WO2020199467A1 - 网络主机间通信负载调控方法、电子装置及可读存储介质 - Google Patents

网络主机间通信负载调控方法、电子装置及可读存储介质 Download PDF

Info

Publication number
WO2020199467A1
WO2020199467A1 PCT/CN2019/102185 CN2019102185W WO2020199467A1 WO 2020199467 A1 WO2020199467 A1 WO 2020199467A1 CN 2019102185 W CN2019102185 W CN 2019102185W WO 2020199467 A1 WO2020199467 A1 WO 2020199467A1
Authority
WO
WIPO (PCT)
Prior art keywords
host node
host
node pair
similarity
connections
Prior art date
Application number
PCT/CN2019/102185
Other languages
English (en)
French (fr)
Inventor
刘洪晔
Original Assignee
平安科技（深圳）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 平安科技（深圳）有限公司 filed Critical 平安科技（深圳）有限公司
Publication of WO2020199467A1 publication Critical patent/WO2020199467A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00—Network arrangements or protocols for supporting network services or applications
    • H04L67/01—Protocols
    • H04L67/10—Protocols in which an application is distributed across nodes in the network
    • H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers

Definitions

  • This application relates to the field of cloud monitoring technology, and in particular to a method for controlling communication load between network hosts, an electronic device, and a readable storage medium.
  • the prior art In order to prevent the communication load of the host nodes in the cloud platform from being too large due to too many connections, the prior art generally adopts the method of setting thresholds to adjust the communication resources of the host nodes in the cloud platform, that is, set the maximum value uniformly.
  • the connection threshold When a host node in the cloud platform exceeds the maximum connection threshold, the host node is processed, for example, the work of the host node is suspended.
  • the existing method of monitoring the load adjustment timing of the host node by setting the maximum connection threshold is a very extensive load monitoring adjustment solution, which cannot accurately balance the waste of system resources and excessive system load.
  • the uniformly set maximum connection threshold is set relatively low, it will cause the host node to reserve more processing space, and it will cause a great waste of system resources; if the uniformly set maximum connection threshold is set If it is higher, it may cause the host node system to be overloaded and cause system abnormality or slow response.
  • the prior art also predicts the number of connections between host nodes by considering the business relationship, so as to provide early warning of possible high communication load conditions, due to frequent business changes and the consideration of reference items are usually relatively single, It is often impossible to accurately predict the number of connectable loads between host nodes, which often results in misjudgments.
  • the main purpose of this application is to provide a method, device, and storage medium for controlling communication load between network hosts to more accurately monitor the communication load of the cloud platform communication network and take timely and effective adjustment measures to optimize network communication Load resource configuration.
  • the present application provides a method for controlling communication load between network hosts.
  • the method includes:
  • the first calculation step regularly adjust the communication load between the network hosts. After the communication load adjustment between the network hosts is performed, determine the host node pairs that are not communicatively connected between the included host nodes. After determining a host node pair, obtain each The preset distance corresponds to the number of other host nodes communicatively connected to each host node of the host node pair, and the number corresponding to each preset distance is substituted into the first variable conversion formula to calculate the first similarity corresponding to the host node pair Degree parameter
  • the second calculation step respectively obtain the connection status of each predetermined database and each host node of the host node pair, and convert the connection status of each predetermined database and each host node of the host node pair into a corresponding adjacency matrix, Substituting the adjacency matrix into a predetermined second variable conversion formula to calculate a second similarity parameter corresponding to the host node pair;
  • the third calculation step input the calculated first similarity parameter and the second similarity parameter into a predetermined comprehensive similarity calculation formula to calculate the corresponding comprehensive similarity of the host node pair;
  • Adjustment step According to the calculated comprehensive similarity, the communication load of the included host node is adjusted for the corresponding host node.
  • the present application also provides an electronic device, the electronic device includes a memory and a processor, the memory stores a communication load control program that can run on the processor, the information load control program When executed by the processor, the following steps are implemented:
  • the first calculation step regularly adjust the communication load between the network hosts. After the communication load adjustment between the network hosts is performed, determine the host node pairs that are not communicatively connected between the included host nodes. After determining a host node pair, obtain each The preset distance corresponds to the number of other host nodes communicatively connected to each host node of the host node pair, and the number corresponding to each preset distance is substituted into the first variable conversion formula to calculate the first similarity corresponding to the host node pair Degree parameter
  • the second calculation step respectively obtain the connection status of each predetermined database and each host node of the host node pair, and convert the connection status of each predetermined database and each host node of the host node pair into a corresponding adjacency matrix, Substituting the adjacency matrix into a predetermined second variable conversion formula to calculate a second similarity parameter corresponding to the host node pair;
  • the third calculation step input the calculated first similarity parameter and the second similarity parameter into a predetermined comprehensive similarity calculation formula to calculate the corresponding comprehensive similarity of the host node pair;
  • Adjustment step According to the calculated comprehensive similarity, the communication load of the included host node is adjusted for the corresponding host node.
  • the present application also provides a computer-readable storage medium that includes a communication load control program between network hosts, and the communication load control program between network hosts is executed by a processor to realize the aforementioned network The steps of the method for controlling communication load between hosts.
  • this application regulates the communication load between network hosts at regular intervals. After a host node pair is determined, the other host nodes corresponding to each preset distance and each host node of the host node pair are all communicatively connected.
  • To form the first similarity parameter obtain the connection status of each predetermined database and each host node of the host node pair to form the second similarity parameter, pass the first similarity parameter and the second similarity parameter through Corresponding conversion to obtain a comprehensive similarity, according to the comprehensive similarity for the corresponding host node to control the communication load of the included host nodes, in order to more accurately and effectively control the host node communication load and balance the network resource configuration.
  • FIG. 1 is a hardware structure diagram of an embodiment of an electronic device for implementing communication load control between network hosts in this application;
  • FIG. 2 is a functional module diagram of an embodiment of a communication load control program between network hosts in FIG. 1;
  • FIG. 3 is an implementation flowchart of an embodiment of a method for controlling communication load between network hosts according to this application;
  • FIG. 4 is a schematic diagram of an application environment of an embodiment of a method for controlling communication load between network hosts according to the present application.
  • the electronic device 1 includes a memory 11, a processor 12, and a network interface 13.
  • the memory 11 stores a communication load control program 10 between network hosts that can be executed by the processor 12.
  • the electronic device 1 may be a terminal device with storage and computing functions, such as a server, a smart phone, a tablet computer, a portable computer, and a desktop computer.
  • the server when the electronic device 1 is a server, the server may be one or more of a rack server, a blade server, a tower server, or a cabinet server.
  • the memory 11 includes at least one type of readable storage medium.
  • the at least one type of readable storage medium may be a non-volatile storage medium such as flash memory, hard disk, multimedia card, card-type memory, and the like.
  • the readable storage medium may be an internal storage unit of the electronic device 1, such as a hard disk of the electronic device 1.
  • the readable storage medium may also be the external memory 11 of the electronic device 1, such as a plug-in hard disk or a smart memory card (Smart Media Card, SMC) equipped on the electronic device 1. , Secure Digital (SD) card, Flash Card (Flash Card), etc.
  • SD Secure Digital
  • Flash Card Flash Card
  • the processor 12 may be a central processing unit (CPU), a microprocessor or any other applicable data processing chip, and is used to run program codes or process data stored in the memory 11, for example, perform communication between network hosts Load control program 10 etc.
  • CPU central processing unit
  • microprocessor any other applicable data processing chip
  • the network interface 13 may include a standard wired interface and a wireless interface (such as a Wi-Fi interface). It is usually used to establish a common connection between the electronic device 1 and other electronic devices.
  • FIG. 1 only shows the electronic device 1 with components 11-13 and the communication load control program 10 between network hosts, but it should be understood that the electronic device 1 may include more or fewer components.
  • the processor 12 implements the following steps when executing the communication load control program 10 between network hosts stored in the memory 11:
  • the first calculation step regularly adjust the communication load between the network hosts. After the communication load adjustment between the network hosts is performed, determine the host node pairs that are not communicatively connected between the included host nodes. After determining a host node pair, obtain each The preset distance corresponds to the number of other host nodes communicatively connected to each host node of the host node pair, and the number corresponding to each preset distance is substituted into the first variable conversion formula to calculate the first similarity corresponding to the host node pair Degree parameter
  • the second calculation step respectively obtain the connection status of each predetermined database and each host node of the host node pair, and convert the connection status of each predetermined database and each host node of the host node pair into a corresponding adjacency matrix, Substituting the adjacency matrix into a predetermined second variable conversion formula to calculate a second similarity parameter corresponding to the host node pair;
  • the third calculation step input the calculated first similarity parameter and the second similarity parameter into a predetermined comprehensive similarity calculation formula to calculate the corresponding comprehensive similarity of the host node pair;
  • Adjustment step According to the calculated comprehensive similarity, the communication load of the included host node is adjusted for the corresponding host node.
  • FIG. 2 it is a functional module diagram of an embodiment of the communication load control program 10 between network hosts in FIG. 1.
  • the communication load control program 10 between the network hosts is divided into a plurality of functional modules, which are stored in the memory 11 and executed by the processor 12, so as to more accurately and effectively control the communication load of host nodes in a timely and effective manner.
  • Balance network resource allocation The “module” referred to in this application refers to a series of computer program instruction sets capable of completing specific functions.
  • the communication load control program 10 between network hosts is divided into: a first calculation module 100, a second calculation module 110, a third calculation module 120, and a control module 130.
  • the communication load control program 10 between the network hosts is divided into a first calculation module 100, a second calculation module 110, a third calculation module 120, and a control module 130, just for more details. It clearly expresses the functions that the communication load control program 10 can achieve, and is not used to limit the communication load control program 10 between network hosts that can or must be divided into a first calculation module 100, a second calculation module 110, and a second calculation module.
  • the third calculation module 120 and the control module 130 for those skilled in the art, in other embodiments, the communication load control program 10 between the network hosts can be easily divided into functional modules different from this embodiment. Do repeat.
  • the first calculation module 100 is configured to perform communication load adjustment between network hosts at regular intervals (for example, every 10 minutes), and after performing communication load adjustment between network hosts, determine the host nodes included in the host nodes that are not connected to each other. Yes, after determining a host node pair (for example, host node I 1 and host node I 2 ), respectively obtain the number of other host nodes that are in communication with each host node of the host node pair corresponding to each preset distance (for example, if the host node I 1, I 2 a distance between the connecting link comprises 2 I 1 -BI 2, I 1 -CI 2, I 1 -DI 2 that three (B, C, D are three Host node), then the number of other host nodes that are in communication with each host node I 1 and I 2 of the host node pair corresponding to distance 2 is 3; if the distance between the host node pair I 1 and I 2 is 3
  • the connection link includes I 1 -E 1 -E 2 -I 2 , I 1 -F 1
  • the second calculation module 110 is configured to obtain the connection status of each predetermined database (for example, database U 1 , U 2 , U 3 , U 4 ) and each host node of the host node pair I 1 , I 2 ( for example, I 1 host nodes and databases U 3 connection is indicated with 1, no connection is represented by 0), the respective predetermined database of the host node I 1, I each host node 2 I 1 or I 2
  • the connection state of is converted into a corresponding adjacency matrix, and the adjacency matrix is substituted into a predetermined second variable conversion formula to calculate the second similarity parameter S 2 corresponding to the host node pair.
  • the third calculation module 120 is configured to input the calculated first similarity parameter S 1 and the second similarity parameter S 2 into a predetermined comprehensive similarity calculation formula to calculate the corresponding host node pair The comprehensive similarity S.
  • the control module 130 is configured to control the communication load of the included host node for the corresponding host node according to the calculated comprehensive similarity S.
  • the adjustment module 130 adjusts the communication load of the included host node for the corresponding host node according to the calculated comprehensive similarity S, including: if one is determined If the comprehensive similarity corresponding to the host node pair I 1 and I 2 is greater than the preset threshold (for example, the preset threshold is 0.15), the host is determined according to the first mapping relationship between the predetermined comprehensive similarity interval and the maximum number of host connections The node pairs the corresponding maximum number of host connections (for example, the maximum number of host connections is 100), and according to the determined maximum number of host connections, controls the number of connections of other host nodes connected to the host node in the host node pair, so that The number of connections of other host nodes connected to the host node in the host node pair is less than or equal to the determined maximum number of host connections.
  • the preset threshold for example, the preset threshold is 0.15
  • the step of adjusting the communication load of the included host node for the corresponding host node by the control module 130 according to the calculated comprehensive similarity S includes:
  • the comprehensive similarity is sorted in descending order, and the preset order (for example, the top 10) is determined.
  • the elastically adjustable host node pairs corresponding to the similarity (for example, the numbers of the elastically adjustable host node pairs are I 1 and I 2 , I 5 and I 6 , I 7 and I 8 , I 12 and I 13 , I 15 and I 16 , I 17 and I 18 , I 19 and I 20 , I 22 and I 23 , I 25 and I 26 , I 37 and I 38 );
  • the number of flexible host connections corresponding to each flexible controllable host node pair finds out the number of flexible host connections corresponding to each flexible controllable host node pair (for example, the number of flexible host connections corresponding to I 1 and I 2 is 30.
  • the number of flexible host connections corresponding to I 5 and I 6 is 35
  • the number of flexible host connections corresponding to I 7 and I 8 is 32
  • the number of flexible host connections corresponding to I 12 and I 13 is 33
  • I 15 and I 16 correspond to
  • the number of elastic host connections corresponding to I 17 and I 18 is 37
  • the number of elastic host connections corresponding to I 19 and I 20 is 39
  • the number of elastic host connections corresponding to I 22 and I 23 is 36.
  • the number of flexible host connections corresponding to I 25 and I 26 is 34
  • the number of flexible host connections corresponding to I 37 and I 38 is 31);
  • the number of connections of other host nodes connected to the host node in each elastically adjustable host node pair is controlled in accordance with the corresponding number of elastic host connections, so that the host nodes of each elastically adjustable host node pair are connected to each other.
  • the number of connections of other host nodes (for example, the number of elastic host connections corresponding to I 1 and I 2 is 30, and the number of connections of other host nodes connected to I 1 is 25) is less than or equal to the corresponding number of elastic host connections.
  • the second variable conversion formula is:
  • B is the adjacency matrix
  • i represents the i-th row of the adjacency matrix
  • k represents the k-th column of the adjacency matrix
  • ⁇ is the preset weight value
  • S 2 represents the second similarity parameter.
  • the calculation formula of the adjacency matrix is: Where R is a predetermined connection state matrix of the connection state of each host node of the database and the host node pair, and R T is the transpose matrix of R.
  • the present application provides a method, an electronic device, and a readable storage medium for regulating communication load between network host nodes, which are based on a multi-layer network similarity algorithm and regulate the communication load of a cloud platform host node.
  • all running host nodes on the cloud platform are collectively classified as a host node cluster, and the host node cluster is abstractly divided into a multi-layer network.
  • the multi-layer network includes two levels of networks. The layer network is divided from the physical level, and the host nodes that communicate with each other are divided into the first layer network.
  • the second layer network is divided according to the topological structure in the graph theory. It is a bipartite graph network (or called a hidden network) in nature.
  • the database is equivalent to the virtual host node, and all databases form a virtual host node cluster.
  • the host node clusters in the first layer network communicate with each other, the host node clusters are a subset of the bipartite graph network set, and the virtual host node clusters are the complement of the host node clusters in the bipartite graph network set.
  • the determined host node pair I 1 and I 2 (I 1 is one node in the host node, and I 2 is the other host node in the host node).
  • a node) is connected in common (including direct connection and indirect connection), the more the number of other host nodes, and the closer the distance between the determined host node pair I 1 and I 2 and the other host nodes connected in common.
  • I 1 and I 2 are commonly connected to the database I
  • the greater the number the greater the possibility that the determined host node pair I 1 and I 2 that are not communicatively connected are connected to each other. For example, if the determined host node pair I 1 and I 2 that are not connected in communication are similar host nodes on the same business line, then the determined host node pair I 1 and I 2 that are not connected in communication perform business items or processes Will be very similar, so the same database node (such as U1 or/and U3) will be called, resulting in a greater possibility of connection between these two nodes.
  • the determined similarity (a measure of similarity) between host node pairs I 1 and I 2 that are not communicatively connected can be used to predict the determined failure.
  • No communication connection to the communication master node 1, connection possibility between I 2 the host node determines not connected to the communication I 1, the higher the degree of similarity between the I 2 I, it is determined that the host is connected to node
  • the connection between I 1 and I 2 is more likely, and the similarity between host nodes can be measured by the conditions of other host nodes connected together and the database connected together.
  • the analysis is divided into two parts here.
  • the first part is the determined host node pair I 1 and I 2 that are not connected to the load control object, and the other part is divided by I. 1, I 2 of the other host nodes, including a node connected to the host order (directly with the I 1, I 2 nodes connected to the host), second order node connected to the host (separated by a node and indirectly the I 1, I 2 is connected to the master node) , And even N-level connected host nodes (host nodes connected to I 1 and I 2 indirectly by N-1 nodes).
  • first-order connected host nodes second-order connected host nodes, and even N-order connected host nodes
  • the prediction of the connection possibility between I 1 and I 2 will be more accurate for the determined host node that is not connected.
  • the multi-dimensional information of first-order connected host nodes, second-order connected host nodes, and even N-order connected host nodes are used to calculate and analyze the similarity between host node pairs I 1 and I 2 that are not connected to communication.
  • the formula is as follows :
  • S 1 is the first similarity parameter in the first layer network between the host node pair I 1 and I 2 , and I 1 , I 2 are determined host node pairs that are not communicatively connected under the cloud platform; , X and y are any two host nodes under the cloud platform; ⁇ is a preset and adjustable weight parameter, 0 ⁇ 1; (A n ) xy is the pre-defined host node pair that is not connected Set the number of other host nodes that are in communication with each host node of the host node pair corresponding to a distance of n, and n is a natural number.
  • a xy is a first-order connected host node
  • (A 2 ) xy is a second-order connected host node.
  • (A 3 ) xy is the third-order phase connection point, if the connection link between the host node pair I 1 and I 2 with a distance of 3 includes I 1 -J 1 -J 2 -I 2 , I 1 -K 1 -K 2 -I 2 (J 1 , J 2 , K 1 , K 2 are four host nodes respectively), then the distance 3 corresponds to each host node I 1 and I 2 of the host node pair are all communicatively connected
  • the similarity between the host node I 1 and itself should be 1. But because the host node will not have a network connection with itself, and in order to avoid affecting subsequent calculations, we set the similarity of the host node itself to 0 .
  • the above formula balances the algorithm complexity and accuracy of the first-layer network by adjusting the weight parameter ⁇ and the highest connection order n. For example, in some scenarios, if you need to consider the greater impact of low-level connections, that is, between the determined host node pairs I 1 and I 2 that are not connected to the first-level connected host nodes, the second-level connected host nodes, etc.
  • the similarity degree of the determined host node that is not connected to the communication has a greater influence on the possibility of the final connection between I 1 and I 2
  • the value of the weight parameter ⁇ can be adjusted to weaken the similarity S 1 The influence of the calculation result, thereby reducing the influence of the higher-order phase connection point on the predicted result of the final connection possibility between I 1 and I 2 on the determined host node that is not connected.
  • the adjacency matrix formed by the connection status of the host node and the database can be represented by B.
  • Each row in R represents a host node set, including host nodes I 1 and I 2
  • each column Represents a database set, including databases U 1 , U 2 , U 3 , U 4.
  • a host node When a host node is connected to a database, it is represented by 1, and if it is not connected, it is represented as 0.
  • the adjacency matrix shown in the matrix table 1 of Figure (a) for example, the database U 1 and the host node I 2 are connected as 1, and the database U 4 and the host node I 2 are not connected as 0.
  • the adjacency matrix is spliced according to the following formula including the transposed matrix to obtain the adjacency matrix B shown in matrix table 2 of Figure (b).
  • the first similarity parameter S 1 in the first layer network is multiplied by the second similarity parameter S 2 in the second layer network, Obtain the final comprehensive similarity S of the determined host node pair that is not connected in communication.
  • step S300-step S330 an implementation flowchart of an embodiment of a communication load control method according to this application.
  • the processor 12 executes the computer program of the communication load control program 10 stored in the memory 11
  • the method for realizing communication load control includes: step S300-step S330, combined with the schematic diagram of the application environment of the communication load control method of FIG. 4 Explain the realization of this application.
  • the first calculation module 100 regularly performs (for example, every 10 minutes) communication load adjustment between network hosts. After performing the communication load adjustment between network hosts, it determines the host node pairs that are not communicatively connected between the included host nodes. After determining a host node pair (for example, host node I 1 and host node I 2 ), respectively obtain the number of other host nodes communicatively connected with each host node of the host node pair corresponding to each preset distance (for example, if host node I 1, I 2 a distance between the connecting link comprises 2 I 1 -BI 2, I 1 -CI 2, I 1 -DI 2 that three (B, C, D are three host node), Then the number of other host nodes connected with a distance of 2 between host nodes I 1 and I 2 is 3), and the number corresponding to each preset distance is substituted into the first variable conversion formula to calculate the corresponding first similarity of the host node pair parameter.
  • a host node pair for example, host node I 1 and host node I 2
  • the second computing module 110 obtains the connection status of each predetermined database (for example, databases U 1 , U 2 , U 3 , U 4 ) and each host node of the host node pair (for example, I 1 host node and database U 3 is represented by 1 if there is a connection, and is represented by 0 if there is no connection), the connection status of each predetermined database and each host node of the host node pair is converted into a corresponding adjacency matrix, and the adjacency matrix is substituted into the predetermined adjacency matrix The second variable conversion formula is calculated to obtain the second similarity parameter corresponding to the host node pair.
  • each predetermined database for example, databases U 1 , U 2 , U 3 , U 4
  • each host node of the host node pair for example, I 1 host node and database U 3 is represented by 1 if there is a connection, and is represented by 0 if there is no connection
  • I 1 host node and database U 3 is represented by 1 if there is a connection,
  • the third calculation module 120 inputs the calculated first similarity parameter S 1 and the second similarity parameter S 2 into a predetermined comprehensive similarity calculation formula to calculate the corresponding comprehensive similarity of the host node pair. Degree S.
  • the control module 130 adjusts the communication load of the included host node for the corresponding host node according to the calculated comprehensive similarity.
  • the adjustment module 130 adjusts the communication load of the included host node for the corresponding host node according to the calculated comprehensive similarity S, including: if one is determined If the comprehensive similarity corresponding to the host node pair is greater than the preset threshold, the maximum number of host connections corresponding to the host node pair is determined according to the first mapping relationship between the predetermined comprehensive similarity interval and the maximum number of host connections, and according to the determined maximum The number of host connections controls the number of connections of other host nodes connected to the host node of the host node pair, so that the number of connections of other host nodes connected to the host node of the host node pair is less than or equal to the determined maximum host Number of connections.
  • the step of adjusting the communication load of the included host node for the corresponding host node by the control module 130 according to the calculated comprehensive similarity S includes:
  • each determined host node After each determined host node has calculated the corresponding comprehensive similarity, sort the comprehensive similarities in descending order, and determine the elastically adjustable host node corresponding to the comprehensive similarity in the preset order of sorting. Correct.
  • the number of flexible host connections corresponding to each flexible controllable host node pair is found.
  • the number of connections of other host nodes connected to the host node in each elastically adjustable host node pair is controlled in accordance with the corresponding number of elastic host connections, so that the host nodes of each elastically adjustable host node pair are connected to each other.
  • the number of connections of other host nodes is less than or equal to the number of corresponding elastic host connections.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)
  • Computer And Data Communications (AREA)

Abstract

本申请涉及云监控技术领域，揭露了一种网络主机间通信负载调控方法、电子装置及可读存储介质，通过获得与主机节点对的各个主机节点均通信连接的其它主机节点的数量以形成第一相似度参数，通过获得各个预先确定的数据库与主机节点对的各个主机节点的连接状态以形成第二相似度参数，进而通过第一相似度参数和第二相似度参数计算得到主机节点对对应的综合相似度，根据综合相似度为对应的主机节点对包含的主机节点的通信负载进行调控。利用本申请揭露的技术方案，能更准确并及时有效地调控主机节点通信负载并均衡网络资源配置。

Description

网络主机间通信负载调控方法、电子装置及可读存储介质
本申请要求于2019年04月04日提交中国专利局、申请号为201910269166.6、发明名称为“网络主机间通信负载调控方法、电子装置及可读存储介质”的中国专利申请的优先权，其全部内容通过引用结合在申请中。
技术领域
本申请涉及云监控技术领域，尤其涉及一种网络主机间通信负载调控方法、电子装置及可读存储介质。
背景技术
随着云平台的发展，云平台内的主机数量也在大量增加，各个主机节点间的连接方式也会变得越来越复杂。为了防止云平台内主机节点由于连接过多导致通信负载过大的情况发生，现有技术中普遍采用设定阈值的方式来对云平台内主机节点的通信资源进行调节处理，即统一设定最大连接阈值，当云平台内某个主机节点超过该最大连接阈值时，再对该主机节点进行处理，例如暂停该主机节点的工作。现有的这种通过设置最大连接阈值的方式监控主机节点的负载调整时机，是非常粗放的一种负载监控调整方案，在系统资源浪费和系统负载过大之间无法准确的进行平衡，例如，若将统一设定的最大连接阈值设置的比较低，则会导致主机节点预留出较多的处理空间，又会带来很大的系统资源浪费；若将统一设定的最大连接阈值设置的比较高，则可能会导致主机节点系统过载而导致系统异常或者响应缓慢。虽然，现有技术中也有通过考虑业务关系的方式对主机节点间的连接数进行预测，从而对可能出现的高通信负载情况进行预警，但由于业务变更频繁，且考虑的参考项通常较为单一，往往无法对主机节点间的可连接负载数量做出准确的预测，因此经常造成误判。
发明内容
鉴于以上内容，本申请的主要目的在于提供一种网络主机间通信负载调控方法、装置及存储介质，以更加准确地对云平台通信网络通信负载量进行监控并采取及时有效的调控措施优化网络通信负载资源配置。
为实现上述目的，本申请提供一种网络主机间通信负载调控方法，该方法包括：
第一计算步骤：定时进行网络主机间通信负载调控，在进行网络主机间通信负载调控后，确定包含的主机节点之间未通信连接的主机节点对，在确定一个主机节点对后，分别获得各个预设距离对应的与该主机节点对的各个主机节点均通信连接的其它主机节点的数量，将各个预设距离对应的数量代入第一变量转换公式以计算得到该主机节点对对应的第一相似度参数；
第二计算步骤：分别获得各个预先确定的数据库与该主机节点对的各个主机节点的连接状态，将各个预先确定的数据库与该主机节点对的各个主机节点的连接状态转换成对应的邻接矩阵，将所述邻接矩阵代入预先确定的第二变量转换公式以计算得到该主机节点对对应的第二相似度参数；
第三计算步骤：将计算得到的所述第一相似度参数和第二相似度参数输入到预先确定的综合相似度计算公式，以计算得到该主机节点对对应的综合相似度；
调控步骤：根据计算得到的综合相似度，为对应的主机节点对包含的主机节点的通信负载进行调控。
为实现上述目的，本申请还提供一种电子装置，所述电子装置包括存储器和处理器，所述存储器上存储有可在所述处理器上运行的通信负载调控程序，所述信负载调控程序被所述处理器执行时实现如下步骤：
第一计算步骤：定时进行网络主机间通信负载调控，在进行网络主机间通信负载调控后，确定包含的主机节点之间未通信连接的主机节点对，在确定一个主机节点对后，分别获得各个预设距离对应的与该主机节点对的各个主机节点均通信连接的其它主机节点的数量，将各个预设距离对应的数量代入第一变量转换公式以计算得到该主机节点对对应的第一相似度参数；
第二计算步骤：分别获得各个预先确定的数据库与该主机节点对的各个主机节点的连接状态，将各个预先确定的数据库与该主机节点对的各个主机节点的连接状态转换成对应的邻接矩阵，将所述邻接矩阵代入预先确定的第二变量转换公式以计算得到该主机节点对对应的第二相似度参数；
第三计算步骤：将计算得到的所述第一相似度参数和第二相似度参数输入到预先确定的综合相似度计算公式，以计算得到该主机节点对对应的综合 相似度；
调控步骤：根据计算得到的综合相似度，为对应的主机节点对包含的主机节点的通信负载进行调控。
为实现上述目的，本申请还提供一种计算机可读存储介质，所述计算机可读存储介质中包括网络主机间通信负载调控程序，该网络主机间通信负载调控程序被处理器执行时实现上述网络主机间通信负载调控方法的步骤。
相较现有技术，本申请通过定时进行网络主机间通信负载调控，在确定一个主机节点对后，分别获得各个预设距离对应的与该主机节点对的各个主机节点均通信连接的其它主机节点的数量以形成第一相似度参数，分别获得各个预先确定的数据库与该主机节点对的各个主机节点的连接状态以形成第二相似度参数，将第一相似度参数和第二相似度参数经过相应转换得到综合相似度，根据综合相似度为对应的主机节点对包含的主机节点的通信负载进行调控，以更准确并及时有效地调控主机节点通信负载并均衡网络资源配置。
附图说明
图1为本申请实现网络主机间通信负载调控的电子装置一实施例的硬件结构图；
图2为图1中网络主机间通信负载调控程序一实施例的功能模块图；
图3为本申请网络主机间通信负载调控方法一实施例的实施流程图；
图4为本申请网络主机间通信负载调控方法一实施例的应用环境示意图。
本申请目的实现、功能特点及优点将结合实施例，参照附图做进一步说明。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处所描述的具体实施例仅用以解释本申请，并不用于限定本申请。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其它实施例，都属于本申请保护的范围。
需要说明的是，在本申请中涉及“第一”、“第二”等的描述仅用于描述目的，而不能理解为指示或暗示其相对重要性或者隐含指明所指示的技术特征的数量。由此，限定有“第一”、“第二”的特征可以明示或者隐含地包括至少一个该 特征。另外，各个实施例之间的技术方案可以相互结合，但是必须是以本领域普通技术人员能够实现为基础，当技术方案的结合出现相互矛盾或无法实现时应当认为这种技术方案的结合不存在，也不在本申请要求的保护范围之内。
参照图1所示，为本申请实现网络主机间通信负载调控的电子装置一实施例的硬件结构图。在本实施例中，电子装置1包括存储器11、处理器12、网络接口13，存储器11中存储有可被处理器12执行的网络主机间通信负载调控程序10。
所述电子装置1可以是服务器、智能手机、平板电脑、便携计算机、桌上型计算机等具有存储和运算功能的终端设备。在本申请的一个实施例中，当电子装置1为服务器时，该服务器可以是机架式服务器、刀片式服务器、塔式服务器或机柜式服务器等的一种或几种。
所述存储器11包括至少一种类型的可读存储介质。所述至少一种类型的可读存储介质可为如闪存、硬盘、多媒体卡、卡型存储器等的非易失性存储介质。在一些实施例中，所述可读存储介质可以是所述电子装置1的内部存储单元，例如该电子装置1的硬盘。在另一些实施例中，所述可读存储介质也可以是所述电子装置1的外部存储器11，例如所述电子装置1上配备的插接式硬盘，智能存储卡(Smart Media Card，SMC)，安全数字(Secure Digital，SD)卡，闪存卡(Flash Card)等。
所述处理器12可以是中央处理器(Central Processing Unit，CPU)，微处理器或其它任意适用的数据处理芯片，用于运行存储器11中存储的程序代码或处理数据，例如执行网络主机间通信负载调控程序10等。
网络接口13可以包括标准的有线接口、无线接口(如WI-FI接口)。通常用于在该电子装置1与其它电子设备之间建立共同连接。
图1仅示出了具有组件11-13以及网络主机间通信负载调控程序10的电子装置1，但是应理解的是，电子装置1可以包括更多或者更少的组件。
在本申请的一个实施例中，处理器12执行存储器11中存储的网络主机间通信负载调控程序10时实现如下步骤：
第一计算步骤：定时进行网络主机间通信负载调控，在进行网络主机间通信负载调控后，确定包含的主机节点之间未通信连接的主机节点对，在确 定一个主机节点对后，分别获得各个预设距离对应的与该主机节点对的各个主机节点均通信连接的其它主机节点的数量，将各个预设距离对应的数量代入第一变量转换公式以计算得到该主机节点对对应的第一相似度参数；
第二计算步骤：分别获得各个预先确定的数据库与该主机节点对的各个主机节点的连接状态，将各个预先确定的数据库与该主机节点对的各个主机节点的连接状态转换成对应的邻接矩阵，将所述邻接矩阵代入预先确定的第二变量转换公式以计算得到该主机节点对对应的第二相似度参数；
第三计算步骤：将计算得到的所述第一相似度参数和第二相似度参数输入到预先确定的综合相似度计算公式，以计算得到该主机节点对对应的综合相似度；
调控步骤：根据计算得到的综合相似度，为对应的主机节点对包含的主机节点的通信负载进行调控。
为了更好的阐述说明所述电子装置1中各个组成部件的功能及相互间的功能配合关系，以下结合图2-3进行详细描述。
参照图2所示，为图1中网络主机间通信负载调控程序10一实施例的功能模块图。所述网络主机间通信负载调控程序10被分割为多个功能模块，该多个功能模块被存储于存储器11中，并由处理器12执行，以更准确并及时有效地调控主机节点通信负载并均衡网络资源配置。本申请所称的“模块”是指能够完成特定功能的一系列计算机程序指令集。在本实施例中，所述网络主机间通信负载调控程序10被分割为：第一计算模块100、第二计算模块110、第三计算模块120、调控模块130。应该理解的是：在本实施例中，将所述网络主机间通信负载调控程序10分割成第一计算模块100、第二计算模块110、第三计算模块120、调控模块130，仅仅是为了更清楚的表达出所述通信负载调控程序10所能实现的功能，并不用于限定所述网络主机间通信负载调控程序10仅能或者必须分割成第一计算模块100、第二计算模块110、第三计算模块120以及调控模块130，对本领域的技术人员来说，可以在其它实施例中，轻易将所述网络主机间通信负载调控程序10分割成与本实施例不同的功能模块，在此不做赘述。
所述第一计算模块100，用于定时(例如，每间隔10分钟)进行网络主机间通信负载调控，在进行网络主机间通信负载调控后，确定包含的主机节 点之间未通信连接的主机节点对，在确定一个主机节点对后(例如，主机节点I 1和主机节点I 2)，分别获得各个预设距离对应的与该主机节点对的各个主机节点均通信连接的其它主机节点的数量(例如，若主机节点对I 1、I 2之间距离为2的连接链路包括I 1-B-I 2、I 1-C-I 2、I 1-D-I 2这三条(B，C，D分别为三个主机节点)，则距离2对应的与该主机节点对的各个主机节点I 1和I 2均通信连接的其它主机节点的数量为3；若主机节点对I 1、I 2之间距离为3的连接链路包括I 1-E 1-E 2-I 2、I 1-F 1-F 2-I 2这二条(E 1，E 2，F 1，F 2分别为四个主机节点，则距离3对应的与该主机节点对的各个主机节点I 1和I 2均通信连接的其它主机节点的数量为4))，将各个预设距离对应的数量代入第一变量转换公式以计算得到该主机节点对对应的第一相似度参数S 1。
所述第二计算模块110，用于分别获得各个预先确定的数据库(例如数据库U 1、U 2、U 3、U 4)与该主机节点对I 1、I 2的各个主机节点的连接状态(例如，I 1主机节点和数据库U 3有连接则用1表示，无连接则用0表示)，将各个预先确定的数据库与该主机节点对I 1、I 2的各个主机节点I 1或I 2的连接状态转换成对应的邻接矩阵，将所述邻接矩阵代入预先确定的第二变量转换公式以计算得到该主机节点对对应的第二相似度参数S 2。
所述第三计算模块120，用于将计算得到的所述第一相似度参数S 1和第二相似度参数S 2输入到预先确定的综合相似度计算公式，以计算得到该主机节点对对应的综合相似度S。
所述调控模块130，用于根据计算得到的综合相似度S，为对应的主机节点对包含的主机节点的通信负载进行调控。
可选地，在本申请的一个实施例中，所述调控模块130根据计算得到的综合相似度S，为对应的主机节点对包含的主机节点的通信负载进行调控的步骤包括：若一个确定的主机节点对I 1、I 2对应的综合相似度大于预设阈值(例如，预设阈值为0.15)，则根据预先确定的综合相似度区间与最大主机连接数的第一映射关系，确定该主机节点对对应的最大主机连接数(例如，最大主机连接数为100)，并根据确定的最大主机连接数对与该主机节点对中的主机节点连接的其它主机节点的连接数量进行控制，以使与该主机节点对中的主机节点连接的其它主机节点的连接数量小于或者等于确定的最大主机连接数。
可选地，在本申请的一个实施例中，所述调控模块130根据计算得到的 综合相似度S，为对应的主机节点对包含的主机节点的通信负载进行调控的步骤包括：
在各个确定的主机节点对对应的综合相似度计算完毕后，对各个综合相似度按照从大到小的顺序排序，确定排序在前的预设次序(例如，排序在前的10个)的综合相似度对应的可弹性调控的主机节点对(例如，可弹性调控的主机节点对的编号分别为I 1和I 2，I 5和I 6，I 7和I 8，I 12和I 13，I 15和I 16，I 17和I 18，I 19和I 20，I 22和I 23，I 25和I 26，I 37和I 38)；
根据预先确定的综合相似度区间与弹性主机连接数的第二映射关系，找出各个可弹性调控的主机节点对对应的弹性主机连接数(例如，I 1和I 2对应的弹性主机连接数为30，I 5和I 6对应的弹性主机连接数为35，I 7和I 8对应的弹性主机连接数为32，I 12和I 13对应的弹性主机连接数为33，I 15和I 16对应的弹性主机连接数为38，I 17和I 18对应的弹性主机连接数为37，I 19和I 20对应的弹性主机连接数为39，I 22和I 23对应的弹性主机连接数为36，I 25和I 26对应的弹性主机连接数为34，I 37和I 38对应的弹性主机连接数为31)；
分别对各个可弹性调控的主机节点对中的主机节点连接的其它主机节点的连接数量按照对应的弹性主机连接数进行控制，以分别使与各个可弹性调控的主机节点对中的主机节点连接的其它主机节点的连接数量(例如，I 1和I 2对应的弹性主机连接数为30，则I 1连接的其它主机节点的连接数量为25)小于或者等于对应的弹性主机连接数。
可选地，所述第一变量转换公式为:S 1＝αA xy+α 2(A 2) xy+α 3(A 3) xy+...+α n(A n) xy，其中，x、y为云平台下的任意两个主机节点；α指的是预设的权重参数，0<α<1；(A n) xy指的是预设距离n对应的与主机节点对的各个主机节点均通信连接的其它主机节点的数量，n为正整数，S 1代表第一相似度参数。
可选地，所述第二变量转换公式为：
Figure PCTCN2019102185-appb-000001
其中，B为邻接矩阵，i代表所述邻接矩阵的第i行，k代表所述邻接矩阵的第k列，μ为预设权重值，S 2代表第二相似度参数。
可选地，所述邻接矩阵的计算公式为：
Figure PCTCN2019102185-appb-000002
其中R为预先确定的数据库与主机节点对的各个主机节点的连接状态的连接状态矩阵，R T为R的转置矩阵。
可选地，所述综合相似度计算公式为：S＝S 1×S 2，其中S 1为第一相似度参 数，S 2为第二相似度参数。
本申请提供一种网络主机节点间通信负载调控方法、电子装置及可读存储介质，是基于多层网络相似度算法，并对云平台主机节点的通信负载进行调控。在本申请中，将云平台上的所有运行主机节点都统归为主机节点集群，并对该主机节点集群抽象划分为多层网络，该多层网络包括了两个层次的网络，其中第一层网络从物理层次来划分，将相互之间有进行通信连接的主机节点划分为第一层网络。第二层网络是按图论里的拓扑结构来划分的，性质上是个二部图网络(或称为隐藏网络)，其中数据库等效于虚拟主机节点，所有数据库形成虚拟主机节点集群，并与第一层网络中的主机节点集群进行通信连接，主机节点集群为所述二部图网络集合里的子集，虚拟主机节点集群是所述二部图网络集合里的主机节点集群的补集。
通过对主机节点集群内的连接情况进行分析得出结论：确定的未通信连接的主机节点对I 1、I 2(I 1是该主机节点中的一个节点，I 2是该主机节点中的另一个节点)共同相连(包括直接相连和间接相连)的其它主机节点数量越多，且确定的未通信连接的主机节点对I 1、I 2和其共同相连的其它主机节点之间的距离越近(相隔节点越少)，则该确定的未通信连接的主机节点对I 1、I 2互相连接的可能性越大；而确定的未通信连接的主机节点对I 1、I 2共同相连的数据库越多，则该确定的未通信连接的主机节点对I 1、I 2互相连接的可能性越大。例如，若确定的未通信连接的主机节点对I 1、I 2为同一业务线上的相似主机节点，则这确定的未通信连接的主机节点对I 1、I 2执行的业务项或处理事情会很相似，因此会调用相同的数据库节点(例如U1或/与U3)，导致这两个节点之间的连接可能性也较大。
基于上述分析，本申请中针对云平台下数量众多的主机节点，可通过确定的未通信连接的主机节点对I 1、I 2之间的相似度(对相似性的度量)来预测确定的未通信连接的主机节点对I 1、I 2之间的连接可能性，确定的未通信连接的主机节点对I 1、I 2之间的相似度越高，则确定的未通信连接的主机节点对I 1、I 2之间的连接可能性越大,而主机节点间的相似度可以通过共同相连的其它主机节点以及共同相连数据库的情况来衡量。
进一步地，针对所述第一层网络的所有主机节点，这里分两部分来分析,第一部分是作为负载调控对象的确定的未通信连接的主机节点对I 1、I 2，另一 部分是除I 1、I 2的其它主机节点，包括一阶相连主机节点(直接与I 1、I 2相连的主机节点)，二阶相连主机节点(相隔一个节点间接与I 1、I 2相连的主机节点)，乃至N阶相连主机节点(相隔N-1个节点间接与I 1、I 2相连的主机节点)。
如果应用一阶相连主机节点，二阶相连主机节点，乃至N阶相连主机节点这些所包含的多维度信息来计算与分析确定的未通信连接的主机节点对I 1、I 2之间的相似度，则针对确定的未通信连接的主机节点对I 1、I 2之间产生连接可能性大小的预测会更加准确。其中，应用一阶相连主机节点，二阶相连主机节点，乃至N阶相连主机节点这些多维度信息来计算分析确定的未通信连接的主机节点对I 1、I 2之间的相似度的公式如下：
S 1＝αA xy+α 2(A 2) xy+a 3(A 3) xy+...+α n(A n) xy
其中，S 1为主机节点对I 1、I 2之间在所述第一层网络中的第一相似度参数，I 1、I 2为云平台下的确定的未通信连接的主机节点对；，x、y为云平台下的任意两个主机节点；α是预设可调的权重参数，0<α<1；(A n) xy是在确定一个未通信连接的主机节点对后，预设距离为n的对应的与该主机节点对的各个主机节点均通信连接的其它主机节点的数量，n为自然数。
即A xy是一阶相连主机节点，(A 2) xy是二阶相连主机节点。例如，对于二阶相连主机节点，若主机节点I 1、I 2之间距离为2的连接链路包括I 1-X-I 2、I 1-Y-I 2、I 1-Z-I 2这三条(X，Y，Z分别为三个主机节点)，则距离2对应的与该主机节点对的各个主机节点I 1和I 2均通信连接的其它主机节点的数量为3，即(A 2) xy＝3。
又例如，(A 3) xy是三阶相连接点，若主机节点对I 1、I 2之间距离为3的连接链路包括I 1-J 1-J 2-I 2、I 1-K 1-K 2-I 2这二条(J 1，J 2，K 1，K 2分别为四个主机节点)，则距离3对应的与该主机节点对的各个主机节点I 1和I 2均通信连接的其它主机节点的数量为4，即(A 3) xy＝4。特别需指出，主机节点I 1和它自己的相似度应该为1.但因为主机节点不会和自己产生网络连接,而为避免对后续计算产生影响,我们将主机节点本身的相似度设置为0。
需要说明的是，上述公式是通过对权重参数α以及最高连接阶数n的调整来平衡第一层网络的算法复杂度与准确度的。例如，在一些场景中，若需考虑低阶连接的影响较大，即确定的未通信连接的主机节点对I 1、I 2之间在一阶相连主机节点、二阶相连主机节点等之间的相似度对这确定的未通信连接 的主机节点对I 1、I 2之间最终连接的可能性影响较大，则可调低权重参数α的值，以削弱高阶相连接点对相似度S 1计算结果的影响，进而减小高阶相连接点对确定的未通信连接的主机节点对I 1、I 2之间最终连接可能性的预测结果的影响。
相反，若需考虑更多维度的信息来计算相似度，则可调高权重参数α的值，以加强高阶相连接点对第一相似度参数S 1计算结果的影响，进而增强高阶相连接点对确定的未通信连接的主机节点对I 1、I 2之间最终连接可能性的预测结果的影响。
针对所述第二层网络，主机节点与数据库的连接状态所形成的邻接矩阵可以用B来表示，所述R中的每一行代表一个主机节点集，包括主机节点I 1和I 2，每一列代表一个数据库集，包括数据库U 1、U 2、U 3、U 4，当一个主机节点和一个数据库存在连接时用1表示,未连接则表示为0。如图(a)的矩阵表1所示的邻接矩阵，例如，数据库U 1和主机节点I 2连接表示为1，数据库U 4和主机节点I 2未连接则表示为0。然后根据如下包含转置矩阵的公式拼接所述邻接矩阵，得到图(b)的矩阵表2所示的邻接矩阵B。
Figure PCTCN2019102185-appb-000003
Figure PCTCN2019102185-appb-000004
接着对照所述邻接矩阵B使用下面的第二变量转换公式计算出确定的未通信连接的主机节点对I 1、I 2在第二层网络的第二相似度参数，其中μ为预设权重值，S 2代表第二相似度参数，并得到图(c)所示的矩阵表3。
Figure PCTCN2019102185-appb-000005
Figure PCTCN2019102185-appb-000006
对于确定的未通信连接的主机节点对I 1、I 2，所述第一层网络中的第一相似度参数S 1和所述第二层网络中的第二相似度参数S 2相乘，得到该确定的未通信连接的主机节点对最终的综合相似度S。
参照图3所示，为本申请通信负载调控方法一实施例的实施流程图。在本实施例中，处理器12执行存储器11中存储的通信负载调控程序10的计算机程序时实现通信负载调控方法包括：步骤S300-步骤S330，并结合图4的通信负载调控方法的应用环境示意图对本申请的实现加以阐述。
S300,第一计算模块100定时进行(例如，每间隔10分钟)网络主机间通信负载调控，在进行网络主机间通信负载调控后，确定包含的主机节点之间未通信连接的主机节点对，在确定一个主机节点对(例如，主机节点I 1和主机节点I 2)后，分别获得各个预设距离对应的与该主机节点对的各个主机节点均通信连接的其它主机节点的数量(例如，若主机节点I 1、I 2之间距离为2的连接链路包括I 1-B-I 2、I 1-C-I 2、I 1-D-I 2这三条(B，C，D分别为三个主机节点)，则主机节点I 1、I 2之间距离为2的连接其它主机节点数为3)，将各个预设距离对应的数量代入第一变量转换公式以计算得到该主机节点对对应的第一相似度参数。
S310,第二计算模块110分别获得各个预先确定的数据库(例如数据库U 1、U 2、U 3、U 4)与该主机节点对的各个主机节点的连接状态(例如，I 1主机节点和数据库U 3有连接则用1表示，无连接则用0表示)，将各个预先确定的数据库与该主机节点对的各个主机节点的连接状态转换成对应的邻接矩阵，将所述邻接矩阵代入预先确定的第二变量转换公式以计算得到该主机节点对对应的第二相似度参数。
S320，第三计算模块120将计算得到的所述第一相似度参数S 1和第二相似度参数S 2输入到预先确定的综合相似度计算公式，以计算得到该主机节点对对应的综合相似度S。
S330，调控模块130根据计算得到的综合相似度，为对应的主机节点对包含的主机节点的通信负载进行调控。
可选地，在本申请的一个实施例中，所述调控模块130根据计算得到的综合相似度S，为对应的主机节点对包含的主机节点的通信负载进行调控的步骤 包括：若一个确定的主机节点对对应的综合相似度大于预设阈值，则根据预先确定的综合相似度区间与最大主机连接数的第一映射关系，确定该主机节点对对应的最大主机连接数，并根据确定的最大主机连接数对与该主机节点对中的主机节点连接的其它主机节点的连接数量进行控制，以使与该主机节点对中的主机节点连接的其它主机节点的连接数量小于或者等于确定的最大主机连接数。
可选地，在本申请的一个实施例中，所述调控模块130根据计算得到的综合相似度S，为对应的主机节点对包含的主机节点的通信负载进行调控的步骤包括：
在各个确定的主机节点对对应的综合相似度计算完毕后，对各个综合相似度按照从大到小的顺序排序，确定排序在前的预设次序的综合相似度对应的可弹性调控的主机节点对。
根据预先确定的综合相似度区间与弹性主机连接数的第二映射关系，找出各个可弹性调控的主机节点对对应的弹性主机连接数。
分别对各个可弹性调控的主机节点对中的主机节点连接的其它主机节点的连接数量按照对应的弹性主机连接数进行控制，以分别使与各个可弹性调控的主机节点对中的主机节点连接的其它主机节点的连接数量小于或者等于对应的弹性主机连接数。
通过以上的实施方式的描述，本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现，当然也可以通过硬件，但很多状态下前者是更佳的实施方式。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品存储在如上所述的一个存储介质(如ROM/RAM、磁碟、光盘)中，包括若干指令用以使得一台终端设备(可以是手机，计算机，服务器，或者网络设备等)执行本申请各个实施例所述的方法。
以上仅为本申请的优选实施例，并非因此限制本申请的专利范围，凡是利用本申请说明书及附图内容所作的等效结构或等效流程变换，或直接或间接运用在其它相关的技术领域，均同理包括在本申请的专利保护范围内。

Claims (20)

  1. 一种网络主机间通信负载调控方法，应用于电子装置，其特征在于，所述方法包括：
    第一计算步骤：定时进行网络主机间通信负载调控，在进行网络主机间通信负载调控后，确定包含的主机节点之间未通信连接的主机节点对，在确定一个主机节点对后，分别获得各个预设距离对应的与该主机节点对的各个主机节点均通信连接的其它主机节点的数量，将各个预设距离对应的数量代入第一变量转换公式以计算得到该主机节点对对应的第一相似度参数；
    第二计算步骤：分别获得各个预先确定的数据库与该主机节点对的各个主机节点的连接状态，将各个预先确定的数据库与该主机节点对的各个主机节点的连接状态转换成对应的邻接矩阵，将所述邻接矩阵代入预先确定的第二变量转换公式以计算得到该主机节点对对应的第二相似度参数；
    第三计算步骤：将计算得到的所述第一相似度参数和第二相似度参数输入到预先确定的综合相似度计算公式，以计算得到该主机节点对对应的综合相似度；
    调控步骤：根据计算得到的综合相似度，为对应的主机节点对包含的主机节点的通信负载进行调控。
  2. 根据权利要求1所述的网络主机间通信负载调控方法，其特征在于，所述调控步骤包括：
    若一个确定的主机节点对对应的综合相似度大于预设阈值，则根据预先确定的综合相似度区间与最大主机连接数的第一映射关系，确定该主机节点对对应的最大主机连接数，并根据确定的最大主机连接数对与该主机节点对中的主机节点连接的其它主机节点的连接数量进行控制，以使与该主机节点对中的主机节点连接的其它主机节点的连接数量小于或者等于确定的最大主机连接数。
  3. 根据权利要求1所述的网络主机间通信负载调控方法，其特征在于，所述调控步骤包括：
    在各个确定的主机节点对对应的综合相似度计算完毕后，对各个综合相似度按照从大到小的顺序排序，确定排序在前的预设次序的综合相似度对应 的可弹性调控的主机节点对；
    根据预先确定的综合相似度区间与弹性主机连接数的第二映射关系，找出各个可弹性调控的主机节点对对应的弹性主机连接数；
    分别对各个可弹性调控的主机节点对中的主机节点连接的其它主机节点的连接数量按照对应的弹性主机连接数进行控制，以分别使与各个可弹性调控的主机节点对中的主机节点连接的其它主机节点的连接数量小于或者等于对应的弹性主机连接数。
  4. 根据权利要求1所述的网络主机间通信负载调控方法，其特征在于，所述第一变量转换公式为:S 1＝αA xy+α 2(A 2) xy+α 3(A 3) xy+...+α n(A n) xy，其中，x、y为云平台下的任意两个主机节点；α指的是预设的权重参数，0<α<1；(A n) xy指的是预设距离n对应的与主机节点对的各个主机节点均通信连接的其它主机节点的数量，n为正整数，S 1代表第一相似度参数。
  5. 根据权利要求1所述的网络主机间通信负载调控方法，其特征在于，所述第二变量转换公式为：
    Figure PCTCN2019102185-appb-100001
    其中，B为邻接矩阵，i代表所述邻接矩阵的第i行，k代表所述邻接矩阵的第k列，μ为预设权重值，S 2代表第二相似度参数。
  6. 根据权利要求5所述的网络主机间通信负载调控方法，其特征在于，所述邻接矩阵的计算公式为：
    Figure PCTCN2019102185-appb-100002
    其中R为预先确定的数据库与主机节点对的各个主机节点的连接状态的连接状态矩阵，R T为R的转置矩阵。
  7. 根据权利要求1-6任一项所述的网络主机间通信负载调控方法，其特征在于，所述综合相似度计算公式为：S＝S 1×S 2，其中S 1为第一相似度参数，S 2为第二相似度参数。
  8. 一种电子装置，所述电子装置包括存储器和处理器，其特征在于，所述存储器上存储有可在所述处理器上运行的网络主机间通信负载调控程序，所述网络主机间通信负载调控程序被所述处理器执行时实现如下步骤：
    第一计算步骤：定时进行网络主机间通信负载调控，在进行网络主机间通信负载调控后，确定包含的主机节点之间未通信连接的主机节点对，在确定一个主机节点对后，分别获得各个预设距离对应的与该主机节点对的各个主机节点均通信连接的其它主机节点的数量，将各个预设距离对应的数量代入第一变量转换公式以计算得到该主机节点对对应的第一相似度参数；
    第二计算步骤：分别获得各个预先确定的数据库与该主机节点对的各个主机节点的连接状态，将各个预先确定的数据库与该主机节点对的各个主机节点的连接状态转换成对应的邻接矩阵，将所述邻接矩阵代入预先确定的第二变量转换公式以计算得到该主机节点对对应的第二相似度参数；
    第三计算步骤：将计算得到的所述第一相似度参数和第二相似度参数输入到预先确定的综合相似度计算公式，以计算得到该主机节点对对应的综合相似度；
    调控步骤：根据计算得到的综合相似度，为对应的主机节点对包含的主机节点的通信负载进行调控。
  9. 如权利要求8所述的电子装置，其特征在于，所述调控步骤包括：
    若一个确定的主机节点对对应的综合相似度大于预设阈值，则根据预先确定的综合相似度区间与最大主机连接数的第一映射关系，确定该主机节点对对应的最大主机连接数，并根据确定的最大主机连接数对与该主机节点对中的主机节点连接的其它主机节点的连接数量进行控制，以使与该主机节点对中的主机节点连接的其它主机节点的连接数量小于或者等于确定的最大主机连接数。
  10. 根据权利要求8所述的电子装置，其特征在于，所述调控步骤包括：
    在各个确定的主机节点对对应的综合相似度计算完毕后，对各个综合相似度按照从大到小的顺序排序，确定排序在前的预设次序的综合相似度对应的可弹性调控的主机节点对；
    根据预先确定的综合相似度区间与弹性主机连接数的第二映射关系，找出各个可弹性调控的主机节点对对应的弹性主机连接数；
    分别对各个可弹性调控的主机节点对中的主机节点连接的其它主机节点的连接数量按照对应的弹性主机连接数进行控制，以分别使与各个可弹性调控的主机节点对中的主机节点连接的其它主机节点的连接数量小于或者等于对应的弹性主机连接数。
  11. 根据权利要求8所述的电子装置，其特征在于，所述第一变量转换公式为:S 1＝αA xy+α 2(A 2) xy+α 3(A 3) xy+...+α n(A n) xy，其中，x、y为云平台下的任意两个主机节点；α指的是预设的权重参数，0<α<1；(A n) xy指的是预设距离n对应的与主机节点对的各个主机节点均通信连接的其它主机节点的数量，n为 正整数，S 1代表第一相似度参数。
  12. 根据权利要求8所述的电子装置，其特征在于，所述第二变量转换公式为：
    Figure PCTCN2019102185-appb-100003
    其中，B为邻接矩阵，i代表所述邻接矩阵的第i行，k代表所述邻接矩阵的第k列，μ为预设权重值，S 2代表第二相似度参数。
  13. 根据权利要求12所述的电子装置，其特征在于，所述邻接矩阵的计算公式为：
    Figure PCTCN2019102185-appb-100004
    其中R为预先确定的数据库与主机节点对的各个主机节点的连接状态的连接状态矩阵，R T为R的转置矩阵。
  14. 根据权利要求9-13任一项所述的电子装置，其特征在于，所述综合相似度计算公式为：S＝S 1×S 2，其中S 1为第一相似度参数，S 2为第二相似度参数。
  15. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质内存储有网络主机间通信负载调控程序，所述网络主机间通信负载调控程序被处理器所执行时实现如权利要求1至6中任一项所述的网络主机间通信负载调控方法的步骤：
    第一计算步骤：定时进行网络主机间通信负载调控，在进行网络主机间通信负载调控后，确定包含的主机节点之间未通信连接的主机节点对，在确定一个主机节点对后，分别获得各个预设距离对应的与该主机节点对的各个主机节点均通信连接的其它主机节点的数量，将各个预设距离对应的数量代入第一变量转换公式以计算得到该主机节点对对应的第一相似度参数；
    第二计算步骤：分别获得各个预先确定的数据库与该主机节点对的各个主机节点的连接状态，将各个预先确定的数据库与该主机节点对的各个主机节点的连接状态转换成对应的邻接矩阵，将所述邻接矩阵代入预先确定的第二变量转换公式以计算得到该主机节点对对应的第二相似度参数；
    第三计算步骤：将计算得到的所述第一相似度参数和第二相似度参数输入到预先确定的综合相似度计算公式，以计算得到该主机节点对对应的综合相似度；
    调控步骤：根据计算得到的综合相似度，为对应的主机节点对包含的主机节点的通信负载进行调控。
  16. 根据权利要求15所述的计算机可读存储介质，其特征在于，所述调控步骤包括：
    若一个确定的主机节点对对应的综合相似度大于预设阈值，则根据预先确定的综合相似度区间与最大主机连接数的第一映射关系，确定该主机节点对对应的最大主机连接数，并根据确定的最大主机连接数对与该主机节点对中的主机节点连接的其它主机节点的连接数量进行控制，以使与该主机节点对中的主机节点连接的其它主机节点的连接数量小于或者等于确定的最大主机连接数。
  17. 根据权利要求15所述的计算机可读存储介质，其特征在于，所述调控步骤包括：
    在各个确定的主机节点对对应的综合相似度计算完毕后，对各个综合相似度按照从大到小的顺序排序，确定排序在前的预设次序的综合相似度对应的可弹性调控的主机节点对；
    根据预先确定的综合相似度区间与弹性主机连接数的第二映射关系，找出各个可弹性调控的主机节点对对应的弹性主机连接数；
    分别对各个可弹性调控的主机节点对中的主机节点连接的其它主机节点的连接数量按照对应的弹性主机连接数进行控制，以分别使与各个可弹性调控的主机节点对中的主机节点连接的其它主机节点的连接数量小于或者等于对应的弹性主机连接数。
  18. 根据权利要求15所述的计算机可读存储介质，其特征在于，所述第一变量转换公式为:S 1＝αA xy+α 2(A 2) xy+α 3(A 3) xy+...+α n(A n) xy，其中，x、y为云平台下的任意两个主机节点；α指的是预设的权重参数，0<α<1；(A n) xy指的是预设距离n对应的与主机节点对的各个主机节点均通信连接的其它主机节点的数量，n为正整数，S 1代表第一相似度参数。
  19. 根据权利要求15所述的计算机可读存储介质，其特征在于，所述第二变量转换公式为：
    Figure PCTCN2019102185-appb-100005
    其中，B为邻接矩阵，i代表所述邻接矩阵的第i行，k代表所述邻接矩阵的第k列，μ为预设权重值，S 2代表第二相似度参数。
  20. 根据权利要求19所述的计算机可读存储介质，其特征在于，所述邻接矩阵的计算公式为：
    Figure PCTCN2019102185-appb-100006
    其中R为预先确定的数据库与主机节点对的各个主机节点的连接状态的连接状态矩阵，R T为R的转置矩阵。
PCT/CN2019/102185 2019-04-04 2019-08-23 网络主机间通信负载调控方法、电子装置及可读存储介质 WO2020199467A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910269166.6A CN110290165B (zh) 2019-04-04 2019-04-04 网络主机间通信负载调控方法、电子装置及可读存储介质
CN201910269166.6 2019-04-04

Publications (1)

Publication Number Publication Date
WO2020199467A1 true WO2020199467A1 (zh) 2020-10-08

Family

ID=68001309

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/102185 WO2020199467A1 (zh) 2019-04-04 2019-08-23 网络主机间通信负载调控方法、电子装置及可读存储介质

Country Status (2)

Country Link
CN (1) CN110290165B (zh)
WO (1) WO2020199467A1 (zh)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120166611A1 (en) * 2010-12-24 2012-06-28 Kim Mi-Jeom Distributed storage system including a plurality of proxy servers and method for managing objects
CN103595780A (zh) * 2013-11-08 2014-02-19 中国人民解放军理工大学 基于消重的云计算资源调度方法
CN106020724A (zh) * 2016-05-20 2016-10-12 南京邮电大学 一种基于数据映射算法的近邻存储方法
CN108600348A (zh) * 2018-04-11 2018-09-28 青岛通产伟博大数据运营有限公司 一种新的去中心化教育资源存储系统和方法
CN109194703A (zh) * 2018-06-29 2019-01-11 平安科技（深圳）有限公司 云平台主机间通信负载的处理方法、电子装置及介质
CN109327540A (zh) * 2018-11-16 2019-02-12 平安科技（深圳）有限公司 电子装置、服务器负载均衡方法及存储介质

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7460558B2 (en) * 2004-12-16 2008-12-02 International Business Machines Corporation System and method for connection capacity reassignment in a multi-tier data processing system network
US20140074553A1 (en) * 2012-09-13 2014-03-13 Truecar, Inc. System and method for constructing spatially constrained industry-specific market areas
CN104077279B (zh) * 2013-03-25 2019-02-05 中兴通讯股份有限公司 一种并行社区发现方法和装置
US9819565B2 (en) * 2015-01-26 2017-11-14 Ciena Corporation Dynamic policy engine for multi-layer network management
CN105939387A (zh) * 2016-06-23 2016-09-14 中国南方电网有限责任公司 基于网省地统一调度的多级负载均衡及服务器角色切换方法
CN106817251B (zh) * 2016-12-23 2020-05-19 烟台中科网络技术研究所 一种基于节点相似度的链路预测方法及装置
CN107124306B (zh) * 2017-04-24 2019-11-05 电子科技大学 网络功能虚拟化环境下的内容交付网络服务器优化部署方法
CN107222410B (zh) * 2017-06-29 2020-09-15 广州杰赛科技股份有限公司 链接预测的方法、装置、终端及计算机可读存储介质

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120166611A1 (en) * 2010-12-24 2012-06-28 Kim Mi-Jeom Distributed storage system including a plurality of proxy servers and method for managing objects
CN103595780A (zh) * 2013-11-08 2014-02-19 中国人民解放军理工大学 基于消重的云计算资源调度方法
CN106020724A (zh) * 2016-05-20 2016-10-12 南京邮电大学 一种基于数据映射算法的近邻存储方法
CN108600348A (zh) * 2018-04-11 2018-09-28 青岛通产伟博大数据运营有限公司 一种新的去中心化教育资源存储系统和方法
CN109194703A (zh) * 2018-06-29 2019-01-11 平安科技（深圳）有限公司 云平台主机间通信负载的处理方法、电子装置及介质
CN109327540A (zh) * 2018-11-16 2019-02-12 平安科技（深圳）有限公司 电子装置、服务器负载均衡方法及存储介质

Also Published As

Publication number Publication date
CN110290165B (zh) 2022-01-28
CN110290165A (zh) 2019-09-27

Similar Documents

Publication Publication Date Title
WO2018076791A1 (zh) 一种资源负载均衡控制方法及集群调度器
US10051056B2 (en) Resource planning method, system, and apparatus for cluster computing architecture
US9641431B1 (en) System and methods for utilization-based balancing of traffic to an information retrieval system
WO2015196686A1 (zh) 一种数据存储方法以及一种数据存储管理服务器
CN113485792B (zh) 一种kubernetes集群内Pod调度方法、终端设备及存储介质
CN106407052A (zh) 一种检测磁盘的方法及装置
CN114513470B (zh) 网络流量控制方法、装置、设备及计算机可读存储介质
CN114143326A (zh) 负载调节方法、管理节点以及存储介质
CN112002949A (zh) 一种动力电池主动均衡的监测管理方法及系统
CN104243531A (zh) 一种数据处理的方法、装置及系统
US10713096B2 (en) System and method for handling data skew at run time
CN112637286B (zh) 网络架构调整方法、装置、系统和计算机可读存储介质
TW202001611A (zh) 多元流動網路之可靠度計算方法及其系統
WO2020199467A1 (zh) 网络主机间通信负载调控方法、电子装置及可读存储介质
CN107357714B (zh) 一种基于监控平台的故障分析方法和装置
WO2020000724A1 (zh) 云平台主机间通信负载的处理方法、电子装置及介质
JP6581282B1 (ja) 人工知能を利用した倒産確率算出システム
CN113254190B (zh) 基于负载能力的动态流调度方法、系统和计算机存储介质
CN105204945A (zh) 一种在大数据背景下的负载平衡装置
US10089149B2 (en) Method for scheduling multiple periodic requests and scheduling device
CN105450751B (zh) 一种提高云计算环境稳定性的系统
CN110851249A (zh) 一种数据导出的方法及设备
US20230179486A1 (en) Method and apparatus related to network analysis
CN115268800B (zh) 基于计算路由重定向的数据处理方法及数据存储系统
JP6756680B2 (ja) 情報処理装置、情報処理方法、及び情報処理プログラム

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19922934

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19922934

Country of ref document: EP

Kind code of ref document: A1