Search

CN-115658230-B - Cloud data center high-efficiency container arrangement method and system

CN115658230BCN 115658230 BCN115658230 BCN 115658230BCN-115658230-B

Abstract

The invention discloses a cloud data center high-efficiency container arrangement method and system, which comprise the steps of obtaining overall information of a cloud data center, including the number of servers, the upper limit of resources of each server, the minimum power of each server and the maximum power of each server, and the average migration network bandwidth budget provided by the cloud data center, obtaining information of all containers in a section of continuous time slot, including the number of containers, the resource request quantity of each container, the mirror image size of each container, the running time of each container and the migration attribute of each container, constructing an optimization problem aiming at minimizing the average energy consumption of the cloud data center according to the information of all containers in the section of continuous time slot and the overall information of the cloud data center, and solving the optimization problem to obtain a container arrangement decision of the section of continuous time slot. The method and the device can obtain the current time slot container arrangement decision under the condition that the number of future container creation requests and the container specification parameters cannot be accurately known and various resource constraint conditions are met, and are efficient, energy-saving and environment-friendly.

Inventors

  • QIAN ZHUZHONG
  • WEI SHENGJIE

Assignees

  • 南京大学

Dates

Publication Date
20260508
Application Date
20221027

Claims (8)

  1. 1. A cloud data center high-efficiency container arrangement method, characterized by comprising the following steps: Acquiring overall information of a cloud data center, wherein the overall information comprises the number of servers, the upper limit of resources of each server, the minimum power of each server and the maximum power of each server, and the average migration network bandwidth budget is provided by the cloud data center; acquiring information of all containers in a section of continuous time slot, wherein the information comprises the number of the containers, the resource request quantity of each container, the mirror image size of each container, the running time of each container and the migration attribute of each container, and the section of continuous time slot is from time slot 1 to time slot T, and T is more than 1; According to the information of all containers in the section of continuous time slot and the overall information of the cloud data center, constructing and solving an optimization problem aiming at minimizing the average energy consumption of the cloud data center, and taking the solving result as a container arrangement decision of the section of continuous time slot, wherein the container arrangement decision is an arrangement deployment position of each container in each time slot from a starting time slot 1 to a time slot T; the construction of the optimization problem (1) aiming at minimizing the average energy consumption of the cloud data center is as follows: optimization target: Constraint conditions: (1.1) for any one of the containers Scheduling decision constraints: (1.2) for any one of the containers Scheduling decision constraints during run time: (1.3) for any one of the containers The decision constraint is arranged outside the running time: (1.4) for any one server Computing resource constraints: (1.5) for any one server Memory resource constraint: (1.6) for any one of the containers Migration attribute constraints: (1.7) long-term migration network bandwidth constraints: in the formula, Representing a slave integer To integers of Is a collection of (3); Representing the total number of time slots of the time axis; The total number of servers in the cloud data center; from time slot 1 to time slot The total number of running instances of the inner container; For time slot t inner container Whether or not to arrange to a server Is a decision of (2); the sum of power consumption of all servers in the cloud data center in the time slot t; Is a container The first time slot of operation; Is a container The last time slot of operation; Is a container Is used for calculating the resource request quantity; Is a container Is a memory resource request amount; Is a server Is calculated by the total amount of resources; Is a server Is a total amount of memory resources; Is a container A migration attribute characterization variable of (1); bandwidth expense of the migration network caused by scheduling containers in time slot t; An average migration network bandwidth budget provided for the cloud data center; wherein the sum of power consumption of all servers in the cloud data center in time slot t The method comprises the following steps: Wherein: Wherein: in the formula, Representation server Computing resource usage at time slot t; representation server Energy consumption at time slot t; representation server Is the lowest power consumption of (1); representation server Maximum power consumption of (2); wherein bandwidth overhead of the migration network caused by the container arrangement in time slot t The method comprises the following steps: Wherein the method comprises the steps of Is a container Characterization variable of whether migration occurs at time slot t: In the middle of Indicating container Is a mirror image of the size of (a).
  2. 2. The cloud data center high-efficiency container orchestration method according to claim 1, wherein translating the optimization problem (1) that is built with the goal of minimizing average energy consumption of the cloud data center, comprises: decomposing the optimization problem (1) into each time slot, and at the beginning of each time slot, actually solving the following optimization problem (2): optimization target: ) constraint conditions including constraint conditions (1.1) to constraint conditions (1.6) In the formula, Is a virtual queue variable, its initial value ; Is a preset weight adjustment parameter.
  3. 3. The cloud data center high-efficiency container arrangement method according to claim 2, wherein solving the container arrangement optimization problem (2) in the present time slot to obtain a final container arrangement decision of the present time slot specifically comprises: L1) obtaining an initial container layout decision for the time slot ; L2) initializing container orchestration decisions by loop iteration On the basis of (a) to obtain the final container arrangement decision of the time slot ; L3) final container scheduling decision according to the obtained present time slot Updating virtual queue variables 。
  4. 4. The method for efficient container organization for cloud data centers as recited in claim 3, Said step L1) obtaining an initial container scheduling decision for the time slot Comprising: L11) obtaining a container orchestration decision for the previous time slot; L12) on the basis of the container arrangement decision of the previous time slot, keeping the arrangement position of the operated container unchanged, and for each newly arrived container in the time slot, arranging the container to a server with available computing resources and available memory resources meeting the requirements of the container resources at equal probability randomly, thereby obtaining an initial container arrangement decision of the time slot 。
  5. 5. The method for efficient container organization for cloud data centers as recited in claim 3, Said step L2) initializing container layout decisions by loop iteration On the basis of (a) to obtain the final container arrangement decision of the time slot Comprising making decisions on the initial container Proceeding with The sub-loop iterates, continuously updating the container arrangement decisions until a final container arrangement decision is obtained The loop iteration specifically comprises the following steps: In the first place After the round-robin iteration is finished, the obtained container arrangement decision is that ; Based on container arrangement decisions In newly arrived containers or containers which are running and have migration attribute characterization variables of 1, an equal probability randomly selects one container ; Among all the servers with available computing resources and available memory resources meeting the container requirement, one server is randomly selected with equal probability ; The container is put into Is set as a server Thereby at On the basis of (a) obtaining container arrangement decisions ; For container orchestration decisions And Respectively carrying calculation expressions related to optimization targets of the optimization problem (2) to obtain results And ; Order the Wherein For the policy transfer smoothing parameter(s), Representation selection As a means of Probability of (2) is such that Is a function of the probability of (1), Is that To Is a function of the probability of (1), Is that ; Completion of After the minor loop iteration, the final container arrangement decision of the time slot is obtained 。
  6. 6. The method for efficient container organization for cloud data centers as recited in claim 3, Said step L3) is based on the resulting final container layout decision for the present time slot Updating virtual queue variables The method specifically comprises the following steps: Final container scheduling decision based on the resulting present time slot Calculating bandwidth expense of the migration network caused by arranging containers in the time slot according to the bandwidth expense Updating virtual queue variables Wherein [ ] + refers to max {, 0}.
  7. 7. A cloud data center high-efficiency container orchestration system, comprising: The system initial unit is configured to initialize the original numerical value of the dynamic parameter; an input acquisition unit configured to acquire information as input to other modules; A decision storage unit configured to acquire and store container arrangement decisions for each time slot; The arrangement decision feedback unit is configured to return a deployment effect corresponding to an arrangement decision of a certain container; An arrangement decision generation unit configured to obtain a container arrangement decision within the present time slot; optimization target: Constraint conditions: (1.1) for any one of the containers Scheduling decision constraints: (1.2) for any one of the containers Scheduling decision constraints during run time: (1.3) for any one of the containers The decision constraint is arranged outside the running time: (1.4) for any one server Computing resource constraints: (1.5) for any one server Memory resource constraint: (1.6) for any one of the containers Migration attribute constraints: (1.7) long-term migration network bandwidth constraints: in the formula, Representing a slave integer To integers of Is a collection of (3); Representing the total number of time slots of the time axis; The total number of servers in the cloud data center; from time slot 1 to time slot The total number of running instances of the inner container; For time slot t inner container Whether or not to arrange to a server Is a decision of (2); the sum of power consumption of all servers in the cloud data center in the time slot t; Is a container The first time slot of operation; Is a container The last time slot of operation; Is a container Is used for calculating the resource request quantity; Is a container Is a memory resource request amount; Is a server Is calculated by the total amount of resources; Is a server Is a total amount of memory resources; Is a container A migration attribute characterization variable of (1); bandwidth expense of the migration network caused by scheduling containers in time slot t; An average migration network bandwidth budget provided for the cloud data center; The input acquisition unit comprises a data center information acquisition module and a container information acquisition module; the data center information acquisition module is configured to acquire total information of the cloud data center, wherein the total information comprises the number of servers, the upper limit of resources of each server, the minimum power of each server, the maximum power of each server and the average migration network bandwidth budget provided by the cloud data center; The system comprises a container information acquisition module, a storage module and a storage module, wherein the container information acquisition module is configured to acquire all container information from a starting time slot to a current time slot, wherein the container information comprises the number of containers, the resource request quantity of each container, the mirror image size of each container, the running time of each container and the migration attribute of each container; the arrangement decision feedback unit comprises an arrangement decision energy consumption feedback module and an arrangement decision migration expense feedback module; wherein the arrangement decision energy consumption feedback module is configured to return, for a container arrangement decision in a time slot, the sum of all server power consumption in the cloud data center under the container arrangement decision in the time slot according to the total information of the cloud data center The method comprises the following steps: Wherein: Wherein: in the formula, Representation server Computing resource usage at time slot t; representation server Energy consumption at time slot t; representation server Is the lowest power consumption of (1); representation server Maximum power consumption of (2); The total number of servers in the cloud data center; from time slot 1 to time slot The total number of running instances of the inner container; For time slot t inner container Whether or not to arrange to a server Is a decision of (2); Is a container Is used for calculating the resource request quantity; wherein the orchestration decision migration expense feedback module is configured to return, for a certain orchestration decision of a container in a time slot, all migration network bandwidth expense caused by the certain orchestration decision of the time slot according to final orchestration decision information of the previous time slot The method comprises the following steps: Wherein the method comprises the steps of Is a container Characterization variable of whether migration occurs at time slot t: in the formula, Representing a slave integer To integers of Is a collection of (3); Indicating container Is the mirror size of (a); Is a container The first time slot of operation; Is a container The last time slot of operation; Is a container Is a memory resource request amount.
  8. 8. The cloud data center efficient container orchestration system according to claim 7, wherein the orchestration decision generation unit comprises an optimization problem construction module and an optimization problem solving module: The optimization problem constructing module is configured to construct the optimization problem of the arrangement of the time slot container by adopting the method as claimed in claim 1 or 2 according to the container request reached by the time slot and the container arrangement decision of the cloud data center of the last time slot; The optimization problem solving module is configured to solve the arrangement optimization problem in the time slot by adopting the method as claimed in any one of claims 3-6, so as to obtain the final arrangement decision of the time slot.

Description

Cloud data center high-efficiency container arrangement method and system Technical Field The invention relates to the field of cloud computing, in particular to a container arrangement method and system of a cloud data center. Background Cloud computing is a novel business computing model that pools resources of servers in a data center through virtualization technology, thereby providing computing resources in the form of on-demand services over the internet. Enterprises or individual users deploy the business cloud according to the self requirements, and the enterprises or individual users do not need to purchase, configure or manage resources by themselves, and pay according to the quantity and time of actually used resources. Cloud computing makes it easier for businesses and individuals to deploy applications and reduces various maintenance costs. Compared with the traditional virtual machine, the container has the advantages of high start-stop speed, light weight, strong expansibility, good isolation and the like, so that the container becomes an alternative technology of the virtual machine and is widely used in the cloud data center. The container creates container instances by mirroring, each container instance having its own resource request, run-time and migration attributes. As cloud computing demands continue to increase, the scale of cloud data centers is also rapidly increasing, exposing the problem of high energy consumption of cloud data centers. The energy consumption of cloud data centers is comprised of many aspects, including server energy consumption. The energy consumption of a single server is related to the utilization rate of the computing resources of the server and the self energy consumption attribute, and the unloaded server can enter a dormant state to reduce the energy consumption. Inside a cloud data center, due to the fact that the number of containers is continuously increased and a proper high-efficiency container arrangement strategy is lacked, the containers are scattered on different servers, a large number of servers are in a low-resource utilization state, and accordingly energy consumption of the data center is increased. There is a need for a container arrangement method that reduces energy consumption in cloud data centers. However, the number of container creation requests is changed continuously over time, so that the actual number of inferred requests of future users cannot be accurately known in advance, and the optimal high-performance container arrangement strategy in long time is difficult to solve due to different running time, resource request parameters and migration attribute characteristics among containers. The existing arrangement strategy of the cloud data center container is simple to take the arrangement strategy of the virtual machine into consideration, and the migration attribute of the container is not considered, so that the energy-saving effect of the arrangement strategy cannot reach an ideal condition, and the energy consumption of the cloud data center cannot be obviously reduced. The above problems are to be solved. Disclosure of Invention The invention aims to solve the defects of the prior art, provides a cloud data center high-efficiency container arrangement method and simultaneously provides a cloud data center high-efficiency container arrangement system. The technical scheme is that in order to solve the technical problems, the high-efficiency container arrangement method for the cloud data center comprises the following steps: Acquiring overall information of a cloud data center, wherein the overall information comprises the number of servers, the upper limit of resources of each server, the minimum power of each server and the maximum power of each server, and the average migration network bandwidth budget is provided by the cloud data center; acquiring information of all containers in a section of continuous time slot, wherein the information comprises the number of the containers, the resource request quantity of each container, the mirror image size of each container, the running time of each container and the migration attribute of each container, and the section of continuous time slot is from time slot 1 to time slot T, and T is more than 1; and constructing and solving an optimization problem aiming at minimizing the average energy consumption of the cloud data center according to the information of all containers in the section of continuous time slot and the overall information of the cloud data center, and taking the solving result as a container arrangement decision of the section of continuous time slot, wherein the container arrangement decision is an arrangement deployment position of each container in each time slot from the beginning time slot 1 to the time slot T. The invention also provides a cloud data center high-efficiency container arrangement system, which comprises: The system initial unit is configured