment; The sensor connects the data center and the server, and is used to monitor the status of each optical interconnect resource in the data center in real time, and send the first status information of each optical interconnect resource in the data center to the server; The optical interconnect resource migration device connects the data center and the server, and is used to migrate the optical interconnect resources to be migrated according to the optical interconnect resource migration plan sent by the server. The server is used to perform the method as described in any one of claims 1 to 6.

Images