A method and device for energy saving in data center network

Abstract

The present invention discloses a method and device for energy saving in a data center network, and relates to the field of communication technology. The method comprises: when the data center networking configuration is completed, an initial allocation calculation of the member path selection probability is performed; during the packet receiving process, when an early congestion notification message is received, a congestion update calculation of the member path selection probability is performed; during the packet sending process, when a message is successfully sent, a packet sending update calculation of the member path selection probability is performed; and the member path selection probability is periodically detected, and an inactive port is selected and closed. The present invention calculates the member path selection probability through each distributed switch node, and realizes dynamic calculation of the bandwidth utilization of the ECMP path; at the same time, according to the calculated member path selection probability, the inactive port and the inactive switch can be dynamically selected and closed, so as to achieve the optimal planning of the energy consumption of the whole network, thereby solving the problem that the network energy saving effect in the prior art is not ideal.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a method and device for energy saving in a data center network. Background Art

[0002] With the development of large-scale Layer 2 network technology in data centers, the technology of using the underlay Layer 3 network to carry overlay user services has been widely used. In order to achieve high bandwidth and high availability in data centers, ECMP (Equal Cost Multi-path) technology is often used in networking to implement underlay routing selection. Generally, 2 N Equivalent paths (N=0~6).

[0003] At present, the traditional ECMP bandwidth calculation and energy-saving control working mechanism of data centers is: using a hybrid superposition neural network model to predict network traffic in the data center, using intelligent algorithms for deep reinforcement learning to optimize bandwidth allocation and routing selection in the network, and realizing energy-saving control of the data center network. However, in actual application, the existing solutions still have the following problems:

[0004] Problem 1: The existing solution cannot dynamically calculate the bandwidth utilization of the ECMP path, and does not consider the processing of redundant switch paths and corresponding switch ports, resulting in unsatisfactory energy saving effects for the entire network;

[0005] Problem 2: In the existing traffic prediction model, the detection cycle is fixed, which makes it difficult to cope with millisecond-level network burst traffic, which may lead to errors in link bandwidth demand calculation, thus affecting message transmission delay;

[0006] Question 3: The system collects data from each node in the network and calculates the centralized control model. The calculation time complexity is large and it is impossible to timely predict the real-time congestion traffic in the network, and thus it is impossible to select the optimal export link for packet sending based on the link traffic. Summary of the invention

[0007] The object of the present invention is to provide a data center network energy saving method and device, which at least solves one of the above-mentioned technical problems.

[0008] To achieve the above objectives, in a first aspect, an embodiment of the present invention provides a method for energy saving in a data center network, the method comprising:

[0009] When the data center networking configuration is completed, the initial allocation calculation of the member path selection probability is performed; during the packet receiving process, when the early congestion notification message is received, the congestion update calculation of the member path selection probability is performed; during the packet sending process, when the message is successfully sent, the packet sending update calculation of the member path selection probability is performed; the member path selection probability is periodically detected, and the inactive ports are selected and closed.

[0010] As a preferred implementation, the initial allocation calculation of member path selection probability specifically includes:

[0011] Perform routing calculation and generate an equal-cost multi-path group index, which includes M member path ports; assign member path port IDs according to the order in which the member path ports are added; after creating a new flow index, assign an initial member path selection probability, recorded as FP i , i is the member path port ID number; when i=M, FP M =(100-M+1)%; when i=1, 2, ..., M-1, FP i =1%.

[0012] As a preferred implementation scheme, the congestion update calculation of the member path selection probability specifically includes:

[0013] Determine whether the port i that receives the early congestion notification message satisfies i=M; if so, FP M =FP M -γ / 2、FP M-1 =FP M-1 +γ / 2, γ represents the proportion of early congestion notification messages; if not satisfied, FP i =FP i -γ / 2、FP M =FP M +γ / 2.

[0014] As a preferred implementation, the γ is updated periodically according to a timer, including: initializing γ=0; collecting statistical information of the port according to the timing cycle of the timer to obtain the latest collected data γ new ; According to γ=γ old *(1-α)+γ new *α, update calculation γ; where γ old is the data collected last time, α=0.02 is a constant, indicating the moving weighting coefficient.

[0015] As a preferred implementation, the packet update calculation of the member path selection probability specifically includes: judging whether the port i that successfully sends the message satisfies i=M; if so, no update is required; if not, FPi =FP i +1%, FP M =FP M -1%.

[0016] As a preferred implementation, when a member path port needs to be added or deleted, a port change update calculation of the member path selection probability is performed, including:

[0017] When member path port j is added, FP j =FP M -1%, M=j, FP M-1 =1%;

[0018] When member path port j is deleted, if j is not equal to M, then FP M =FP M +FP j , M=M-1; and for i>j, then FP i-1 =FP i , for i<j, then FP i No change; if j is equal to M, then FP j-1 =FP j-1 +FP j , M=j-1.

[0019] As a preferred implementation scheme, the inactive ports are selected and closed, specifically including: setting a port shutdown threshold K; when the member path selection probability corresponding to the port is less than or equal to K, the port is shut down; when all ports under the switch are shut down, the switch is shut down.

[0020] As a preferred embodiment, the method further comprises:

[0021] During the packet receiving process, when an early congestion notification message is received, a congested link rerouting message is generated; after receiving the congested link rerouting message, the flow on the congested link is triggered to readjust the routing, and the flow index is readjusted to the new member path port according to the member path port selection probability, and the new member path port corresponding to the flow index is recorded.

[0022] As a preferred implementation, during the packet sending process, the member paths in the current equal-cost multi-path group index are sorted according to their selection probabilities, and the path with the highest probability is selected for sending the traffic.

[0023] In a second aspect, an embodiment of the present invention further provides a data center network energy-saving device based on the method in the embodiment of the first aspect. The device includes an initial allocation calculation module, a congestion update calculation module, a packet update calculation module and a switch control module arranged in a switch;

[0024] The initial allocation calculation module is used to: perform initial allocation calculation of member path selection probability after the data center networking configuration is completed;

[0025] The congestion update calculation module is used to: during the packet receiving process, when an early congestion notification message is received, perform congestion update calculation of member path selection probability;

[0026] The packet sending update calculation module is used to: during the packet sending process, when the message is successfully sent, perform packet sending update calculation of the member path selection probability;

[0027] The switch control module is used to periodically detect the member path selection probability, select inactive ports and close them.

[0028] The beneficial effects of the present invention are:

[0029] (1) In the present invention, the member path selection probability can be calculated on the switch to dynamically select inactive links and inactive switches, and the inactive links and inactive switches can be shut down to achieve optimal utilization of the data center network bandwidth and achieve the purpose of optimal energy saving and consumption reduction. It has good economic benefits and meets practical application needs.

[0030] (2) During the packet receiving process, the present invention utilizes the active congestion detection mechanism in the network to realize millisecond-level burst traffic perception, and can timely reschedule traffic according to the member path selection probability after sensing congestion, thereby reducing the message transmission delay; it solves the problem that the prior art is difficult to cope with millisecond-level network burst traffic, resulting in errors in the calculation of link bandwidth requirements, which in turn affects the message transmission delay.

[0031] (3) In the present invention, the switch is controlled in a distributed manner, and during the packet sending process, the optimal egress link can be dynamically selected for packet sending according to the calculated member path selection probability, thereby solving the problem that the existing centralized control model has a large computational time complexity and cannot timely predict the real-time congested traffic in the network, and thus cannot select the optimal egress link for packet sending according to the link traffic. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a schematic diagram of an embodiment of the present invention applicable to a typical three-layer data center network;

[0033] Figure 2 A flowchart of a method for energy saving in a data center network according to an embodiment of the present invention;

[0034] Figure 3 Schematic diagram of ECMP routing from L1 to L2 in a three-tier data center according to an embodiment of the present invention;

[0035] Figure 4Flow chart of the packet receiving process in an embodiment of the present invention;

[0036] Figure 5 The figure is a flow chart of the packet issuing process in an embodiment of the present invention. DETAILED DESCRIPTION

[0037] In view of the many problems existing in the practical application of the traditional ECMP bandwidth calculation and energy-saving control mechanism of data centers in the prior art, for example, the energy-saving effect of the entire network is not ideal; it is difficult to cope with the network burst traffic at the millisecond level, affecting the message transmission delay; it is impossible to select the optimal export link for packet transmission according to the link traffic, etc. The present invention aims to provide a method and device for energy saving in a data center network, which at least solves one of the above technical problems to meet the actual application needs.

[0038] The main design ideas are as follows: when the data center networking configuration is completed, the initial allocation calculation of the member path selection probability is performed; during the packet receiving process, when the early congestion notification message is received, the congestion update calculation of the member path selection probability is performed; during the packet sending process, when the message is successfully sent, the packet sending update calculation of the member path selection probability is performed; the member path selection probability is periodically detected, and the inactive ports are selected and closed.

[0039] In the above main design scheme, the bandwidth utilization of the ECMP path can be dynamically calculated by calculating the member path selection probability of each distributed switch node (including the initial allocation calculation and various subsequent update calculations); at the same time, according to the calculated member path selection probability, inactive ports and inactive switches can be dynamically selected, and the switch control of the entire switch and the switch port can be realized to achieve the optimal planning of the energy consumption of the entire network, thereby solving the problem that the network energy saving effect in the prior art is not ideal.

[0040] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, the technical solutions of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0041] However, it should be noted that the examples to be introduced below are only some specific examples, and are not intended to limit the embodiments of the present invention to the following specific steps, values, conditions, data, sequences, etc. Those skilled in the art can apply the concept of the present invention to construct more embodiments not mentioned in this specification by reading this specification.

[0042] Embodiment 1

[0043] The technical solution provided in this embodiment can be applied to Figure 1 In the typical three-tier data center network shown in Figure 1, see Figure 1As shown, by applying the technical solution provided in this embodiment, inactive links and inactive switches can be dynamically selected through member path selection probability calculation on the switch, and inactive links and inactive switches can be shut down to achieve optimal utilization of data center network bandwidth and achieve optimal energy saving and consumption reduction.

[0044] Specifically, see Figure 2 As shown, this embodiment provides a data center network energy saving method, including the following steps:

[0045] A. After the data center networking configuration is completed, the initial allocation calculation of the member path selection probability is performed. It can be understood that a member path selection probability is designed in this embodiment to indicate the selection probability of the flow index (FLOW ID) at each member path port ID. By calculating the member path selection probability (including the initial allocation calculation and the subsequent update calculation) by each distributed switch node, the purpose of dynamically calculating the bandwidth utilization of the ECMP path can be achieved.

[0046] Exemplarily, as an optional implementation, in step A, the initial allocation calculation of the member path selection probability is performed, which specifically includes the following operations:

[0047] A1. Perform routing calculation and generate an ECMP Group ID, where the ECMP Group ID includes M member path ports. Figure 2 As an example, the ECMP routing diagram from L1 to L2 of the three-tier data center shown in the figure, where server 1 and server 4 communicate, the traffic path needs to pass through the L1 switch and the L2 switch, and there are 2 N equivalent paths (N=3), such as P1, P2...P7, P8; then, Figure 2 In the example, L1 switch-1 calculates the route to 200.1.1.0 / 24 and obtains ECMP Group ID = 1, which contains 8 member ports.

[0048] A2. According to the order in which the member path ports are added, member path port IDs are allocated, such as i (i=1, 2, ..., M). For example, Figure 2 In the example, if the ports corresponding to the path from P1 to P8 are added in sequence, then ID = 1, 2, 3, ... 8.

[0049] A3. After creating a new flow index (FLOW ID), the selection probability of the flow index (FLOW ID) on each member path port ID is assigned, that is, the member path selection probability (Flow Probability of i), recorded as FP i ; Among them, FP M=(100-M+1)%, FP i =1% (i=1, 2, ..., M-1). For example, Figure 2 In the example, L1 switch-1 creates a new flow from source IP=100.1.1.1 to destination IP=200.1.1.1, FLOW ID=1024, FP8=(100-8+1)%=93%, FP i =1% (i=1, 2, ..., 7).

[0050] B. During the packet receiving process, when an Early Congestion Notification (ECN) message is received, the congestion update calculation of the member path selection probability is performed.

[0051] Exemplarily, as a preferred implementation, in step B, the congestion update calculation of the member path selection probability is performed, which specifically includes the following operations:

[0052] Determine whether the port i receiving the early congestion notification (ECN) message satisfies i=M;

[0053] If satisfied, then FP M =FP M -γ / 2、FP M-1 =FP M-1 +γ / 2, γ represents the proportion of early congestion notification (ECN) messages;

[0054] If not satisfied, FP i =FP i -γ / 2、FP M =FP M +γ / 2.

[0055] Furthermore, in practical applications, the γ needs to be updated periodically according to a timer, specifically including:

[0056] (1) Initialize γ = 0;

[0057] (2) According to the timing cycle of the timer (the default timing cycle of the timer is 2ms), collect the statistical information of the port and obtain the latest collected data γ new ;

[0058] (3) According to γ ​​= γ old *(1-α)+γ new *α, update calculation γ; where γ old is the data collected last time, α=0.02 is a constant, indicating the moving weighting coefficient.

[0059] C. During the packet sending process, when the message is successfully sent, the packet sending update calculation of the member path selection probability is performed.

[0060] Exemplarily, as a preferred implementation, in step C, performing a packet update calculation of the member path selection probability specifically includes the following operations:

[0061] Determine whether the port i that successfully sent the message satisfies i=M;

[0062] If satisfied, then FP M No change, that is, no update is required under the current conditions;

[0063] If not satisfied, FP i =FP i +1%, FP M =FP M -1%.

[0064] Furthermore, in actual applications, member path ports may need to be added or deleted. Therefore, in this embodiment, when member path ports need to be added or deleted, port change update calculation of member path selection probability is also performed. Specifically, it includes:

[0065] When member path port j is added, FP j =FP M -1%, M=j, FP M-1 =1%;

[0066] When member path port j is deleted, if j is not equal to M, then FP M =FP M +FP j , M=M-1; and for i>j, then FP i-1 =FP i , that is, for all ports greater than j, their FP changes to the FP of the port ID one digit smaller; for i<j, the FP i No change;

[0067] If j is equal to M, then FP j-1 =FP j-1 +FP j , M=j-1.

[0068] In order to more clearly understand the various calculation processes of member path selection probabilities in this embodiment, the following describes the member path selection probabilities using Table 1 as an example. Figure 2 The change process of member path selection probability FP under different operation steps in the scenario.

[0069] Table 1

[0070]

[0071]

[0072] D. Periodically check the member path selection probability, select inactive ports and close them.

[0073] Exemplarily, as a preferred implementation, in step D, selecting an inactive port and shutting it down specifically includes the following operations:

[0074] (1) Setting the port shutdown threshold K; in practical applications, the value range of K is 1% to 10%, preferably K = 2%;

[0075] (2) When the member path selection probability corresponding to the port is less than or equal to K (that is, when the selection probability in the equal-cost multi-path group ECMP GROUP to which the port belongs is less than or equal to K), shut down the port;

[0076] (3) When all ports on the switch are shut down, shut down the switch. Figure 2 As shown, by detecting the member path selection probability, the inactive link P8 and the inactive switch L2 switch-8 are selected.

[0077] Furthermore, in actual applications, when the member path selection probability is periodically detected, if the member path selection probability corresponding to a closed port is greater than K, the port and the switch to which it is located will be reopened. In addition, it is understandable that in actual applications, when a port is closed or opened, the port change update calculation of the member path selection probability will be performed according to the situation where the member path port needs to be deleted or added. The specific update process is described above and will not be repeated here. At the same time, when a port is closed or opened, the Flow ID will be readjusted to the new member path port, and the new member path port corresponding to the Flow ID will be recorded.

[0078] As can be seen from the above content, in this embodiment, a member path selection probability is designed to indicate the selection probability of the flow index (FLOW ID) at each member path port ID. By calculating the member path selection probability through each distributed switch node, including the initial allocation calculation and subsequent various update calculations, the purpose of dynamically calculating the bandwidth utilization of the ECMP path can be achieved. In addition, by periodically detecting the member path selection probability, inactive ports and inactive switches can be dynamically selected, and inactive ports and inactive switches can be closed to achieve optimal utilization of the data center network bandwidth and achieve the purpose of optimal energy saving and consumption reduction.

[0079] Embodiment 2

[0080] This embodiment provides a method for energy saving in a data center network, and its basic steps are the same as those in the first embodiment, except that, as a preferred implementation method, in order to further solve the problem that the existing technology is difficult to cope with millisecond-level network burst traffic, resulting in errors in link bandwidth demand calculation, which in turn affects the message transmission delay, this embodiment, during the packet receiving process, uses the ECN message detection mechanism to achieve millisecond-level burst traffic perception, and can timely reschedule traffic according to the member path selection probability after sensing congestion, thereby reducing the message transmission delay. Specifically, step B of this embodiment also includes the following operations:

[0081] During the packet receiving process, when an Early Congestion Notification (ECN) message is received, a congested link rerouting message is generated;

[0082] After receiving the congested link rerouting message, the flow (FLOW) on the congested link will be triggered to readjust the route, that is, if the member path port link is congested, the Flow ID will be readjusted to the new member path port according to the member path port selection probability, and the new member path port corresponding to the Flow ID will be recorded.

[0083] It can be understood that during the packet receiving process, this embodiment utilizes the active congestion detection mechanism in the network to achieve millisecond-level burst traffic perception; after sensing congestion, it can timely reschedule traffic based on the member path selection probability, thereby reducing message transmission delay.

[0084] For example, see Figure 4 As shown, as an optional implementation, the packet receiving process may specifically include the following operations:

[0085] S101, the port receives the message and performs packet analysis;

[0086] S102, query the FLOW ID table according to the five-tuple of the message (source IP address, destination IP address, protocol type, source port, destination port); if there is a FLOW ID, obtain the FLOW ID; if not, generate a new FLOW ID and add it to the FLOW ID table;

[0087] S103, determining whether the ECN bit of the received message is set, if so, determining that an early congestion notification (ECN) message is received, and proceeding to step S104; otherwise, proceeding to step S106;

[0088] S104, performing congestion update calculation of member path selection probability and generating a congested link rerouting message;

[0089] S105, after receiving the congested link rerouting message, trigger the flow (FLOW) on the congested link to re-adjust the route;

[0090] S106: Execute other forwarding processing procedures of the inbound port, such as L2 service processing.

[0091] Embodiment 3

[0092] This embodiment provides a method for energy saving in a data center network, and its basic steps are the same as those of the second embodiment, except that, as a preferred implementation mode, in order to further solve the problem that the existing centralized control model calculation has a large calculation time complexity, cannot timely predict the real-time congested traffic in the network, and thus cannot select the optimal export link for packet sending according to the link traffic, this embodiment can dynamically select the optimal export link for packet sending according to the calculated member path selection probability during the packet sending process.

[0093] Specifically, in step C of this embodiment, during the packet sending process, the member paths in the current equal-cost multi-path group index (ECMP Group ID) will be sorted and the path with the highest probability will be selected to send the flow; at the same time, the message will be encapsulated by VXLAN, and the FLOW ID will be encapsulated into the Group ID field of the VXLAN header.

[0094] For example, see Figure 5 As shown, as an optional implementation, the package issuance process may specifically include the following operations:

[0095] S201, query the routing table to obtain the destination ECMP Group ID;

[0096] S202, sorting the paths according to the current member selection probabilities in the destination ECMP Group ID, and selecting the path with the highest probability to send the traffic;

[0097] S203, perform VXLAN encapsulation on the message, and encapsulate the FLOW ID into the Group ID field of the VXLAN header;

[0098] S204: After the message is successfully sent, a packet update calculation of the member path selection probability is performed.

[0099] Embodiment 4

[0100] Based on the same inventive concept, an embodiment of the present invention also provides a data center network energy-saving device, which includes an initial allocation calculation module, a congestion update calculation module, a packet update calculation module and a switch control module arranged in a switch.

[0101] The initial allocation calculation module is used to perform the initial allocation calculation of the member path selection probability after the data center networking configuration is completed. The congestion update calculation module is used to perform the congestion update calculation of the member path selection probability when the early congestion notification message is received during the packet receiving process. The packet sending update calculation module is used to perform the packet sending update calculation of the member path selection probability when the message is successfully sent during the packet sending process. The switch control module is used to periodically detect the member path selection probability, select the inactive port and close it.

[0102] It can be understood that, by using the device of this embodiment, the member path selection probability can be calculated through each distributed switch node, including the initial allocation calculation and subsequent various update calculations, so as to achieve the purpose of dynamically calculating the bandwidth utilization of the ECMP path. In addition, by periodically detecting the member path selection probability, inactive ports and inactive switches can be dynamically selected, and inactive ports and inactive switches can be closed, so as to achieve the optimal utilization of the data center network bandwidth and achieve the purpose of optimal energy saving and consumption reduction.

[0103] It should be noted that the various variations and specific examples in the aforementioned method embodiments are also applicable to the device of this embodiment. Through the detailed description of the aforementioned method, those skilled in the art can clearly understand the implementation method of the device of this embodiment, so for the sake of brevity of the specification, it will not be described in detail here.

[0104] Note: The above-mentioned specific embodiments are only examples rather than limitations, and those skilled in the art can, based on the concept of the present invention, merge and combine some steps and devices from the various embodiments described separately above to achieve the effects of the present invention. Such merged and combined embodiments are also included in the present invention, and such merges and combinations are not described one by one here.

[0105] The advantages, strengths, effects, etc. mentioned in the embodiments of the present invention are only examples, not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the above specific details disclosed in the embodiments of the present invention are only for the purpose of illustration and facilitating understanding, not limitation, and the above details do not limit the embodiments of the present invention to be implemented by adopting the above specific details.

[0106] The block diagrams of the devices, apparatuses, equipment, and systems involved in the embodiments of the present invention are only illustrative examples, and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagrams. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems may be connected, arranged, and configured in any manner. Words such as "including," "comprising," "having," and the like are open words, meaning "including but not limited to," and may be used interchangeably therewith. The words "or" and "and" used in the embodiments of the present invention refer to the words "and / or," and may be used interchangeably therewith, unless the context clearly indicates otherwise. The words "such as" used in the embodiments of the present invention refer to the phrase "such as but not limited to," and may be used interchangeably therewith.

[0107] The step flow charts and the above method descriptions in the embodiments of the present invention are only illustrative examples, and are not intended to require or imply that the steps of each embodiment must be performed in the order given. As will be appreciated by those skilled in the art, the order of the steps in the above embodiments can be performed in any order. Words such as "thereafter", "then", "next", etc. are not intended to limit the order of the steps; these words are only used to guide the reader through the description of these methods. In addition, any reference to a singular element, such as using the article "a", "an", or "the", is not to be construed as limiting the element to the singular.

[0108] In addition, the steps and devices in each embodiment of the present invention are not limited to be implemented in a certain embodiment. In fact, according to the concept of the present invention, relevant partial steps and partial devices in each embodiment of this document can be combined to conceive new embodiments, and these new embodiments are also included in the scope of the present invention.

[0109] Each operation in the embodiment of the present invention may be performed by any appropriate means capable of performing the corresponding functions, which may include various hardware and / or software components and / or modules, including but not limited to hardware circuits or processors.

[0110] The method of the embodiment of the present invention includes one or more actions for implementing the above-mentioned method. The method and / or action can be interchangeable with each other without departing from the scope of the claims. In other words, unless a specific order of actions is specified, the order and / or use of specific actions can be modified without departing from the scope of the claims.

[0111] The functions in the embodiments of the present invention can be implemented by hardware, software, firmware or any combination thereof. If implemented in software, the functions can be stored as one or more instructions on a tangible computer-readable medium. The storage medium can be any available tangible medium that can be accessed by a computer. By way of example and not limitation, such a computer-readable medium may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices or any other tangible medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer. As used herein, a disk and a disc include a compact disc (CD), a laser disc, an optical disc, a DVD (Digital Versatile Disc), a floppy disc and a blue disc, wherein the disc reproduces data magnetically, and the disc reproduces data optically using a laser.

[0112] Thus, a computer program product may perform the operations presented herein. For example, such a computer program product may be a computer-readable tangible medium having instructions tangibly stored (and / or encoded) thereon, which instructions may be executed by one or more processors to perform the operations described herein. The computer program product may include packaging materials.

[0113] Other examples and implementations are within the scope and spirit of the embodiments of the invention and the appended claims. For example, due to the nature of software, the functions described above can be implemented using software executed by a processor, hardware, firmware, hard wiring, or any combination of these. Features that implement the functions can also be physically located in various locations, including being distributed so that parts of the functions are implemented at different physical locations.

[0114] Various changes, substitutions and modifications of the techniques described herein may be made by those skilled in the art without departing from the teachings defined by the appended claims. In addition, the scope of the claims of the present disclosure is not limited to the specific aspects of the processes, machines, manufactures, components of events, means, methods and actions described above. Currently existing or later to be developed processes, machines, manufactures, components of events, means, methods or actions that perform substantially the same functions or achieve substantially the same results as the corresponding aspects described herein may be utilized. Thus, the appended claims include such processes, machines, manufactures, components of events, means, methods or actions within their scope.

[0115] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present invention. Therefore, the present invention is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0116] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present invention to the forms disclosed herein. Although a number of example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof. And the content not described in detail in this specification belongs to the prior art known to those skilled in the art.

Claims

1. A data center network energy saving method, characterized in that: The method includes: When the data center network configuration is completed, the initial allocation calculation of member path selection probability is performed; During the packet receiving process, when an early congestion notification message is received, the congestion update calculation of the member path selection probability is performed; During the packet sending process, when the message is successfully sent, the packet sending update calculation of the member path selection probability is performed; Periodically check the member path selection probability, select inactive ports and close them; The initial allocation calculation of member path selection probability specifically includes: Performing routing calculation and generating an equal-cost multi-path group index, wherein the equal-cost multi-path group index includes M member path ports; According to the order in which the member path ports are added, member path port IDs are allocated; After creating a new flow index, the initial member path selection probability is assigned, denoted as FP i , i is the member path port ID number; when i=M, FP M =(100-M+1)%; when i=1, 2, ..., M-1, FP i =1%.

2. The data center network energy saving method according to claim 1, characterized in that: The congestion update calculation of the member path selection probability specifically includes: Determine whether the port i receiving the early congestion notification message satisfies i=M; If satisfied, then FP M =FP M -γ / 2、FP M-1 =FP M-1 +γ / 2, γ represents the proportion of early congestion notification messages; If not satisfied, FP i =FP i -γ / 2、FP M =FP M +γ / 2.

3. The data center network energy saving method according to claim 2, characterized in that: The γ is periodically updated according to a timer, including: Initialize γ = 0; According to the timing cycle of the timer, collect the statistical information of the port and get the latest collected dataγ new ; According to γ=γ old *(1-α)+γ new *α, update calculation γ; where γ old is the data collected last time, α=0.02 is a constant, indicating the moving weighting coefficient.

4. The data center network energy saving method according to claim 1, characterized in that: The packet updating calculation of the member path selection probability specifically includes: Determine whether the port i that successfully sent the message satisfies i=M; If satisfied, no update is required; If not satisfied, FP i =FP i +1%, FP M =FP M -1%.

5. The data center network energy saving method according to claim 1, characterized in that: When a member path port needs to be added or deleted, the port change update calculation of the member path selection probability is performed, including: When member path port j is added, FP j =FP M -1%, M=j, FP M-1 =1%; When member path port j is deleted, if j is not equal to M, then FP M =FP M +FP j , M=M-1; and for i>j, then FP i-1 =FP i , for i<j, then FP i No change; If j is equal to M, then FP j-1 =FP j-1 +FP j , M=j-1.

6. The data center network energy saving method according to claim 1, characterized in that: The selecting and closing of inactive ports specifically includes: Set the port shutdown threshold K; When the member path selection probability corresponding to the port is less than or equal to K, shut down the port; When all ports on the switch are shut down, the switch is shut down.

7. The data center network energy saving method according to claim 1, characterized in that: The method further includes: During the packet receiving process, when an early congestion notification message is received, a congested link rerouting message is generated; After receiving the congested link rerouting message, the flow on the congested link is triggered to readjust the routing, the flow index is readjusted to the new member path port according to the member path selection probability, and the new member path port corresponding to the flow index is recorded.

8. The data center network energy saving method according to claim 1, characterized in that: In the packet sending process, the member paths in the current equal-cost multi-path group index are sorted according to their selection probabilities, and the path with the maximum probability is selected for sending the flow.

9. A data center network energy saving device based on the method according to any one of claims 1 to 8, characterized in that: The device comprises an initial allocation calculation module, a congestion update calculation module, a packet update calculation module and a switch control module arranged in the switch; The initial allocation calculation module is used to: perform initial allocation calculation of member path selection probability after the data center networking configuration is completed; The congestion update calculation module is used to: during the packet receiving process, when an early congestion notification message is received, perform congestion update calculation of member path selection probability; The packet sending update calculation module is used to: during the packet sending process, when the message is successfully sent, perform packet sending update calculation of the member path selection probability; The switch control module is used to: periodically detect the member path selection probability, select inactive ports and close them; The initial allocation calculation module performs initial allocation calculation of member path selection probability, specifically including: Performing routing calculation and generating an equal-cost multi-path group index, wherein the equal-cost multi-path group index includes M member path ports; According to the order in which the member path ports are added, member path port IDs are allocated; After creating a new flow index, the initial member path selection probability is assigned, denoted as FP i , i is the member path port ID number; when i=M, FP M =(100-M+1)%; when i=1, 2, ..., M-1, FP i =1%.

Images