A data center network dynamic energy-saving scheduling method and system under SDN environment

By detecting the flow information type in the SDN environment and using key interval priority algorithms and static load balancing mechanisms to differentiate the elephant flow and mouse flow, the problem of ineffective energy saving in the existing technology is solved, and the resource energy of the data center network is realized Reduced consumption and improved computing efficiency.

CN114398174BActive Publication Date: 2024-11-08NANJING UNIV OF POSTS & TELECOMM
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111633860.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-29
Publication Date
2024-11-08
Estimated Expiration
2041-12-29

AI Technical Summary

Technical Problem

In the absence of prediction, the prior art cannot effectively realize the energy-saving scheduling of the data center network through static load balancing methods, especially the scheduling of elephant streams and mouse streams cannot achieve the best energy-saving effect.

Method used

In the SDN environment, the flow information type is detected through the switch and the SDN controller, and the key interval priority algorithm is used for dynamic scheduling for elephant flow, and the mouse flow is scheduled in combination with the static load balancing mechanism to realize differentiated scheduling of elephant flow and mouse flow. , reduce the energy consumption of data center network resources.

Benefits of technology

By distinguishing between elephant flow and mouse flow, a reasonable scheduling scheme is adopted, the energy consumption of the data center network is significantly reduced, computing efficiency is improved, and the number of flow entries dynamically scheduled is reduced.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114398174B_ABST
    Figure CN114398174B_ABST
Patent Text Reader

Abstract

The present invention discloses a data center network dynamic energy-saving scheduling method under an SDN environment, comprising: (1) a switch forwards flow information to a data center network, and forwards the flow information to an SDN controller; (2) the SDN controller detects the received flow information, and if it is a mouse flow, it is scheduled using a static load balancing mechanism; if it is an elephant flow, it enters step (3); (3) for the received elephant flow, it is scheduled together with the elephant flow that has not been fully transmitted before, and a critical interval priority algorithm is used to calculate the scheduling order and speed for the elephant flow in the set; (4) the SDN controller issues a flow table according to the scheduling result, and issues the elephant flow and the mouse flow to corresponding links to achieve energy-saving scheduling. The present invention reasonably distinguishes the elephant flow and the mouse flow through the SDN controller, reduces the energy consumption of data center network resources, greatly reduces the number of flow entries for dynamic scheduling, and improves the calculation efficiency of the SDN controller.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of data center network traffic scheduling, and in particular to a data center network dynamic energy-saving scheduling method and system under an SDN environment. Background Art

[0002] The current data center network traffic characteristics show the distribution characteristics of elephant flows and mouse flows. The elephant flows include video, download transmission, games, etc., and the mouse flows include instant text communication services, e-commerce, etc. Mouse flows have the characteristics of large number and small bandwidth occupation, while elephant flows have the characteristics of small number and large bandwidth occupation. Currently, the scheduling of the two flows generally adopts a static load balancing method for scheduling, but in the case of unpredictable conditions, it cannot achieve effective energy-saving scheduling. Summary of the invention

[0003] Purpose of the invention: In view of the above problems, the purpose of the present invention is to provide a dynamic energy-saving scheduling method for a data center network in an SDN environment, which improves the processing speed of the network center by using different scheduling schemes for elephant flows and mouse flows; another purpose of the present invention is to provide a scheduling system for implementing the above-mentioned dynamic energy-saving scheduling method for a data center network.

[0004] Technical solution: A data center network dynamic energy-saving scheduling method in an SDN environment of the present invention comprises:

[0005] (1) The switch forwards the flow information to the data center network. The data center network analyzes the capacity information of each link and then forwards the flow information to the SDN controller. The SDN controller establishes the network topology structure based on the data information.

[0006] (2) The SDN controller detects the received flow information. If it is a mouse flow, it uses the static load balancing mechanism to schedule it; if it is an elephant flow, it goes to step (3);

[0007] (3) For the received elephant flows, schedule them together with the elephant flows that have not been transmitted before, and modify the relevant information of the elephant flows that have not been transmitted before. For the flow information arriving at the current moment and the flow at the previous moment that recalculates the amount of data to be transmitted, a unified flow scheduling scheme is designed, so as to optimally schedule the elephant flows in real time and dynamically, and use the critical interval priority algorithm to calculate the scheduling order and speed for the elephant flows in the set;

[0008] (4) The SDN controller issues a flow table based on the scheduling results and distributes the elephant flow and mouse flow to the corresponding links to achieve energy-saving scheduling.

[0009] Further, the step of using the critical interval priority algorithm to calculate the scheduling order and speed for the elephant flows in the set includes:

[0010] (301) All the elephant flows J on each link e e (t0) Sort by deadline, J e (t0) is the set of elephant flows arriving at the data center network, expressed as:

[0011]

[0012]

[0013] in Represents the flow The source and destination nodes, Represents the start time and the latest end time of the flow;

[0014] (302) Divide the time into different intervals I according to the cut-off time points, and merge the same cut-off time points;

[0015] (303) Traverse all intervals on all links and calculate the energy density δ(I,e) on each interval, which is expressed as:

[0016]

[0017] In the formula, δ max (I,e) is the maximum energy consumption density, a~b is the available time in the interval [a,b], w i ′ is the virtual weight, w i ′=w i ·(|P i |) 1 / τ ,j i ∈J, is the set of links that the flow transmission path passes through, and τ is a constant related to the link type;

[0018] (304) Find the maximum energy consumption density δ max The interval corresponding to (I,e) is defined as the key interval I * =[a,b], the key interval is the interval with the maximum energy density;

[0019] (305) Schedule the flows in the critical interval in the order of earliest deadline first, calculate the minimum transmission rate of the flow, and set the interval corresponding to the minimum transmission rate as the transmission interval Scheduling of elephant flows;

[0020] (306) The elephant flows and time intervals for which the scheduling scheme has been configured are removed, and new key intervals are iteratively searched until all flows are transmitted.

[0021] Further, the step of the SDN controller detecting the received flow information includes:

[0022] Calculate the total transmission capacity of the flow information The expression is:

[0023]

[0024] in, The information flow calculated at time t0 Transmission rate;

[0025] According to the required traffic size, a capacity threshold x is predefined. If there is If It is the elephant flow.

[0026] Furthermore, after the mouse flow is scheduled through the static load balancing mechanism, the mouse flow is deleted from the set.

[0027] A data center network dynamic energy-saving scheduling system in an SDN environment of the present invention comprises:

[0028] A switch module is used to forward flow information into the data center network and receive flow information from the data center network;

[0029] A receiving module, used for receiving flow information sent to the data center network;

[0030] The SDN controller module is used to determine the type of flow information and send the flow information to the corresponding scheduling module;

[0031] Static load balancing calculation module, used to schedule the mouse flow sent by the SDN controller;

[0032] Dynamic energy-saving flow scheduling module, used to schedule the elephant flow sent by the SDN controller;

[0033] The stream sending module is used to send the scheduled stream information to the corresponding link.

[0034] Beneficial effect: Compared with the prior art, the present invention has the following significant advantages: the present invention reasonably distinguishes between elephant flows and mouse flows through the SDN controller, adopts a key interval priority scheduling scheme for elephant flows, and adopts the original static scheduling scheme for mouse flows, and utilizes the energy-saving effects of the two scheduling schemes to reduce the energy consumption of data center network resources, greatly reduce the number of dynamically scheduled flow entries, and improve the computing efficiency of the SDN controller. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a flow chart of the energy-saving scheduling method of the present invention;

[0036] Figure 2 Scheduling flow chart for elephant flow;

[0037] Figure 3 is a topological diagram of an embodiment;

[0038] Figure 4 This is a flow scheduling diagram;

[0039] Figure 5 It is a structural schematic diagram of the energy-saving scheduling system of the present invention. DETAILED DESCRIPTION

[0040] The flow chart of a data center network dynamic energy-saving scheduling method in an SDN environment described in this embodiment is as follows: Figure 1 As shown, the scheduling method includes the following steps:

[0041] (1) The switch forwards the flow information to the data center network. The data center network analyzes the capacity information of each link and then forwards the flow information to the SDN controller. The SDN controller establishes the network topology structure based on the data information.

[0042] (2) The SDN controller detects the received flow information. If it is a rat flow, it is scheduled using the static load balancing mechanism. After the rat flow is scheduled by the static load balancing mechanism, the rat flow is deleted from the set; if it is an elephant flow, it goes to step (3);

[0043] The steps to detect whether it is an elephant flow include:

[0044] Calculate the total transmission capacity of the flow information The expression is:

[0045]

[0046] in, The information flow calculated at time t0 Transmission rate;

[0047] According to the required traffic size, a capacity threshold x is predefined. If there is If It is the elephant flow.

[0048] (3) For the received elephant flow, schedule it together with the elephant flow that has not been transmitted before, and modify the relevant information of the elephant flow that has not been transmitted before. For the flow information arriving at the current moment and the flow at the previous moment that recalculates the amount of data to be transmitted, a unified flow scheduling scheme is designed, so as to dynamically and optimally schedule the elephant flow in real time. The critical interval priority algorithm is used to calculate the scheduling order and speed for the elephant flows in the set. The flow chart is shown in the following figure. Figure 2 As shown:

[0049] (301) All the elephant flows J on each link e e (t0) Sort by deadline, J e (t0) is the set of elephant flows arriving at the data center network, expressed as:

[0050]

[0051]

[0052] in Represents the flow The source and destination nodes, Representative Flow The start time and the latest end time;

[0053] (302) Divide the time into different intervals I according to the cut-off time points, and merge the same cut-off time points;

[0054] (303) Traverse all intervals on all links and calculate the energy density δ(I,e) on each interval, which is expressed as:

[0055]

[0056] In the formula, δ max (I,e) is the maximum energy consumption density, a~b is the available time in the interval [a,b], w i ′ is the virtual weight, w i ′=w i ·(|P i |) 1 / τ ,j i ∈J, is the set of links that the flow transmission path passes through, and τ is a constant related to the link type;

[0057] (304) Find the maximum energy consumption density δ max The interval corresponding to (I,e) is defined as the key interval I * =[a,b], the key interval is the interval with the maximum energy density;

[0058] (305) Schedule the flows in the critical interval in the order of earliest deadline first, calculate the minimum transmission rate of the flow, and set the interval corresponding to the minimum transmission rate as the transmission interval Scheduling of elephant flows;

[0059] (306) The elephant flows and time intervals for which the scheduling scheme has been configured are removed, and new key intervals are iteratively searched until all flows are transmitted.

[0060] (4) The SDN controller issues a flow table based on the scheduling results and distributes the elephant flow and mouse flow to the corresponding links to achieve energy-saving scheduling.

[0061] The effect of scheduling elephant flows using a key interval priority scheduling scheme is described in detail below in conjunction with an embodiment.

[0062] Establish a network topology diagram based on the data information between data streams, such as Figure 3 As shown in the figure, there are 5 switches Switch1-5 and 6 links Link1-6. In a long period of time, according to the statistical law of data center network traffic, specific switches and related ports are opened. Figure 3 The switches and links in the figure are actually active devices. The set of active switches is represented as {s1, s2, s3, s5}, and the set of active links is represented as {l1, l3, l4}. Assume that there are currently three flow information transmitted to the data center network through the switch. The three flow information are represented as J1 = [1, 3, 0, 4, 4], J2 = [2, 5, 1, 3, 6], and J3 = [3, 5, 1, 4, 15]. All three flow information are transmitted using a single route. The path link sets of the three flow information are represented as P1 = {l1, l3}, P2 = {l4}, and P3 = {l3, l4}, where P1 = {l1, l3} indicates that the first flow information passes through links Link1 and Link3, P2 = {l4} indicates that the second flow information passes through link Link4, and P3 = {l3, l4} indicates that the third flow information passes through links Link3 and Link4. The time range of the above three flow information is set to 5 unit time, and each link contains 11 unit bandwidth. The dynamic scheduling mechanism and the static scheduling mechanism are applied to the scheduling of the elephant flow respectively, such as Figure 4As shown in the figure, Scheme 1 and Scheme 2 represent dynamic scheduling mechanisms, and Scheme 3 represents a static scheduling mechanism. It is assumed that the flow speed is a constant in each calculation, that the flow transmission rate is an integer multiple of the unit bandwidth, and that each flow needs to be forwarded before the deadline in a shared link manner. According to the link rate energy consumption model, it can be calculated that the energy consumed by Scheme 1 is 15σ+278μ, the energy consumed by Scheme 2 is 15σ+266μ, and the energy consumed by Scheme 3 is 15σ+264μ. The static flow scheduling mechanism consumes the least energy because it knows the information of all flows in advance, and assumes that the flow information prediction is accurate, so it can make more energy-saving scheduling decisions. Although the dynamic shared flow scheduling mechanism has a slightly increased energy consumption, it does not need to predict the information of other flows. The energy consumption of Scheme 2 is lower than that of Scheme 1 because the transmission of flows is more balanced. From the above three transmission mechanisms, we can see that when the traffic cannot be accurately predicted, energy consumption can be effectively reduced by adjusting the transmission rate and transmission interval of the flow to make it more evenly transmitted on the link.

[0063] The present invention provides a data center network dynamic energy-saving scheduling system under an SDN environment, and the structural diagram is as follows: Figure 5 As shown, the system includes:

[0064] A switch module is used to forward flow information into the data center network and receive flow information from the data center network;

[0065] A receiving module, used for receiving flow information sent to the data center network;

[0066] The SDN controller module is used to determine the type of flow information and send the flow information to the corresponding scheduling module;

[0067] Static load balancing calculation module, used to schedule the mouse flow sent by the SDN controller;

[0068] Dynamic energy-saving flow scheduling module, used to schedule the elephant flow sent by the SDN controller;

[0069] The stream sending module is used to send the scheduled stream information to the corresponding link.

Claims

1. A data center network dynamic energy-saving scheduling method in an SDN environment, characterized in that: The scheduling method includes: (1) The switch forwards the flow information to the data center network. The data center network analyzes the capacity information of each link and then forwards the flow information to the SDN controller. The SDN controller establishes the network topology structure based on the data information. (2) The SDN controller detects the received flow information. If it is a mouse flow, it uses the static load balancing mechanism to schedule it; if it is an elephant flow, it goes to step (3); (3) For the received elephant flows, schedule them together with the elephant flows that have not been fully transmitted before, and modify the relevant information of the elephant flows that have not been fully transmitted before. Use the key interval priority algorithm to calculate the scheduling order and speed for the elephant flows in the set; (4) The SDN controller issues a flow table based on the scheduling results, and distributes the elephant flow and mouse flow to the corresponding links to achieve energy-saving scheduling; The step of using the key interval priority algorithm to calculate the scheduling order and speed for the elephant flow in the set includes: (301) All the elephant flows J on each link e e (t0) Sort by deadline, J e (t0) is the set of elephant flows arriving at the data center network, expressed as: in Represents the flow The source and destination nodes, Representative Flow The start time and the latest end time, Represents the size of the flow; (302) Divide the time into different intervals I according to the cut-off time points, and merge the same cut-off time points; (303) Traverse all intervals on all links and calculate the energy density δ(I,e) on each interval, which is expressed as: In the formula, δ max (I,e) is the maximum energy consumption density, t0 represents the current time, t end represents the latest deadline of all flows, a~b is the available time in the interval [a,b], w i ′ is the virtual weight, w i ′=w i ·(|P i |) 1 / τ ,j i ∈J,P (t0,i) is the set of links that the flow transmission path passes through, τ is a constant related to the link type, and w i is the size of the i-th flow; (304) Find the maximum energy consumption density δ max The interval corresponding to (I,e) is defined as the key interval I * =[a,b], the key interval is the interval with the maximum energy density; (305) Schedule the flows in the critical interval in the order of earliest deadline first, calculate the minimum transmission rate of the flow, and set the interval corresponding to the minimum transmission rate as the transmission interval Scheduling of elephant flows; (306) The elephant flows and time intervals for which the scheduling scheme has been configured are removed, and new key intervals are iteratively searched until all flows are transmitted.

2. The scheduling method according to claim 1, characterized in that: The step of the SDN controller detecting the received flow information includes: Calculate the total transmission capacity of the flow information The expression is: in, The information flow calculated at time t0 Transmission rate; According to the required traffic size, a capacity threshold x is predefined. If there is If It is the elephant flow.

3. The scheduling method according to claim 1, characterized in that: When the mouse flow is scheduled through the static load balancing mechanism, the mouse flow is deleted from the set.

4. A system for implementing the scheduling method according to any one of claims 1 to 3, characterized in that: The system includes: A switch module is used to forward flow information into the data center network and receive flow information from the data center network; A receiving module, used for receiving flow information sent to the data center network; The SDN controller module is used to determine the type of flow information and send the flow information to the corresponding scheduling module; Static load balancing calculation module, used to schedule the mouse flow sent by the SDN controller; Dynamic energy-saving flow scheduling module, used to schedule the elephant flow sent by the SDN controller; The stream sending module is used to send the scheduled stream information to the corresponding link.