Load regulation method, apparatus, device, and storage medium

Abstract

The present disclosure provides a load regulation method, device, equipment and storage medium, relating to the technical field of computer. The method comprises: acquiring multi-source heterogeneous parameters of the load, constructing a congestion potential energy function based on the multi-source heterogeneous parameters and a preset algorithm, and determining whether congestion occurs in a preset time period based on the congestion potential energy function and a deep reinforcement intelligent model. By acquiring the multi-source heterogeneous parameters in the RDMA technology and predicting whether congestion occurs in the future time period based on the multi-source heterogeneous parameters, the data transmission efficiency of RDMA in a wide area scenario is improved.

Description

Technical Field

[0001] This disclosure relates to the field of communication technology, and in particular to a load regulation method, apparatus, device and storage medium. Background Technology

[0002] Remote Direct Memory Access (RDMA) technology has become a core technology for high-speed data transmission within data centers due to its advantages of low latency, high throughput, and low CPU overhead. However, its deployment in wide-area scenarios faces many bottlenecks. On the one hand, when wide-area RDMA transmits across geographical regions, it is affected by the high round-trip time (RTT) and large bandwidth-delay product (BDP) of long-distance links. Existing single-path RDMA transmission schemes in data centers cannot utilize the multi-core optical fiber and wavelength division multiplexing resources of wide-area optical transmission networks. The link bandwidth utilization is low, the load balancing effect is poor, and it is difficult to meet the massive data transmission needs across data centers in ultra-large-scale intelligent computing networks. On the other hand, the congestion control of existing wide-area RDMA is mostly passive response-based, that is, adjusting the rate by detecting congestion characteristics such as packet loss, queue congestion, and RTT mutation. However, the long latency of wide-area links leads to a significant lag in congestion feedback. Passive adjustment is prone to causing a sharp drop in link throughput and a significant increase in flow completion time, which cannot adapt to the low latency and high reliability transmission requirements of wide-area RDMA. Summary of the Invention

[0003] This disclosure provides a load regulation method, apparatus, device, and storage medium, which improves data transmission efficiency based on RDMA technology in wide-area scenarios to at least a certain extent.

[0004] Other features and advantages of this disclosure will become apparent from the following detailed description, or may be learned in part from practice of this disclosure.

[0005] According to one aspect of this disclosure, a load regulation method is provided, comprising: Obtain multi-source heterogeneous parameters of the load. These multi-source heterogeneous parameters include at least one of the following: time-based fiber core utilization, wavelength utilization, optical layer resource load parameters, queue length data, congestion data, packet transmission rate data, round-trip delay data, and bandwidth delay data. Congestion potential function is constructed based on multi-source heterogeneous parameters and a preset algorithm; The congestion potential function and a deep-reinforcement intelligent model are used to determine whether congestion will occur within a preset time period.

[0006] In one embodiment of this disclosure, multi-source heterogeneous parameters are normalized based on an intelligent model to obtain parameters with the same dimensions; Congestion potential function is constructed based on multi-source heterogeneous parameters and a pre-defined algorithm, including: Congestion potential function is constructed based on parameters with the same dimensions and a pre-defined algorithm.

[0007] In one embodiment of this disclosure, the method further includes: If congestion occurs within a predetermined time period, determine whether there are any idle wavelengths or fiber cores. Data is transmitted based on the available wavelength or fiber core when there is an available wavelength or fiber core.

[0008] In one embodiment of this disclosure, transmitting data based on an idle wavelength or fiber core includes: Data with the same IP address is transmitted using wavelengths or fiber cores with transmission paths differing within a preset range, so that the transmission time of data with the same IP address is within a preset range.

[0009] In one embodiment of this disclosure, determining whether congestion occurs within a preset time period based on a congestion potential function and a deep reinforcement intelligent model includes: Determine the value and derivative of the congestion potential function corresponding to the congestion potential function; The congestion potential function value and its derivative are input into the deep augmented intelligent model to obtain the model output, which indicates whether congestion has occurred within a preset time period.

[0010] In one embodiment of this disclosure, the method further includes: A training set is constructed based on historical congestion potential function values, historical derivatives, and historical congestion data; Input the training set into the untrained deep augmentation intelligent model to obtain the model output; Train the untrained deep augmentation intelligent model based on the model output and loss function; When the loss function value corresponding to the loss function converges, a deep reinforcement intelligent model is obtained.

[0011] According to another aspect of this disclosure, a load regulation device is provided, comprising: The acquisition module is used to acquire multi-source heterogeneous parameters of the load. The multi-source heterogeneous parameters include at least one of the following: time-based fiber core utilization, wavelength utilization, optical layer resource load parameters, queue length data, congestion data, packet transmission rate data, round-trip delay data, and bandwidth delay data. The module is used to construct the congestion potential function based on multi-source heterogeneous parameters and preset algorithms; The first determination module is used to determine whether congestion will occur within a preset time period based on the congestion potential function and the deep reinforcement intelligent model.

[0012] In one embodiment of this disclosure, the apparatus further includes: The normalization module is used to normalize multi-source heterogeneous parameters based on the intelligent model to obtain parameters with the same dimensions. Build modules, including: The building unit is used to construct the congestion potential function based on parameters with the same dimensions and a preset algorithm.

[0013] In one embodiment of this disclosure, the apparatus further includes: The second determining module is used to determine whether there is an idle wavelength or fiber core when congestion occurs within a predetermined time period. A transmission module for transmitting data based on an idle wavelength or fiber core when such an idle wavelength or fiber core is available.

[0014] In one embodiment of this disclosure, the transmission module includes: The transmission unit is used to transmit data with the same IP address using wavelengths or fiber cores with transmission paths that differ within a preset range, so that the transmission time of the data with the same IP address is within a preset range.

[0015] In one embodiment of this disclosure, the apparatus further includes: The control module is used to control the packet transmission rate when there are no available wavelengths or fiber cores.

[0016] In one embodiment of this disclosure, the first determining module includes: The determination unit is used to determine the value and derivative of the congestion potential function corresponding to the congestion potential function; The input unit is used to input the congestion potential function value and its derivative into the deep augmentation intelligent model to obtain the model output result, which indicates whether congestion has occurred within a preset time period.

[0017] In one embodiment of this disclosure, the apparatus further includes: The module is used to build a training set based on historical congestion potential function values, historical derivatives, and historical congestion data. The first training module is used to input the training set into the untrained deep augmented intelligence model and obtain the model output results; The second training module trains the untrained deep augmented intelligence model based on the model output and the loss function. The module is used to obtain a deep reinforcement intelligent model when the loss function value corresponding to the loss function converges.

[0018] According to another aspect of this disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to perform the above-described load control method by executing the executable instructions.

[0019] According to another aspect of this disclosure, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the above-described load control method.

[0020] The load regulation method, apparatus, device, and storage medium provided in the embodiments of this disclosure acquire multi-source heterogeneous parameters of the load, construct a congestion potential function based on the multi-source heterogeneous parameters and a preset algorithm, and determine whether congestion will occur within a preset time period based on the congestion potential function and a deep enhanced intelligent model. By acquiring multi-source heterogeneous parameters in RDMA technology and predicting whether congestion will occur in future time periods based on the aforementioned multi-source heterogeneous parameters, the data transmission efficiency of RDMA in wide-area scenarios is improved.

[0021] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0022] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0023] Figure 1 This diagram illustrates a load control system architecture according to an embodiment of the present disclosure. Figure 2 This diagram illustrates a load regulation method according to an embodiment of the present disclosure. Figure 3 This diagram illustrates another load regulation method according to an embodiment of the present disclosure. Figure 4 This diagram illustrates a flowchart of yet another load regulation method according to an embodiment of the present disclosure; Figure 5 This diagram illustrates a flowchart of yet another load regulation method according to an embodiment of the present disclosure; Figure 6 This diagram illustrates a flowchart of yet another load regulation method according to an embodiment of the present disclosure; Figure 7 This diagram illustrates a flowchart of yet another load regulation method according to an embodiment of the present disclosure; Figure 8 This diagram illustrates a flowchart of yet another load regulation method according to an embodiment of the present disclosure; Figure 9 A structural diagram of a load regulation device according to an embodiment of this disclosure is shown; Figure 10 A structural block diagram of an electronic device according to an embodiment of the present disclosure is shown. Detailed Implementation

[0024] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that this disclosure will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0025] Furthermore, the accompanying drawings are merely illustrative of this disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0026] It should be understood that the steps described in the method embodiments of this disclosure may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this disclosure is not limited in this respect.

[0027] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0028] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0029] It should be noted that, where there is no conflict, the embodiments of this disclosure and the technical features in the embodiments can be combined with each other.

[0030] The specific implementation of the embodiments of this disclosure will now be described in detail with reference to the accompanying drawings.

[0031] Figure 1A schematic diagram of a load control system structure according to an embodiment of the present disclosure is shown. This system can apply the load control methods or load control processing devices in various embodiments of the present disclosure.

[0032] like Figure 1 As shown, the load control system 10 may include a data acquisition module 101 and a load control module 102. The data acquisition module 101 and the load control module 102 may be located on two different devices. The data acquisition module 101 may be a module on an electronic device with data acquisition capabilities, such as a recording device with sound collection capabilities, a photographic device with image collection capabilities, or a computer with text information collection capabilities. The load control module 102 may be a module on an electronic device with processing capabilities, such as a computer. The data acquisition module 101 and the load control module 102 may also be located on the same device. For example, the data acquisition module 101 and the load control module 102 may be an input module and a processing module on a computer or mobile phone.

[0033] The data acquisition module 101 and the load control module 102 are connected by a network, which can be a wired network or a wireless network.

[0034] Optionally, the aforementioned wireless or wired networks use standard communication technologies and / or protocols. The network is typically the Internet, but can also be any network, including but not limited to Local Area Networks (LANs), Metropolitan Area Networks (MANs), Wide Area Networks (WANs), mobile, wired or wireless networks, private networks, or any combination of virtual private networks. In some embodiments, technologies and / or formats including Hyper Text Markup Language (HTML), Extensible Markup Language (XML), etc., are used to represent data exchanged over the network. Furthermore, conventional encryption technologies such as Secure Socket Layer (SSL), Transport Layer Security (TLS), Virtual Private Networks (VPNs), and Internet Protocol Security (IPsec) can be used to encrypt all or some links. In other embodiments, custom and / or dedicated data communication technologies can be used to replace or supplement the aforementioned data communication technologies.

[0035] The following describes the case where the data acquisition module 101 and the load control module 102 are located on two different devices.

[0036] The data acquisition module 101 can be located on a terminal device, which can be various electronic devices, including but not limited to smartphones, tablets, laptops, desktop computers, wearable devices, augmented reality devices, virtual reality devices, etc.

[0037] Optionally, the client for the application installed on different terminal devices can be the same, or the client for the same type of application based on different operating systems. Depending on the terminal platform, the specific form of the application client can also differ; for example, the application client can be a mobile client, a PC client, etc.

[0038] The load control module 102 can be located on a server, which can be a server that provides various services, such as a backend management server that supports the operation of devices by users using terminal devices. The backend management server can analyze and process the received requests and other data, and feed the processing results back to the terminal device.

[0039] Optionally, the server can be a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms.

[0040] Those skilled in the art will know that Figure 1 The number of data acquisition modules and load control modules shown is merely illustrative; any number of video acquisition modules and personality recognition modules can be included depending on actual needs. This disclosure does not limit this.

[0041] To address the aforementioned problems, embodiments of this disclosure provide a load regulation method, apparatus, device, and storage medium.

[0042] Figure 2 A flowchart of a load regulation method according to an embodiment of this disclosure is shown.

[0043] like Figure 2 As shown, the method may include: S210, acquire the multi-source heterogeneous parameters of the load. The multi-source heterogeneous parameters include at least one of the following: time-based fiber core utilization, wavelength utilization, optical layer resource load parameters, queue length data, congestion data, packet transmission rate data, round-trip delay data, and bandwidth delay data.

[0044] In some embodiments, the time-based fiber core utilization can be used to represent the degree to which a particular fiber is occupied at time t. It can be determined by the ratio of at least one of the number of allocable channels, subcarriers, and spectral blocks to at least one of the total effective channels and spectral capacity available from the fiber core.

[0045] Wavelength utilization can be used to indicate the degree of utilization of a wavelength channel at time t. It can also be used to indicate the wavelengths already occupied on a fiber. It can be determined based on the ratio of the wavelengths occupied in a given optical path to the wavelengths that the path can provide.

[0046] Optical layer resource load parameters can be used to represent the degree of occupancy of optical layer resources. Optical layer resource load parameters can be determined based on the fiber core utilization rate and wavelength utilization rate at a given time.

[0047] Queue length data can be used to represent the queuing depth of data packets waiting to be serialized and sent onto the link.

[0048] Congestion data can be used to reflect the location, extent, and type of congestion.

[0049] Packet transmission rate data can be used to reflect rate observations on the sending side.

[0050] Round-trip delay data can be used to reflect the total delay sample from the sender to the receiver.

[0051] In some embodiments, the above-mentioned multi-source heterogeneous parameters can be obtained through configured sensors. This disclosure does not limit the method for obtaining multi-source heterogeneous parameters.

[0052] S220 constructs a congestion potential function based on multi-source heterogeneous parameters and a preset algorithm.

[0053] In some embodiments, the preset algorithm may include an algorithm for determining the stability of a dynamic system comprising multiple parameters. This disclosure does not specifically limit the preset algorithm.

[0054] For example, the preset algorithm includes the Lyapunov algorithm, which may include defining a quadratic Lyapunov function for each "backlog" to be stabilized, and using its "drift" to determine the current action at each step, thereby turning "stability" into an optimizable algebraic term.

[0055] In some embodiments, the core premise of the Lyapunov optimization algorithm is to construct a unified state vector that can comprehensively describe the system state, integrate multi-source heterogeneous parameters of the optical layer, IP layer, and queue state, eliminate dimensional differences, and realize unified input of multi-source information.

[0056]

[0057] Wherein, Uc(t) is the fiber core utilization rate at time t, Uw(t) is the wavelength utilization rate at time t, representing the optical layer resource load status; q(t) is the queue length data at time t, representing the buffer congestion level; r(t) is the RDMA packet transmission rate data at time t, RTT(t) is the round-trip delay data at time t, and BDP(t) is the bandwidth delay data at time t. The three together represent the RDMA transmission behavior and wide area link characteristics.

[0058] To quantify the deviation between the current system state and the "ideal steady state without congestion," a pre-defined algorithm constructs a congestion potential function, transforming multi-dimensional heterogeneous parameters into unified non-negative scalar values, thus enabling the measurability, calculability, and comparability of the system's congestion level. Its mathematical expression can be as follows:

[0059] in, This represents the optimal steady-state utilization rate of optical layer resources, i.e., the maximum load level under congestion-free conditions. This represents the maximum capacity of the queue. Let be the weighting coefficient, satisfying It can be dynamically adjusted according to the priority of RDMA services to ensure that key parameters have higher weight.

[0060] The weighting coefficients can be adjusted based on a preset strategy, which may include: In a wide-area, long-distance environment, when the RTT exceeds a certain threshold, the queue length weight coefficient is increased to prioritize suppressing queue congestion and adapt to the high-risk characteristics of queue overflow under long RTT.

[0061] When fiber loss exceeds the threshold, the comprehensive weighting coefficient of optical layer resource utilization is increased to prioritize the balanced utilization of optical layer resources and reduce the impact of transmission loss on congestion.

[0062] When BDP exceeds the threshold, the weighting coefficient of the packet sending rate and BDP matching item is increased to avoid congestion spread caused by excessive expansion of the sending window.

[0063] S230 determines whether congestion will occur within a preset time period based on the congestion potential function and a deep-enhanced intelligent model.

[0064] In some embodiments, after determining the deep reinforcement intelligent model, a preset time period can be input into the deep reinforcement intelligent model to obtain the model output result. The model output result is used to indicate whether congestion occurs within the preset time period.

[0065] The load regulation method provided in the embodiments of this disclosure acquires multi-source heterogeneous parameters of the load, constructs a congestion potential function based on the multi-source heterogeneous parameters and a preset algorithm, and determines whether congestion will occur within a preset time period based on the congestion potential function and a deep enhanced intelligent model. By acquiring multi-source heterogeneous parameters in RDMA technology and predicting whether congestion will occur in future time periods based on the aforementioned multi-source heterogeneous parameters, the data transmission efficiency of RDMA in wide-area scenarios is improved.

[0066] Figure 3 A flowchart of another load regulation method in an embodiment of this disclosure is shown.

[0067] like Figure 3 As shown, the method may include: S310, acquire the multi-source heterogeneous parameters of the load. The multi-source heterogeneous parameters include at least one of the following: moment core utilization, wavelength utilization, optical layer resource load parameters, queue length data, congestion data, packet transmission rate data, round-trip delay data, and bandwidth delay data. S320 normalizes multi-source heterogeneous parameters based on an intelligent model to obtain parameters with the same dimensions.

[0068] In some embodiments, the above-mentioned multi-source heterogeneous parameters can be input into the above-mentioned intelligent model to obtain normalized parameters.

[0069] In some embodiments, having the same units can include parameters that can be calculated using the same computing system. For example, time and distance have different units, while minutes and seconds have the same unit.

[0070] For example, the aforementioned multi-source heterogeneous parameters can be vectorized to obtain vectorized data. The vectorized data are the parameters with the same dimensions.

[0071] S330 constructs a congestion potential function based on parameters with the same dimensions and a preset algorithm.

[0072] S340 determines whether congestion will occur within a preset time period based on the congestion potential function and a deep-enhanced intelligent model.

[0073] The load regulation method provided in this disclosure acquires multi-source heterogeneous parameters of the load, normalizes these parameters, constructs a congestion potential function based on the normalized parameters and a preset algorithm, facilitating the calculation of the congestion potential function, and determines whether congestion will occur within a preset time period based on the congestion potential function and a deep reinforcement intelligent model. By acquiring multi-source heterogeneous parameters in RDMA technology and predicting whether congestion will occur in future time periods based on these parameters, the data transmission efficiency of RDMA in wide-area scenarios is improved.

[0074] Figure 4 A flowchart of another load regulation method according to an embodiment of this disclosure is shown.

[0075] like Figure 4 As shown, the method may include: S410, acquire multi-source heterogeneous parameters of the load, including at least one of the following: moment core utilization, wavelength utilization, optical layer resource load parameters, queue length data, congestion data, packet transmission rate data, round-trip delay data, and bandwidth delay data.

[0076] S420 constructs a congestion potential function based on multi-source heterogeneous parameters and a preset algorithm.

[0077] S430 determines whether congestion will occur within a preset time period based on the congestion potential function and a deep-enhanced intelligent model.

[0078] S440 determines whether there are any idle wavelengths or fiber cores if congestion occurs within a preset time period.

[0079] In some embodiments, an idle wavelength or fiber core may include a wavelength and fiber core with load capacity. The fiber core utilization rate and wavelength utilization rate have been described in the above embodiments. An idle wavelength or fiber core may include a wavelength or fiber core whose fiber core utilization rate and wavelength utilization rate have not reached a preset threshold.

[0080] S450 transmits data based on an available wavelength or fiber core when such a wavelength or fiber core is available.

[0081] The load control method provided in this disclosure acquires multi-source heterogeneous parameters of the load, normalizes these parameters, constructs a congestion potential function based on the normalized parameters and a preset algorithm, facilitating the calculation of the congestion potential function, and determines whether congestion will occur within a preset time period based on the congestion potential function and a deep-enhanced intelligent model. By acquiring multi-source heterogeneous parameters in RDMA technology and predicting whether congestion will occur in future time periods based on these parameters, data is transmitted based on idle wavelengths or fiber cores in the event of congestion, thus improving the data transmission efficiency of RDMA in wide-area scenarios.

[0082] Figure 5 A flowchart of another load regulation method according to an embodiment of this disclosure is shown.

[0083] like Figure 5 As shown, the method may include: S510, acquires multi-source heterogeneous parameters of the load, including at least one of the following: time-based fiber core utilization, wavelength utilization, optical layer resource load parameters, queue length data, congestion data, packet transmission rate data, round-trip delay data, and bandwidth delay data.

[0084] S520 constructs a congestion potential function based on multi-source heterogeneous parameters and a preset algorithm.

[0085] S530 determines whether congestion will occur within a preset time period based on the congestion potential function and a deep-enhanced intelligent model.

[0086] S540 determines whether there are any idle wavelengths or fiber cores if congestion occurs within a preset time period.

[0087] S550, when there is an available wavelength or fiber core, transmits data with the same IP using a wavelength or fiber core with a transmission path difference within a preset range, so that the transmission time of the data with the same IP is within a preset range.

[0088] In some embodiments, data with the same IP address may include data with the same DstIP, or data with the same DstIP and SrcIP. It should be noted that the same DstIP can mean that all this data is destined for the same remote machine / the same IP node. The same DstIP and SrcIP can mean a group of traffic belonging to the same pair of RDMA nodes.

[0089] The load control method provided in this disclosure acquires multi-source heterogeneous parameters of the load, normalizes these parameters, constructs a congestion potential function based on the normalized parameters and a preset algorithm, facilitating the calculation of the congestion potential function, and determines whether congestion will occur within a preset time period based on the congestion potential function and a deep-enhanced intelligent model. By acquiring multi-source heterogeneous parameters in RDMA technology and predicting whether congestion will occur in future time periods based on these parameters, in the event of congestion, data with the same IP is transmitted using wavelengths or fiber cores with transmission path differences within a preset range, so that the transmission time of data with the same IP is within the preset range, thereby improving the data transmission efficiency and quality of RDMA in wide-area scenarios.

[0090] Figure 6 A flowchart of yet another load regulation method according to an embodiment of this disclosure is shown.

[0091] like Figure 6As shown, the method may include: S610, acquire the multi-source heterogeneous parameters of the load, including at least one of the following: moment core utilization, wavelength utilization, optical layer resource load parameters, queue length data, congestion data, packet transmission rate data, round-trip delay data, and bandwidth delay data. S620 constructs a congestion potential function based on multi-source heterogeneous parameters and a preset algorithm; S630 determines whether congestion will occur within a preset time period based on the congestion potential function and a deep enhanced intelligent model.

[0092] S640 determines whether there are any idle wavelengths or fiber cores if congestion occurs within a preset time period.

[0093] The S650 controls the packet transmission rate when there are no available wavelengths or fiber cores.

[0094] In some embodiments, controlling the packet transmission rate may include controlling the rate at which data is transmitted based on RDMA technology. For example, the packet transmission rate may be controlled based on congested data.

[0095] In some embodiments, the derivative of the congestion potential function can be determined, and the packet sending rate can be controlled based on the derivative.

[0096] The load control method provided in this disclosure acquires multi-source heterogeneous parameters of the load, normalizes these parameters, constructs a congestion potential function based on the normalized parameters and a preset algorithm, facilitating the calculation of the congestion potential function, and determines whether congestion will occur within a preset time period based on the congestion potential function and a deep reinforcement intelligent model. By acquiring multi-source heterogeneous parameters in RDMA technology and predicting whether congestion will occur in future time periods based on these parameters, the method controls the packet transmission rate in the event of congestion, thereby improving the data transmission efficiency and quality of RDMA in wide-area scenarios.

[0097] Figure 7 A flowchart of yet another load regulation method according to an embodiment of this disclosure is shown.

[0098] like Figure 7 As shown, the method may include: S710 acquires multi-source heterogeneous parameters of the load, including at least one of the following: moment core utilization, wavelength utilization, optical layer resource load parameters, queue length data, congestion data, packet transmission rate data, round-trip delay data, and bandwidth delay data.

[0099] S720 constructs a congestion potential function based on multi-source heterogeneous parameters and a preset algorithm.

[0100] S730, determine the congestion potential function value and derivative corresponding to the congestion potential function.

[0101] In some embodiments, time data can be input into the congestion potential energy function to obtain the potential energy function value.

[0102] In some embodiments, the derivative of the congestion potential function can be the first derivative.

[0103] For example, the first derivative of the congestion potential function can be:

[0104] When the first derivative is less than 0, the system state tends to converge, gradually approaching a congestion-free steady state, and the risk of congestion continues to decrease.

[0105] When the first derivative is greater than 0, the system state tends to diverge, deviating from the congestion-free steady state, and the risk of congestion increases significantly. It is about to enter an unstable state and an early warning needs to be triggered.

[0106] The S740 inputs the congestion potential function value and its derivative into the deep augmented intelligent model to obtain the model output result, which indicates whether congestion has occurred within a preset time period.

[0107] The load control method provided in this disclosure acquires multi-source heterogeneous parameters of the load, normalizes these parameters, constructs a congestion potential function based on the normalized parameters and a preset algorithm, facilitating the calculation of the congestion potential function, and determines whether congestion will occur within a preset time period based on the congestion potential function and a deep reinforcement intelligent model. By acquiring multi-source heterogeneous parameters in RDMA technology and predicting whether congestion will occur in future time periods based on these parameters, the method controls the packet transmission rate in the event of congestion, thereby improving the data transmission efficiency and quality of RDMA in wide-area scenarios.

[0108] Figure 8 A flowchart of yet another load regulation method according to an embodiment of this disclosure is shown.

[0109] like Figure 8 As shown, the method may include: S810 constructs a training set based on historical congestion potential function values, historical derivatives, and historical congestion data.

[0110] In some embodiments, the method for obtaining historical congestion potential function values ​​and historical derivatives can be the same as the method for obtaining congestion potential function values ​​and derivatives, and will not be described again here.

[0111] Methods for obtaining historical congestion data can include obtaining it through historical measurements.

[0112] The S820 takes the training set as input to an untrained deep augmented intelligence model and obtains the model output.

[0113] S830 trains an untrained deep reinforcement intelligent model based on the model output and loss function.

[0114] In some embodiments, the loss function may include: .

[0115] S840, when the loss function value corresponding to the loss function converges, yields a deep reinforcement intelligent model.

[0116] Based on the same inventive concept, this disclosure also provides a load regulation device, as shown in the following embodiment. Since the principle by which this device embodiment solves the problem is similar to that of the above-described method embodiment, the implementation of this device embodiment can refer to the implementation of the above-described method embodiment, and repeated details will not be elaborated further.

[0117] Figure 9 A structural diagram of a load regulation device according to an embodiment of the present disclosure is shown.

[0118] like Figure 9 As shown, the device 900 may include: The acquisition module 910 is used to acquire multi-source heterogeneous parameters of the load. The multi-source heterogeneous parameters include at least one of the following: time-based fiber core utilization, wavelength utilization, optical layer resource load parameters, queue length data, congestion data, packet transmission rate data, round-trip delay data, and bandwidth delay data. Module 920 is used to construct a congestion potential function based on multi-source heterogeneous parameters and a preset algorithm. The first determining module 930 is used to determine whether congestion will occur within a preset time period based on the congestion potential function and the deep reinforcement intelligent model.

[0119] In one embodiment of this disclosure, the apparatus further includes: The normalization module is used to normalize multi-source heterogeneous parameters based on the intelligent model to obtain parameters with the same dimensions. Build modules, including: The building unit is used to construct the congestion potential function based on parameters with the same dimensions and a preset algorithm.

[0120] In one embodiment of this disclosure, the apparatus further includes: The second determining module is used to determine whether there is an idle wavelength or fiber core when congestion occurs within a predetermined time period. A transmission module for transmitting data based on an idle wavelength or fiber core when such an idle wavelength or fiber core is available.

[0121] In one embodiment of this disclosure, the transmission module includes: The transmission unit is used to transmit data with the same IP address using wavelengths or fiber cores with transmission paths that differ within a preset range, so that the transmission time of the data with the same IP address is within a preset range.

[0122] In one embodiment of this disclosure, the apparatus further includes: The control module is used to control the packet transmission rate when there are no available wavelengths or fiber cores.

[0123] In one embodiment of this disclosure, the first determining module includes: The determination unit is used to determine the value and derivative of the congestion potential function corresponding to the congestion potential function; The input unit is used to input the congestion potential function value and its derivative into the deep augmentation intelligent model to obtain the model output result, which indicates whether congestion has occurred within a preset time period.

[0124] In one embodiment of this disclosure, the apparatus further includes: The module is used to build a training set based on historical congestion potential function values, historical derivatives, and historical congestion data. The first training module is used to input the training set into the untrained deep augmented intelligence model and obtain the model output results; The second training module trains the untrained deep augmented intelligence model based on the model output and the loss function. The module is used to obtain a deep reinforcement intelligent model when the loss function value corresponding to the loss function converges.

[0125] The load control device provided in the embodiments of this disclosure acquires multi-source heterogeneous parameters of the load, constructs a congestion potential function based on the multi-source heterogeneous parameters and a preset algorithm, and determines whether congestion will occur within a preset time period based on the congestion potential function and a deep enhanced intelligent model. By acquiring multi-source heterogeneous parameters in RDMA technology and predicting whether congestion will occur in future time periods based on the aforementioned multi-source heterogeneous parameters, the data transmission efficiency of RDMA in wide-area scenarios is improved.

[0126] The load control device provided in this embodiment can be used to execute the load control methods provided in the above-described method embodiments. The implementation principle and technical effect are similar, and for the sake of simplicity, they will not be described in detail here.

[0127] Those skilled in the art will understand that various aspects of this disclosure can be implemented as a system, method, or program product. Therefore, various aspects of this disclosure can be specifically implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or a combination of hardware and software aspects, collectively referred to herein as a "circuit," "module," or "system."

[0128] The following reference Figure 10 To describe an electronic device 1000 according to such an embodiment of the present disclosure. Figure 10 The electronic device 1000 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments disclosed herein.

[0129] like Figure 10 As shown, the electronic device 1000 is manifested in the form of a general-purpose computing device. The components of the electronic device 1000 may include, but are not limited to: at least one processing unit 1010, at least one storage unit 1020, and a bus 1030 connecting different system components (including storage unit 1020 and processing unit 1010).

[0130] The storage unit stores program code, which can be executed by the processing unit 1010, causing the processing unit 1010 to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of this disclosure. For example, the processing unit 1010 can perform the following steps of the above method embodiments: Obtain historical optical fiber network maintenance parameters for multiple categories and the corresponding maintenance efficiency indicators for each category of historical optical fiber network maintenance parameters; Multiple sample datasets are identified, each containing historical optical fiber network maintenance parameters for any category and their corresponding maintenance efficiency indicators. The model is trained on multiple sample datasets to obtain multiple trained measurement models; Based on a preset algorithm, the indicator data corresponding to each of the multiple measurement models is determined; The target calculation model is determined based on the indicator data corresponding to each calculation model.

[0131] Storage unit 1020 may include readable media in the form of volatile storage units, such as random access memory (RAM) 10201 and / or cache memory 10202, and may further include read-only memory (ROM) 10203.

[0132] Storage unit 1020 may also include a program / utility 10204 having a set (at least one) program module 10205, such program module 10205 including but not limited to: operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0133] Bus 1030 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the multiple bus structures.

[0134] Electronic device 1000 can also communicate with one or more external devices 1040 (e.g., keyboard, pointing device, Bluetooth device, etc.), one or more devices that enable a user to interact with electronic device 1000, and / or any device that enables electronic device 1000 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 1050. Furthermore, electronic device 1000 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 1060. As shown, network adapter 1060 communicates with other modules of electronic device 1000 via bus 1030. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 1000, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0135] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, terminal device, or network device, etc.) to execute the methods according to the embodiments of this disclosure.

[0136] In exemplary embodiments of this disclosure, a computer-readable storage medium is also provided, which may be a readable signal medium or a readable storage medium. A program product capable of implementing the methods described above is stored thereon. In some possible implementations, various aspects of this disclosure may also be implemented as a program product including program code, which, when run on a terminal device, causes the terminal device to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of this disclosure.

[0137] More specific examples of computer-readable storage media in this disclosure may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0138] In this disclosure, a computer-readable storage medium may include a data signal propagated in baseband or as part of a carrier wave, wherein readable program code is carried. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of transmitting, propagating, or transmitting a program for use by or in connection with an instruction execution system, apparatus, or device.

[0139] Optionally, the program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0140] In practice, program code for performing the operations of this disclosure can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0141] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0142] Furthermore, although the steps of the method in this disclosure are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or a step may be broken down into multiple steps.

[0143] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, mobile terminal, or network device, etc.) to execute the methods according to the embodiments of this disclosure.

[0144] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the appended claims.

Claims

1. A load regulation method, characterized in that, include: Obtain multi-source heterogeneous parameters of the load, wherein the multi-source heterogeneous parameters include at least one of the following: moment core utilization, wavelength utilization, optical layer resource load parameters, queue length data, congestion data, packet transmission rate data, round-trip delay data, and bandwidth delay data; Congestion potential function is constructed based on the aforementioned multi-source heterogeneous parameters and a preset algorithm; Based on the congestion potential function and the deep reinforcement intelligent model, it is determined whether congestion will occur within a preset time period.

2. The method according to claim 1, characterized in that, The method further includes: The multi-source heterogeneous parameters are normalized based on the intelligent model to obtain parameters with the same dimensions. The construction of the congestion potential function based on the multi-source heterogeneous parameters and the preset algorithm includes: Congestion potential function is constructed based on parameters with the same dimensions and a preset algorithm.

3. The method according to claim 1, characterized in that, The method further includes: If congestion occurs within a predetermined time period, determine whether there are any idle wavelengths or fiber cores. Data is transmitted based on the available wavelength or fiber core when there is an available wavelength or fiber core.

4. The method according to claim 3, characterized in that, The data transmitted based on idle wavelengths or fiber cores includes: Data with the same IP address is transmitted using wavelengths or fiber cores with transmission paths differing within a preset range, so that the transmission time of data with the same IP address is within a preset range.

5. The method according to claim 3, characterized in that, The method further includes: Control the packet transmission rate when there are no available wavelengths or fiber cores.

6. The method according to claim 1, characterized in that, The step of determining whether congestion occurs within a preset time period based on the congestion potential function and the deep reinforcement intelligent model includes: Determine the value and derivative of the congestion potential function corresponding to the congestion potential function; The congestion potential function value and the derivative are input into the deep enhanced intelligent model to obtain the model output result, which indicates whether congestion occurs within a preset time period.

7. The method according to claim 1, characterized in that, The method further includes: A training set is constructed based on historical congestion potential function values, historical derivatives, and historical congestion data; The training set is input into the untrained deep reinforcement intelligence model to obtain the model output. The untrained deep reinforcement intelligent model is trained based on the model output and the loss function. The deep reinforcement intelligent model is obtained when the loss function value corresponding to the loss function converges.

8. A load regulation device, characterized in that, include: The acquisition module is used to acquire multi-source heterogeneous parameters of the load, which include at least one of the following: time-based fiber core utilization, wavelength utilization, optical layer resource load parameters, queue length data, congestion data, packet transmission rate data, round-trip delay data, and bandwidth delay data. The construction module is used to construct a congestion potential function based on the multi-source heterogeneous parameters and a preset algorithm; The first determining module is used to determine whether congestion occurs within a preset time period based on the congestion potential energy function and the deep reinforcement intelligent model.

9. An electronic device, characterized in that, include: processor; as well as Memory for storing the executable instructions of the processor; The processor is configured to execute the load control method according to any one of claims 1 to 7 by executing the executable instructions.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the load control method according to any one of claims 1 to 7.

Images