CN113946436B - Resource pre-scheduling method based on load balancing - Google Patents

Resource pre-scheduling method based on load balancing Download PDF

Info

Publication number
CN113946436B
CN113946436B CN202110863510.1A CN202110863510A CN113946436B CN 113946436 B CN113946436 B CN 113946436B CN 202110863510 A CN202110863510 A CN 202110863510A CN 113946436 B CN113946436 B CN 113946436B
Authority
CN
China
Prior art keywords
server
user
resource
time
user demand
Prior art date
Legal status (The legal status 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 status listed.)
Active
Application number
CN202110863510.1A
Other languages
Chinese (zh)
Other versions
CN113946436A (en
Inventor
高岭
朱海蓉
郭子正
向东
李妍
许佶鹏
杨旭东
郭红波
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
NORTHWEST UNIVERSITY
Original Assignee
NORTHWEST UNIVERSITY
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 NORTHWEST UNIVERSITY filed Critical NORTHWEST UNIVERSITY
Priority to CN202110863510.1A priority Critical patent/CN113946436B/en
Publication of CN113946436A publication Critical patent/CN113946436A/en
Application granted granted Critical
Publication of CN113946436B publication Critical patent/CN113946436B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5083—Techniques for rebalancing the load in a distributed system
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/48—Program initiating; Program switching, e.g. by interrupt
    • G06F9/4806—Task transfer initiation or dispatching
    • G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
    • G06F9/4881—Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
    • G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
    • G06F9/5038—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering the execution order of a plurality of tasks, e.g. taking priority or time dependency constraints into consideration
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
    • G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
    • G06F9/505—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering the load

Landscapes

  • Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

According to a result of user demand prediction, determining a resource server set matched with the user demand; selecting a resource pre-scheduling server, collecting resource use data of each server according to the running period of the resource pre-scheduling server, and obtaining the current actual utilization rate of the server according to the executing user request on each server; judging whether the server is in a saturated state, if so, selecting the server with the shortest waiting time as a resource pre-dispatching server, otherwise, distributing the user demand prediction result to a server set to obtain the load variance of the server, and selecting the server with the smallest variance as the server for processing the user demand. And finally, selecting a path with the smallest maximum link bandwidth utilization rate as a resource pre-scheduling link. The method selects a resource pre-scheduling server by calculating a server load variance, and selects an optimal resource pre-scheduling link by using a principle of minimizing maximum bandwidth utilization (max-min). The efficiency of resource scheduling is greatly improved, and the problem that resources cannot be fully utilized is solved.

Description

Resource pre-scheduling method based on load balancing
Technical Field
The invention relates to the field of cloud resource scheduling, in particular to a resource pre-scheduling method based on load balancing.
Background
The cloud platform is constructed in different application forms, and digital resources are stored in the service platform. The resource user can access the cloud platform without time and place limitation, and the cloud service platform is efficiently applied to acquire related resources.
In recent years, public digital culture and civil engineering such as national cultural information resource sharing engineering, digital library popularization engineering, public electronic reading room construction plan and the like are implemented, a national service network is basically established, a large-scale cloud digital resource library group and a technical support service platform represented by national public culture cloud are formed, and a foundation is laid for digitization and networking of a public culture service system. However, there are also a few problems with national public culture cloud services, such as: the most reasonable resource allocation method and resource scheduling strategy cannot be found according to the user demands, public culture cloud resources cannot be efficiently scheduled, and the resource utilization rate is low.
Different from the traditional resource scheduling, the development of public culture cloud environment enables more resources to be stored in the cloud, and the resources in the cloud are more required to be reasonably distributed and used because the cloud resource distribution efficiency influences the experience of using cloud computing services by users. In the process of cloud resource scheduling, because the cloud computing environment has high dynamic property and isomerism and different clients have different demands on cloud computing resources, reasonable scheduling is needed to be carried out on the resources from global optimal and overall server resource load balancing, and timely decision of scheduling is considered according to real-time demand prediction of users. In addition, in the process of executing scheduling, static resource allocation and scheduling often cause insufficient or wasted resources, and manual dynamic resource adjustment has obvious hysteresis, so that the execution state of tasks needs to be monitored in real time, and the server saturation is intelligently judged through the resource utilization rate. And finally, selecting a resource pre-scheduling server through server load variance in the resource scheduling process, and selecting an optimal resource pre-scheduling link by using a principle of minimizing the maximum bandwidth utilization rate. The efficiency of the whole system resource scheduling is greatly improved, and the problem that resources cannot be fully utilized is solved.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention aims to provide a resource pre-scheduling method based on load balancing, which is used for determining a resource pre-scheduling server set according to a prediction result of resources required by a user for a time interval in the future and required resource distribution data, combining server load variance, determining a scheduling server based on server saturation and server load variance, and finally determining an optimal pre-scheduling link according to judging the link conditions of a public culture cloud server and a user side server. The method solves the defects of low resource utilization rate, slow scheduling time, unbalanced server resource load and the like of the existing resource scheduling method.
In order to achieve the above purpose, the technical scheme adopted by the invention is as follows:
a resource pre-scheduling method based on load balancing comprises the following steps:
step 1, server set selection:
determining a resource server set matched with the user demand according to the result of the user demand prediction by the following formula,
wherein C is ik N is the total amount of resource k on server i mi (t-1) waiting for M types of user requests on the server i for t-1 time units, M being the upper limit of the types of user requests, D mk D, the amount of consumed resource k for m-class user requests nk Requesting the amount of the resource k required by n for the user;
step 2, calculating the utilization rate of the server:
monitoring equipment is arranged at each server node, the resource use data of each server is collected according to the operation period, the user request which is being executed on each server is analyzed, and the current actual utilization rate of the server is calculated, specifically comprising the following steps:
1) Collecting the number P of user requests being executed on each server according to the characteristic that the public culture cloud server node data changes along with time and the use data of resources according to the operation period i ；
2) According to the number P of user requests being executed by each server i Calculating the actual utilization rate eta of each server i ；
Wherein, the servers can jointly run P in parallel i Individual user requirements, P i More than 0 is a positive integer, N is the number of total servers in public culture cloud, the denominator represents the maximum number of user requests which can be executed by the servers in parallel, and the numerator represents the number of user requirements executed at the current stage, when eta i When=1, the representative server is in saturation; when eta i When=0, no executing user demand exists on the representative server, and the representative server is in a non-saturated state;
step 3, selecting a resource pre-scheduling server:
1) Judging whether the servers are in a saturated state or not according to the actual utilization rate of each server, and selecting a user requirement n to be allocated if the servers are in a non-saturated state;
(1) according to the m-class user demand number in the t time, calculating the total user demand number to be distributed in the t time, wherein the calculation formula is as follows:
L m (t)＝L m (t-1)+A m (t)-H m (t)
wherein L is m (t) is the number of m-class user demands in t time, A m (t) new arrival within t timeUp to m kinds of user demand number, H m (t) is the number of m types of user demands completed in t time, and L (t) is the total number of user demands to be distributed in t time;
(2) if the sum of the total user demand number to be allocated and the m types of user demands waiting to be allocated in the time unit t-1 is not greater than the length of the waiting queue, selecting the user demand with the largest required resource in the user demands to be allocated as n, otherwise, selecting the user demand with the largest task weight to be allocated as n;
wherein Max is ql To wait for the queue length, N mi (t-1) class m resources waiting to be allocated for t-1 time units, W a For the weight of task a to be assigned, T a For the time that the user demand a waits in the waiting sequence to be allocated,the resource amount required by the user demand a to be allocated;
2) Selecting a server which processes the user demand when the server is in a non-saturated state;
(1) the average load of each server when the regional user demand n is distributed to the server set is calculated, and the calculation formula is as follows:
wherein the method comprises the steps ofDenoted as load on t time cell server i, P ni (t) probability of allocation of user demand n to server i for t time units, D mk The amount of consumed resource k for class m user demand, D nk For the amount of resources k required by the user demand N, N mi (t-1) class m resources waiting to be allocated for t-1 time units, avg DC (t) the average load of each server when the user demand N is distributed to the server set, wherein N is the number of public culture cloud total servers;
(2) when the user demand n is distributed to each server, the load variance of the server is calculated, and the calculation formula is as follows:
wherein N is the number of public culture cloud total servers,expressed as load on t time cell server i, avg DC (t) an average load for each server when the user demand n is assigned to the set of servers;
(3) analyzing the load of the server, and selecting the server with the smallest load variance as the server for processing the user demand;
3) If the servers are in a saturated state, the waiting time of all the servers is obtained according to the execution time of the server user requests, the queuing number of the server user requests and the operation user request number of the servers, and the server with the shortest waiting time is selected to process the user demands. Detecting the running state of the resource according to the resource monitoring equipment, and simultaneously obtaining a corresponding parameter value;
the resource waiting time of the server is calculated, and the calculation formula is as follows:
wherein P is i Representing the number of user demands that the server is currently executing,indicating the task queuing number of the server, I i Indicating the time required for the server to fulfill a user's demand;
step 4, link selection:
measuring the balance degree of the flow in the network according to the principle of minimizing the maximum bandwidth utilization rate (max-min), and selecting the path with the minimum maximum link bandwidth utilization rate as a resource scheduling link;
abstract describing public culture cloud to user network topology by using a directed graph G= (V, E), wherein V represents a set of nodes in a network, and E represents a network link set; the number of nodes and the number of links are respectively represented by M and N, namely M= |V| and N= |E|; a path P from a source public culture cloud s to a destination user terminal t is formed by a group of non-repeated links (I 1 ，I 2 ，I 3 ,…,I n ) Composition; for the followingAll correspond to a length value, use +.>A representation;
the sum of all length values that make up the path is:
the bandwidth utilization of path P is:
where len (P) denotes the length of path P, isIndicating that traffic is passing through link l i Bandwidth utilization of the back link.
In the step 4, the new link problem is abstracted as follows: in a directed graph G, a path P from a source public culture cloud s to a destination user t is found, and an upper length limit D is specified to enable the path P to meet the following conditions:
len(P)≤D
the length of each path does not exceed the long limit, the service quality of the user is guaranteed, and the path with the smallest maximum link bandwidth utilization rate is selected as the resource scheduling link.
The beneficial effects of the invention are as follows:
the method selects a resource pre-scheduling server by calculating a server load variance, and selects an optimal resource pre-scheduling link by using a principle of minimizing maximum bandwidth utilization (max-min). The efficiency of resource scheduling is greatly improved, and the problem that resources cannot be fully utilized is solved.
Drawings
FIG. 1 is a schematic flow chart of the present invention;
fig. 2 is a schematic diagram of a resource scheduling network structure.
Detailed Description
The invention is further described below with reference to the drawings and examples.
As shown in fig. 1 and 2, a resource pre-scheduling method based on load balancing specifically includes the following steps:
step 1: and determining a resource server set matched with the user demand according to the result of the user demand prediction.
The calculation formula is as follows:
wherein C is ik N is the total amount of resource k on server i mi (t-1) is t-1 timeThe unit waits for M types of user requests on the server i, M is the upper limit of the types of the user requests, and D mk D, the amount of consumed resource k for m-class user requests nk Requesting the amount of the resource k required by n for the user;
step 2: monitoring equipment is arranged at each server node, the resource use data of each server are collected according to the operation period, the user request which is being executed on each server is analyzed, and the current actual utilization rate of the server is calculated.
Further, the step S2 specifically includes the following steps:
s21: collecting the number P of user requests being executed on each server according to the characteristic that the public culture cloud server node data changes along with time and the use data of resources according to the operation period i ；
S22: according to the number P of user requests being executed by each server i Calculating the actual utilization rate eta of each server i ；
Wherein, the servers can jointly run P in parallel i Individual user requirements, P i More than 0 is a positive integer, N is the number of total servers in public culture cloud, the denominator represents the maximum number of user requests which can be executed by the servers in parallel, and the numerator represents the number of user requirements executed at the current stage, when eta i When=1, the representative server is in saturation; when eta i When=0, no executing user demand exists on the representative server, and the representative server is in a non-saturated state;
step 3: and judging the saturation state of the server according to the current actual utilization rate of the server, if the server is in the unsaturated state, calculating the load variance of the server when the user demand prediction result is distributed to the server set, and selecting the server with the smallest variance as the server for processing the user demand. If the servers are in a saturated state, the waiting time of all the servers is obtained according to the execution time of the server user requests, the queuing number of the server user requests and the operation user request number of the servers, and the server with the shortest waiting time is selected.
Further, the step S3 specifically includes the following steps:
s31: judging whether the servers are in a saturated state or not according to the actual utilization rate of each server, and selecting a user requirement n to be allocated if the servers are in a non-saturated state;
1) According to the m-class user demand number in the t time, calculating the total user demand number to be distributed in the t time, wherein the calculation formula is as follows:
L m (t)＝L m (t-1)+A m (t)-H m (t)
wherein L is m (t) is the number of m-class user demands in t time, A m (t) is the number of m types of user demands newly arrived in t time, H m (t) is the number of m types of user demands completed in t time, and L (t) is the total number of user demands to be distributed in t time;
2) If the sum of the total user demand number to be allocated and the m types of user demands waiting to be allocated in the time unit t-1 is not greater than the length of the waiting queue, selecting the user demand with the largest required resource in the user demands to be allocated as n, otherwise, selecting the user demand with the largest task weight to be allocated as n;
wherein Max is ql To wait for the queue length, N mi (t-1) class m resources waiting to be allocated for t-1 time units, W a For the weight of task a to be assigned, T a For the time that the user demand a waits in the waiting sequence to be allocated,the resource amount required by the user demand a to be allocated;
s32: selecting a server which processes the user demand when the server is in a non-saturated state;
1) The average load of each server when the regional user demand n is distributed to the server set is calculated, and the calculation formula is as follows:
wherein the method comprises the steps ofDenoted as load on t time cell server i, P ni (t) probability of allocation of user demand n to server i for t time units, D mk The amount of consumed resource k for class m user demand, D nk For the amount of resources k required by the user demand N, N mi (t-1) class m resources waiting to be allocated for t-1 time units, avg DC (t) the average load of each server when the user demand N is distributed to the server set, wherein N is the number of public culture cloud total servers;
2) When the user demand n is distributed to each server, the load variance of the server is calculated, and the calculation formula is as follows:
3) Analyzing the load of the server, and selecting the server with the smallest load variance as the server for processing the user demand;
wherein N is the number of public culture cloud total servers,denoted as t time cell serviceLoad on device i, avg DC (t) an average load for each server when the user demand n is assigned to the set of servers;
s32: if the servers are in a saturated state, calculating the waiting time of all the servers, and selecting the server with the shortest waiting time to process the user demands. Detecting the running state of the resource according to the resource monitoring equipment, and simultaneously obtaining a corresponding parameter value;
the resource waiting time of the server is calculated, and the calculation formula is as follows:
wherein P is i Representing the number of user demands that the server is currently executing,indicating the task queuing number of the server, I i Indicating the time required for the server to fulfill a user's demand;
s4: measuring the balance degree of the flow in the network according to the principle of minimizing the maximum bandwidth utilization rate (max-min), and selecting the path with the minimum maximum link bandwidth utilization rate as a resource scheduling link;
and abstractly describing the network topology from the public culture cloud to the user side by using a directed graph G= (V, E). V denotes a set of nodes in the network and E denotes a set of network links. The number of nodes and the number of links are respectively represented by M and N, namely M= |V| and N= |E|. A path P from a source public culture cloud s to a destination user terminal t is formed by a group of non-repeated links (I 1 ，I 2 ，I 3 ,…,I n ) Composition; for the followingAll correspond to a length value, use +.>A representation;
the sum of all length values that make up the path is:
the bandwidth utilization of path P is:
where len (P) denotes the length of path P, isIndicating that traffic is passing through link l i Bandwidth utilization of the back link.
Abstracting the selection of a new link problem as: in a directed graph G, a path P from a source public culture cloud s to a destination user t is found, and an upper length limit D is specified to enable the path P to meet the following conditions:
len(P)≤D
the length of each path does not exceed the long limit, the service quality of the user is guaranteed, and the path with the smallest maximum link bandwidth utilization rate is selected as the resource scheduling link.

Claims (2)

1. The resource pre-scheduling method based on load balancing is characterized by comprising the following steps:
step 1, server set selection:
determining a resource server set matched with the user demand according to the result of the user demand prediction by the following formula,
wherein C is ik N is the total amount of resource k on server i mi (t-1) waiting for M types of user requests on the server i for t-1 time units, M being the upper limit of the types of user requests, D mk D, the amount of consumed resource k for m-class user requests nk Requesting the amount of the resource k required by n for the user;
step 2, calculating the utilization rate of the server:
monitoring equipment is arranged at each server node, the resource use data of each server is collected according to the operation period, the user request which is being executed on each server is analyzed, and the current actual utilization rate of the server is calculated, specifically comprising the following steps:
1) Collecting the number P of user requests being executed on each server according to the characteristic that the public culture cloud server node data changes along with time and the use data of resources according to the operation period i ；
2) According to the number P of user requests being executed by each server i Calculating the actual utilization rate eta of each server i ；
Wherein, the servers can jointly run P in parallel i Individual user requirements, P i More than 0 is a positive integer, N is the number of total servers in public culture cloud, the denominator represents the maximum number of user requests which can be executed by the servers in parallel, and the numerator represents the number of user requirements executed at the current stage, when eta i When=1, the representative server is in saturation; at the current eta i When=0, no executing user demand exists on the representative server, and the representative server is in a non-saturated state;
step 3, selecting a resource pre-scheduling server:
1) Judging whether the servers are in a saturated state or not according to the actual utilization rate of each server, and selecting a user requirement n to be allocated if the servers are in a non-saturated state;
(1) according to the m-class user demand number in the t time, calculating the total user demand number to be distributed in the t time, wherein the calculation formula is as follows:
L m (t)＝L m (t-1)+A m (t)-H m (t)
wherein L is m (t) is the number of m-class user demands in t time, A m (t) is the number of m types of user demands newly arrived in t time, H m (t) is the number of m types of user demands completed in t time, and L (t) is the total number of user demands to be distributed in t time;
(2) if the sum of the total user demand number to be allocated and the m types of user demands waiting to be allocated in the time unit t-1 is not greater than the length of the waiting queue, selecting the user demand with the largest required resource in the user demands to be allocated as n, otherwise, selecting the user demand with the largest task weight to be allocated as n;
wherein Max is ql To wait for the queue length, N mi (t-1) class m resources waiting to be allocated for t-1 time units, W a For the weight of task a to be assigned, T a For the time that the user demand a waits in the waiting sequence to be allocated,the resource amount required by the user demand a to be allocated;
2) Selecting a server which processes the user demand when the server is in a non-saturated state;
(1) the average load of each server when the regional user demand n is distributed to the server set is calculated, and the calculation formula is as follows:
wherein the method comprises the steps ofRepresented as load on t time cell server i, p ni (t) probability of allocation of user demand n to server i for t time units, D mk The amount of consumed resource k for class m user demand, D nk For the amount of resources k required by the user demand N, N mi (t-1) class m resources waiting to be allocated for t-1 time units, avg DC (t) the average load of each server when the user demand N is distributed to the server set, wherein N is the number of public culture cloud total servers;
(2) when the user demand n is distributed to each server, the load variance of the server is calculated, and the calculation formula is as follows:
wherein N is the number of public culture cloud total servers,expressed as load on t time cell server i, avg DC (t) an average load for each server when the user demand n is assigned to the set of servers;
(3) analyzing the load of the server, and selecting the server with the smallest load variance as the server for processing the user demand;
3) If the servers are in a saturated state, according to the execution time of the server user requests, the queuing quantity of the server user requests and the running user request quantity of the servers, waiting time of all the servers is obtained, and the server with the shortest waiting time is selected to process the user demands; detecting the running state of the resource according to the resource monitoring equipment, and simultaneously obtaining a corresponding parameter value;
the resource waiting time of the server is calculated, and the calculation formula is as follows:
wherein P is i Representing the number of user demands that the server is currently executing,indicating the task queuing number of the server, I i Indicating the time required for the server to fulfill a user's demand;
step 4, link selection:
measuring the balance degree of the flow in the network according to the principle of minimizing the maximum bandwidth utilization rate (max-min), and selecting the path with the minimum maximum link bandwidth utilization rate as a resource scheduling link;
abstract describing public culture cloud to user network topology by using a directed graph G= (V, E), wherein V represents a set of nodes in a network, and E represents a network link set; the number of nodes and the number of links are respectively represented by M and N, namely M= |V| and N= |E|; a path P from a source public culture cloud s to a destination user terminal t is formed by a group of non-repeated links (I 1 ，I 2 ，I 3 ,…,I n ) Composition; for the followingAll correspond to a length value, use +.>A representation;
the sum of all length values that make up the path is:
the bandwidth utilization of path P is:
where len (P) denotes the length of path P, isIndicating that traffic is passing through link l i Bandwidth utilization of the back link.
2. The method for pre-scheduling resources based on load balancing according to claim 1, wherein the step 4 abstracts selecting a new link problem as follows: in a directed graph G, a path P from a source public culture cloud s to a destination user t is found, and an upper length limit D is specified to enable the path P to meet the following conditions:
len(P)≤D
the length of each path does not exceed the long limit, the service quality of the user is guaranteed, and the path with the smallest maximum link bandwidth utilization rate is selected as the resource scheduling link.
CN202110863510.1A 2021-07-29 2021-07-29 Resource pre-scheduling method based on load balancing Active CN113946436B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110863510.1A CN113946436B (en) 2021-07-29 2021-07-29 Resource pre-scheduling method based on load balancing

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110863510.1A CN113946436B (en) 2021-07-29 2021-07-29 Resource pre-scheduling method based on load balancing

Publications (2)

Publication Number Publication Date
CN113946436A CN113946436A (en) 2022-01-18
CN113946436B true CN113946436B (en) 2024-03-08

Family

ID=79327727

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110863510.1A Active CN113946436B (en) 2021-07-29 2021-07-29 Resource pre-scheduling method based on load balancing

Country Status (1)

Country Link
CN (1) CN113946436B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115811317A (en) * 2022-10-17 2023-03-17 中国人民大学 Stream processing method and system based on self-adaptive non-decompression direct calculation
CN115617279B (en) * 2022-12-13 2023-03-31 北京中电德瑞电子科技有限公司 Distributed cloud data processing method and device and storage medium
CN116048821B (en) * 2023-04-03 2023-06-16 深圳市昊源诺信科技有限公司 High-utilization AI server and resource allocation method thereof

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102185759A (en) * 2011-04-12 2011-09-14 田文洪 Multi-physical server load equalizing method and device capable of meeting requirement characteristic
CN104023042A (en) * 2013-03-01 2014-09-03 清华大学 Cloud platform resource scheduling method
CN107770096A (en) * 2017-12-11 2018-03-06 国网河南省电力公司信息通信公司 A kind of SDN/NFV network dynamic resource allocation algorithms based on load balancing
CN112650579A (en) * 2020-10-30 2021-04-13 西北大学 Gridding-based cloud resource distribution scheduling method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102185759A (en) * 2011-04-12 2011-09-14 田文洪 Multi-physical server load equalizing method and device capable of meeting requirement characteristic
CN104023042A (en) * 2013-03-01 2014-09-03 清华大学 Cloud platform resource scheduling method
CN107770096A (en) * 2017-12-11 2018-03-06 国网河南省电力公司信息通信公司 A kind of SDN/NFV network dynamic resource allocation algorithms based on load balancing
CN112650579A (en) * 2020-10-30 2021-04-13 西北大学 Gridding-based cloud resource distribution scheduling method

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
一种基于负载均衡的云资源调度方法;陈斌;甘茂林;李娟;;计算机技术与发展(第06期);全文 *
云计算环境下虚拟机资源均衡调度方法研究;李济汉;李素粉;张云勇;房秉毅;;电信科学(第04期);全文 *
基于OpenStack的清华云平台构建与调度方案设计;赵少卡;李立耀;凌晓;徐聪;杨家海;;计算机应用(第12期);全文 *

Also Published As

Publication number Publication date
CN113946436A (en) 2022-01-18

Similar Documents

Publication Publication Date Title
CN113946436B (en) Resource pre-scheduling method based on load balancing
Chen et al. Performance analysis and uplink scheduling for QoS-aware NB-IoT networks in mobile computing
US20170142177A1 (en) Method and system for network dispatching
CN107038071B (en) Storm task flexible scheduling algorithm based on data flow prediction
CN110990159A (en) Historical data analysis-based container cloud platform resource quota prediction method
CN109617826A (en) A kind of storm dynamic load balancing method based on cuckoo search
CN110351571A (en) Live video cloud transcoding resource allocation and dispatching method based on deeply study
CN107566535B (en) Self-adaptive load balancing method based on concurrent access timing sequence rule of Web map service
CN108021447B (en) Method and system for determining optimal resource strategy based on distributed data
WO2024094104A1 (en) Dynamic feedback weighted cloud storage resource scheduling method, apparatus and device
CN116389491B (en) Cloud edge computing power resource self-adaptive computing system
Li et al. Edge cloud resource expansion and shrinkage based on workload for minimizing the cost
CN113822456A (en) Service combination optimization deployment method based on deep reinforcement learning in cloud and mist mixed environment
Kang et al. Application of adaptive load balancing algorithm based on minimum traffic in cloud computing architecture
CN109861850A (en) A method of the stateless cloud workflow load balance scheduling based on SLA
CN105119751A (en) Service evaluation and selection method based on environment real-time perceiving
CN110990160B (en) Static security analysis container cloud elastic telescoping method based on load prediction
CN113596868A (en) 5G network slice resource management mechanism based on SDN and NFV
Zheng et al. Dynamic load balancing and pricing in grid computing with communication delay
CN117493020A (en) Method for realizing computing resource scheduling of data grid
CN110262880B (en) Distributed data center energy consumption overhead optimization-oriented job scheduling method
CN113032146A (en) Robust service supply method for multi-access edge computing environment
Milocco et al. Evaluating the upper bound of energy cost saving by proactive data center management
Li et al. Self-adaptive load-balancing strategy based on a time series pattern for concurrent user access on Web map service
CN117076117A (en) Intelligent media meeting place scheduling method and system based on new communication architecture

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant