CN117938855A - Method and system for supplying and distributing containerized service resources for cloud edge computing network system - Google Patents
Method and system for supplying and distributing containerized service resources for cloud edge computing network system Download PDFInfo
- Publication number
- CN117938855A CN117938855A CN202410162543.7A CN202410162543A CN117938855A CN 117938855 A CN117938855 A CN 117938855A CN 202410162543 A CN202410162543 A CN 202410162543A CN 117938855 A CN117938855 A CN 117938855A
- Authority
- CN
- China
- Prior art keywords
- container
- request information
- cloud
- edge
- request
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 69
- 238000012545 processing Methods 0.000 claims abstract description 45
- 230000008569 process Effects 0.000 claims description 22
- 230000004044 response Effects 0.000 claims description 18
- 238000009826 distribution Methods 0.000 claims description 10
- 238000003860 storage Methods 0.000 claims description 10
- 238000005265 energy consumption Methods 0.000 claims description 9
- 230000000694 effects Effects 0.000 claims description 6
- 230000008859 change Effects 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 abstract description 11
- 238000010586 diagram Methods 0.000 description 10
- 238000004590 computer program Methods 0.000 description 8
- 230000006870 function Effects 0.000 description 7
- 230000008901 benefit Effects 0.000 description 5
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 238000013468 resource allocation Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 230000004075 alteration Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000003190 augmentative effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000011144 upstream manufacturing Methods 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a method and a system for supplying and distributing containerized service resources for a cloud computing network system, wherein the method comprises the following steps: the cloud edge computing network system receives user request information, constructs a global controller and selects an edge cloud for providing resources according to the user request information; an admission controller is constructed on the calculation point, and the admission request information is put into a request queue; and processing the request information in the request queue by adopting an earliest deadline priority scheduling strategy to realize service resource supply allocation. The method of the invention ensures that the calculation request provided by the terminal is timely processed by the calculation points positioned on the paths from the terminal, the edge node and the back-end cloud server, ensures that the request is processed, and provides the highest service quality for the terminal user. The method effectively improves the request rate processed in the service deadline and returned to the original user, improves the resource utilization rate in the edge cloud, and reduces the average number of new container instances started by the computing point.
Description
Technical Field
The invention relates to the technical field of distributed computing, in particular to a method and a system for supplying and distributing containerized service resources for a cloud computing network system.
Background
Currently, based on centralized cloud computing systems deployed on remote large infrastructures, challenges are faced with increasing demands from emerging applications requiring stringent quality of service requirements, such as latency, including autopilot or unmanned, virtual reality or augmented reality, interactive gaming, and the like. The cloud data center resource is excessively depended, so that the problem of difficult guarantee of service quality is caused, a large number of resources such as computing power, storage space and the like which are far away from the edge node of the cloud center and are owned by the user side terminal equipment in the network system are always in an idle state, and massive resources are wasted.
Different from cloud computing, edge computing runs computing tasks on computing resources close to a data source, and cloud computing is lowered to the edge of a network, so that delay of a computing system is effectively reduced, data transmission bandwidth is reduced, and cloud computing center pressure is relieved. An implementation of edge computing is to build a distributed edge cloud infrastructure at the network edge near the end user in order to deploy services at the edge and meet their QoS requirements. An edge cloud consists of distributed edge computation points, each of which has a limited amount of resources compared to the centralized cloud infrastructure.
The cloud edge computing network system is composed of a central cloud, a plurality of edge clouds, a plurality of independent edge servers and a plurality of network terminals, wherein all the levels of computing and storage resources contained in the central cloud server, the network edge cloud server and the network terminals at the rear end are organically aggregated together, so that the respective geographic, performance and cost advantages of the cloud center, the network edge and the user terminals are fully exerted, tasks are decomposed and packaged, then the tasks are orderly deployed on different nodes to finish various complex user applications according to the needs, and the mode goes from a 'resource centralized sharing mode' to a 'distributed mutual sharing mode', so that the computing targets such as high efficiency, low cost and maximized resource utilization are truly achieved.
In a cloud edge computing network system, an application service provider provides its containerized service functions to computing points, while end users submit time-sensitive computing requests with strict response delay deadlines. The response delay is the delay between submitting a computation request to the network and receiving the computation result. The cloud edge computing network system infrastructure represents the responsibility for resource provisioning and allocation for the application service provider, and then charges the application provider for actual resource usage. The resource management module of the cloud edge computing network system can carefully determine the number of instances provided for each application program and select the scheduled computing points so as to improve the resource utilization rate of the computing points and meet the response deadline requirement of the terminal user request.
Disclosure of Invention
The present invention has been made in view of the above-described problems.
Therefore, the technical problems solved by the invention are as follows: the cloud computing network system must consider the request response delay requirement, the service container cold start provisioning overhead and the communication and coordination overhead simultaneously in the service provisioning and terminal request scheduling when achieving the management goal.
In order to solve the technical problems, the invention provides the following technical scheme: the method for supplying and distributing the containerized service resources facing the cloud edge computing network system comprises the following steps: the cloud edge computing network system receives user request information, constructs a global controller and selects an edge cloud for providing resources according to the user request information; an admission controller is constructed on the computing point of the edge cloud, and the admission request information is admitted and put into a request queue; and processing the request information in the request queue by adopting an earliest deadline priority scheduling strategy to realize service resource supply allocation.
As a preferable scheme of the containerized service resource supply and distribution method facing the cloud computing network system, the invention comprises the following steps: the global controller monitors the latest states of all edge clouds, including the input queue size of the edge clouds, the state of an active pool instance, the resource utilization rate, the network bandwidth and delay, the service response time and the energy consumption; the computing point comprises a container activity pool for providing a plurality of containers with different services, when the request information reaches the computing point, the admission controller judges the deadline of the request information and the calculated container instance, and when the time used by the computing point for processing the request information is less than or equal to the deadline of the request information and the computing point has the container instance for processing the request information, the request information is admitted and put into a request queue; when the time for processing the request information by the computing point is longer than the deadline of the request information or the computing point does not have the container instance to process the request information, the computing point refuses the request information, and the refused request information is transferred to the next computing point along the path from the computing point to the back-end cloud, and is judged by an admission controller of the next computing point.
As a preferable scheme of the containerized service resource supply and distribution method facing the cloud computing network system, the invention comprises the following steps: the container instance comprises the steps that the container instance is arranged on a CPU core, the container instance is started to occupy tasks in a period of time slot of the CPU core, and the number of cores occupied by loading a container image on an edge cloud is expressed as:
at the same time, each container in the active pool can only launch one container instance at a time, denoted as:
The containers within the time slot define a set of time-evolving over the active pool instance on the edge cloud, expressed as:
wherein, Representing the number of cores taken to load a container image; f represents a container; /(I)Representing a collection of containers; v denotes a container on the edge cloud; /(I)A collection of containers on edge cloud h; t represents a time slot; /(I)Representing a set of time slots; i represents a remaining time slot; /(I)A time slot overhead representing the launch of a container instance; /(I)A binary decision variable representing running a mirror image of container f as container instance v over time slot i; h represents an edge cloud; /(I)Representing an edge cloud set; /(I)Representing a set of container f instances on an edge cloud h within a time slot t; f' represents other containers than f; the edge cloud processes the requests for pairs by scheduling the available instances of the container on the idle core, i.e. performing the association of the container to the core, the container being in a running state, the number of running instances being defined as:
the number of running instances is limited by the total number of cores of the edge cloud and the number of cores occupied when the mirror edge cloud is loaded, and the constraint is expressed as:
wherein, A binary variable representing the association of container v with the core at time slot t; /(I)The number of thermal instances on edge cloud h for container f for time slot t is represented.
As a preferable scheme of the containerized service resource supply and distribution method facing the cloud computing network system, the invention comprises the following steps: for an edge cloud within a given time slot, the change in queue length of the request queue is expressed as,
For users with the same network connection point, the requests that the container needs to forward to the edge within the time slot t are expressed as:
The requests that arrive at the container of the queue class of the edge cloud at time slot t are expressed as:
wherein, Representing the number of requests for container f on a k-class queue on edge cloud h within time slot t; /(I)Representing a total number of requests for a k-class queue for container f to reach edge cloud h at time slot t; /(I)The total request number of k-type queues of the edge cloud h is served by the container f in the time slot t; /(I)Representing the total number of queue classes; /(I)Representing the number of requests that group g needs to forward to edge cloud h in time slot t with respect to container f; /(I)Representing the demand of the user group g for the container f at time slot t; /(I)Representing a set of user groups defined in terms of the location of the user's connection point to the network; /(I)The request representing container f on group g of edge cloud h arrives at the queue class; deltat gh represents the delay forwarding overhead from group g onto edge cloud h in time slots.
As a preferable scheme of the containerized service resource supply and distribution method facing the cloud computing network system, the invention comprises the following steps: the admission controller comprises that admission acceptance of request information by the edge cloud is associated with ensuring that it gets served within its remaining deadline, the number of requests served from the k=1 queue class must always satisfy the arriving requests and the existing requests in the queue, expressed as:
The service requests of the k queue class must adhere to the queue class priority and the number of active pool instances of the container in the edge cloud and time, expressed as:
wherein, The total number of requests for the k-class queue of the service edge cloud h for container f at time slot t is represented.
As a preferable scheme of the containerized service resource supply and distribution method facing the cloud computing network system, the invention comprises the following steps: the earliest deadline first scheduling policy includes marking the container instance as busy state when the container instance is processing the request, not accepting the request information, and if the computing point has no container instance in non-busy state, the scheduling controller puts the request information into the queuing queue until the container instance in non-busy state appears.
In a second aspect, the invention also provides a containerized service resource supply and distribution system oriented to the cloud edge computing network system, which comprises a global control module, wherein the cloud edge computing network system receives user request information, a global controller is constructed to monitor the latest states of all edge clouds, and the edge clouds for providing resources are selected according to the user request information; the admission control module is used for constructing an admission controller on a computing point of the edge cloud, judging the deadline of the request information and the computed container instance by the admission controller, admitting the request information and putting the request information into a request queue; and the scheduling control module processes the request information in the request queue by adopting an earliest deadline priority scheduling strategy to realize service resource supply allocation.
As a preferable scheme of the containerized service resource supply distribution system facing the cloud computing network system, the invention comprises the following steps: the global controller monitors the latest states of all edge clouds, including the input queue size of the edge clouds, the state of an active pool instance, the resource utilization rate, the network bandwidth and delay, the service response time and the energy consumption; the computing point comprises a container activity pool for providing a plurality of containers with different services, when the request information reaches the computing point, the admission controller judges the deadline of the request information and the calculated container instance, and when the time used by the computing point for processing the request information is less than or equal to the deadline of the request information and the computing point has the container instance for processing the request information, the request information is admitted and put into a request queue; when the time for processing the request information by the computing point is longer than the deadline of the request information or the computing point does not have the container instance to process the request information, the computing point refuses the request information, and the refused request information is transferred to the next computing point along the path from the computing point to the back-end cloud, and is judged by an admission controller of the next computing point.
In a third aspect, the present invention also provides a computing device comprising: a memory and a processor;
the memory is configured to store computer executable instructions, and the processor is configured to execute the computer executable instructions, where the computer executable instructions when executed by the processor implement the steps of the method for allocating containerized service resources for a cloud edge computing network system.
In a fourth aspect, the present invention also provides a computer readable storage medium storing computer executable instructions which when executed by a processor implement the steps of the method for allocating containerized service resources for a cloud edge computing network system.
The invention has the beneficial effects that: the method of the invention ensures that the calculation request provided by the terminal is timely processed by the calculation points positioned on the paths from the terminal, the edge node and the back-end cloud server, ensures that the request is processed, and provides the highest service quality for the terminal user. The method effectively improves the request rate processed in the service deadline and returned to the original user, improves the resource utilization rate in the edge cloud, and reduces the average number of new container instances started by the computing point.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art. Wherein:
fig. 1 is an overall flowchart of a method for allocating containerized service resources for a cloud computing network system according to an embodiment of the present invention;
fig. 2 is a tree topology structure diagram of a cloud edge computing network system according to a second embodiment of the present invention, where the tree topology structure diagram is a method for allocating containerized service resources for the cloud edge computing network system;
fig. 3 is a schematic diagram of a terminal request meeting path of a cloud computing network infrastructure of a method for allocating containerized service resources for a cloud computing network system according to a second embodiment of the present invention;
Fig. 4 is a schematic diagram of a computing point of a container example of a method for distributing a containerized service resource supply for a cloud computing network system according to a second embodiment of the present invention.
Detailed Description
So that the manner in which the above recited objects, features and advantages of the present invention can be understood in detail, a more particular description of the invention, briefly summarized above, may be had by reference to the embodiments, some of which are illustrated in the appended drawings. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
Example 1
Referring to fig. 1, for one embodiment of the present invention, there is provided a method for allocating containerized service resources for a cloud computing network system, including:
S1: and the cloud edge computing network system receives the user request information, constructs a global controller and selects an edge cloud for providing resources according to the user request information.
Further, firstly, selecting an edge cloud for providing resources by a global controller; the global controller is responsible for monitoring and managing the resource allocation of the whole network, and collecting and analyzing the state information of all edge clouds in real time, including the input queue size of the edge clouds, the state of an active pool instance, the resource utilization rate, the network bandwidth and delay, the service response time and the energy consumption.
And by adopting an edge cloud priority strategy, the edge cloud closer to a user bears most containers with high requirements on the deadline, and is only scheduled to a back-end cloud for processing when the edge cloud is overloaded so as to meet the deadline. The adoption of the edge cloud priority strategy means that the system can give priority to task allocation to the edge cloud closest to the user, response time and network congestion can be remarkably reduced, and the method is particularly suitable for delay-sensitive applications.
When the edge cloud cannot process more requests due to overload, the system can schedule part of tasks to the back-end cloud, a flexible load balancing mechanism is provided, and continuity and reliability of service are ensured.
S2: and constructing an admission controller on the computing point of the edge cloud, admitting the request information and putting the request information into a request queue.
Further, each computing point is provided with a container activity pool which comprises a plurality of containers capable of providing different services, when the request information reaches the computing point, the admission controller judges the expiration time of the request information and the calculated container instance, and when the time used by the computing point for processing the request information is less than or equal to the expiration time of the request information and the computing point exists in the container instance, the request information is admitted and put into the request queue.
When the time for processing the request information by the computing point is longer than the deadline of the request information or the computing point does not have the container instance to process the request information, the computing point refuses the request information, and the refused request information is transferred to the next computing point along the path from the computing point to the back-end cloud, and is judged by an admission controller of the next computing point.
A fixed time stamp is recorded in the service request as the remaining deadline. The remaining deadline is initially set by the terminal in its service request as an upper bound for the expected response time due to network and processing delays. Each time a request is sent to a computation point, its remaining deadline is updated by subtracting the estimated round trip time RTT to the computation point.
The calculation point periodically sends the detection data packet to the direct upstream calculation point, and then receives the detection response packet, so that accurate estimation of RTT is realized. The request remaining deadline indicates the remaining time that the request can be used for queuing and processing. Once a request is accepted and queued, its remaining deadline will be updated over time.
For request R, admission is considered if the computing point has a container that satisfies the request. Considering the admission request R includes checking the remaining deadline of R, assuming R is at the tail of the queue. The completion time of a request is the sum of the total latency in the queue and the functional processing time. The processing time of the function is fixed at each calculation point. Accepting request R if the new completion time of all requests does not exceed the remaining deadlines thereof; otherwise, request R is denied.
Further, from a group ofUser requested/>A container in which a group is defined based on its user access points, i.e. all users with the same network connection point belong to the same group. Will groupFor container/>, over time slot tThe number of requests is recorded as/>And associates each request for container f with an expiration time T f, expressed in time slots. The request for container f by the user terminal is considered to be satisfactory only if it is serviced in T f time slots and the result is returned to the user. The request processing time of all containers is fixed to one time slot.
By usingRepresenting a collection of computing points, the computing points in the collection being equipped with hardware capable of carrying any container instance. Each edge cloud/>A set/>, can be maintained in its active poolA container.
In order for a container instance to service a request, it is first arranged on one CPU core, limited to an edge cloud hC h is the maximum number of requests that can be serviced simultaneously in the edge cloud h within a single time slot. In view of edge cloud/>Position, group/>Can access h with round trip time RTT in the L gh time slot.
Starting a container instance can be viewed as taking up a CPU core for a period of time slotInternal tasks.
By usingTo represent a binary decision variable representing a mirror image of the container f as container instance/>, over time slot tRun, i.e. when container v runs f-Otherwise 0. On the edge cloud, the number of cores taken to load one container image is expressed as:
at the same time, each container in the active pool can only launch one container instance at a time, denoted as:
The containers within the time slot define a set of time-evolving over the active pool instance on the edge cloud, expressed as:
wherein, Representing the number of cores taken to load a container image; f represents a container; /(I)Representing a collection of containers; v represents a container on the edge cloud; /(I)A collection of containers on edge cloud h; t represents a time slot; /(I)Representing a set of time slots; i represents a remaining time slot; /(I)A time slot overhead representing the launch of a container instance; /(I)A binary decision variable representing running a mirror image of container f as container instance v over time slot i; h represents an edge cloud; /(I)Representing an edge cloud set; /(I)Representing a set of container f instances on an edge cloud h within a time slot t; f ′ represents a container other than f; during time slot t, the set of instances of container f on edge cloud h depends on the number of instances of f that are newly started and the number of containers that have been reassigned to other containers.
The edge cloud processes the requests for pairs by scheduling the available instances of the container on the free core, i.e. performing the container-to-core association, the container being in an operational state, and within the time slot t, the number of operational instances of container f on edge cloud h is defined as:
the number of running instances is limited by the total number of cores of the edge cloud and the number of cores occupied when the mirror edge cloud is loaded, and the constraint is expressed as:
wherein, A binary variable representing the association of container v with the core at time slot t; /(I)The number of thermal instances on edge cloud h for container f for time slot t is represented.
At each edge cloud a set ofQueues in which arriving requests are placed in the corresponding queues according to the amount of time remaining before their arrival deadlines. If the remaining time slot of a request is 1, the request enters a queue class of k=1. The edge cloud starts servicing requests from the queue class with the lowest remaining time and then continues to process the queue class with higher remaining time, i.e. k=1, 2,3, …, until its processing capacity is reached, keeping all cores in a busy state as much as possible.
For edge clouds within a given time slot, useRepresenting queue class/>The above is for container/>The change in the queue length of the request queue is expressed as:
requests belonging to the k+1 queue class and not serviced within time slot t will be forwarded into the k queue class in time slot t+1 because their time slots are reduced by 1.
For users with the same network connection point, the requests that the container needs to forward to the edge within the time slot t are expressed as:
The container forwarding request matrix defining group g and all services in time slot t is:
Time slot write off for transmitting group g forwarding request to edge cloud h is
When (when)When the queue class of (c) is within the time slot of t+Δt gh, the edge cloud h will receive the request of the container f sent by g at time slot T, and allocate the appropriate time slot for the request. For example, if there are T f =5 time slots and Δt gh =1, each request would take up to k=2 time slots in the queue when forwarded at h.
The requests that arrive at the container of the queue class of the edge cloud at time slot t are expressed as:
wherein, Representing the number of requests for container f on a k-class queue on edge cloud h within time slot t; representing a total number of requests for a k-class queue for container f to reach edge cloud h at time slot t; /(I) The total request number of k-type queues of the edge cloud h is served by the container f in the time slot t; /(I)Representing the total number of queue classes; /(I)Representing the number of requests that group g needs to forward to edge cloud h in time slot t with respect to container f; /(I)Representing the demand of the user group g for the container f at time slot t; /(I)Representing a set of user groups defined in terms of the location of the user's connection point to the network; /(I)The request representing container f on group g of edge cloud h arrives at the queue class; deltat gh represents the delay forwarding overhead from group g onto edge cloud h in time slots.
The admission acceptance of request information by the edge cloud is associated with ensuring that it gets served within its remaining deadline, the number of requests served from the k=1 queue class must always satisfy the arriving requests as well as the existing requests in the queue, expressed as:
The service requests of the k queue class must adhere to the queue class priority and the number of active pool instances of the container in the edge cloud and time, expressed as:
for the highest priority class k=1, the restriction on the request information is expressed as:
wherein, The total number of requests for the k-class queue of the service edge cloud h for container f at time slot t is represented.
S3: and processing the request information in the request queue by adopting an earliest deadline priority scheduling strategy to realize service resource supply allocation.
Further, the scheduling controller determines the order of processing the requests in the request queue. An earliest deadline first scheduling policy, a non-preemptive scheduling approach, is employed and each container instance processes only one request at a time.
While one container instance is processing a request, the container instance will be busy and not be able to service other requests. If there are no other container instances available, the dispatch controller will not be able to schedule and run another request for the same container until the container instance becomes available.
According to the earliest deadline first scheduling strategy, the admission controller only needs to simulate the scheduling after all existing requests join the request R for the decision of the incoming request R so as to calculate the completion time. For a request to meet its deadline, it must be scheduled before its remaining time equals its predicted processing time. Depending on the scheduling policy and the current load of the computation point, an arriving request R with a deadline D (where D. Gtoreq. The current time) that requires a container instance F may be rejected in two cases:
(1) Container instance F is either insufficient or non-existent, while CPU resources are available: the computation point is either that there are no container instances F currently or that there are not enough container instances F in its active pool to handle the request R.
(2) CPU resources are not enough: at least one container instance F is available before D at the computing point, but CPU resources will be allocated to other requests with earlier deadlines.
The present embodiment also provides a computing device comprising, a memory and a processor; the memory is configured to store computer executable instructions, and the processor is configured to execute the computer executable instructions to implement a method for implementing containerized service resource supply allocation for a cloud edge computing network system according to the foregoing embodiment.
The embodiment also provides a containerized service resource supply distribution system facing the cloud edge computing network system, which comprises a global control module, wherein the cloud edge computing network system receives user request information, constructs a global controller to monitor the latest states of all edge clouds, and selects the edge clouds for providing resources according to the user request information; the admission control module constructs an admission controller on the calculation point, and the admission controller judges the deadline of the request information and the calculated container instance, admits the request information and puts the request information into a request queue; and the scheduling control module processes the request information in the request queue by adopting an earliest deadline priority scheduling strategy to realize service resource supply allocation.
Further, the global controller monitors the latest states of all edge clouds, including the input queue size of the edge clouds, the state of the active pool instance, the resource utilization, the network bandwidth and delay, the service response time and the energy consumption; the computing point comprises a container activity pool for providing a plurality of containers with different services, when the request information reaches the computing point, the admission controller judges the deadline of the request information and the calculated container instance, and when the time used by the computing point for processing the request information is less than or equal to the deadline of the request information and the computing point has the container instance for processing the request information, the request information is admitted and put into a request queue; when the time for processing the request information by the computing point is longer than the deadline of the request information or the computing point does not have the container instance to process the request information, the computing point refuses the request information, and the refused request information is transferred to the next computing point along the path from the computing point to the back-end cloud, and is judged by an admission controller of the next computing point.
The container instance comprises the steps that the container instance is arranged on a CPU core, the container instance is started to occupy tasks in a period of time slot of the CPU core, and the number of cores occupied by loading a container image on an edge cloud is expressed as:
at the same time, each container in the active pool can only launch one container instance at a time, denoted as:
The containers within the time slot define a set of time-evolving over the active pool instance on the edge cloud, expressed as:
wherein, Representing the number of cores taken to load a container image; f represents a container; /(I)Representing a collection of containers; v denotes a container on the edge cloud; /(I)A collection of containers on edge cloud h; t represents a time slot; /(I)Representing a set of time slots; i represents a remaining time slot; /(I)A time slot overhead representing the launch of a container instance; /(I)A binary decision variable representing running a mirror image of container f as container instance v over time slot i; h represents an edge cloud; /(I)Representing an edge cloud set; /(I)Representing a set of container f instances on an edge cloud h within a time slot t; f' represents other containers than f;
The edge cloud processes the requests for pairs by scheduling the available instances of the container on the idle core, i.e. performing the association of the container to the core, the container being in a running state, the number of running instances being defined as:
the number of running instances is limited by the total number of cores of the edge cloud and the number of cores occupied when the mirror edge cloud is loaded, and the constraint is expressed as:
wherein, A binary variable representing the association of container v with the core at time slot t; /(I)The number of thermal instances on edge cloud h for container f for time slot t is represented.
For an edge cloud within a given time slot, the change in queue length of the request queue is expressed as,
For users with the same network connection point, the requests that the container needs to forward to the edge within the time slot t are expressed as:
The requests that arrive at the container of the queue class of the edge cloud at time slot t are expressed as:
wherein, Representing the number of requests for container f on a k-class queue on edge cloud h within time slot t; /(I)Representing a total number of requests for a k-class queue for container f to reach edge cloud h at time slot t; /(I)The total request number of k-type queues of the edge cloud h is served by the container f in the time slot t; /(I)Representing the total number of queue classes; /(I)Representing the number of requests that group g needs to forward to edge cloud h in time slot t with respect to container f; /(I)Representing the demand of the user group g for the container f at time slot t; /(I)Representing a set of user groups defined in terms of the location of the user's connection point to the network; /(I)The request representing container f on group g of edge cloud h arrives at the queue class; deltat gh represents the delay forwarding overhead from group g onto edge cloud h in time slots.
The admission controller comprises that admission acceptance of request information by the edge cloud is associated with ensuring that it gets served within its remaining deadline, the number of requests served from the k=1 queue class must always satisfy the arriving requests and the existing requests in the queue, expressed as:
The service requests of the k queue class must adhere to the queue class priority and the number of active pool instances of the container in the edge cloud and time, expressed as:
wherein, The total number of requests for the k-class queue of the service edge cloud h for container f at time slot t is represented.
The earliest deadline first scheduling policy includes marking the container instance as busy state when the container instance is processing the request, not accepting the request information, and if the computing point has no container instance in non-busy state, the scheduling controller puts the request information into the queuing queue until the container instance in non-busy state appears.
The present embodiment also provides a storage medium having stored thereon a computer program which, when executed by a processor, implements a containerized service resource supply allocation method for a cloud edge computing network system as set forth in the above embodiment.
The storage medium proposed in this embodiment belongs to the same inventive concept as the method for provisioning and distributing containerized service resources for a cloud computing network system proposed in the foregoing embodiment, and technical details not described in detail in this embodiment may be referred to the foregoing embodiment, and this embodiment has the same beneficial effects as the foregoing embodiment.
From the above description of embodiments, it will be clear to a person skilled in the art that the present invention may be implemented by means of software and necessary general purpose hardware, but of course also by means of hardware, although in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as a floppy disk, a read only Memory (ReadOnly, a Memory, a ROM), a random access Memory (RandomAccessMemory, RAM), a FLASH Memory (FLASH), a hard disk, or an optical disk of a computer, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method of the embodiments of the present invention.
Example 2
Referring to fig. 2-4, for one embodiment of the present invention, a method for allocating containerized service resources for cloud computing network system is provided, and in order to verify the beneficial effects of the present invention, scientific demonstration is performed through economic benefit calculation and simulation experiments.
The topology architecture of the cloud edge computing network system is abstracted into a tree structure, as shown in fig. 2, the root of the tree is a back-end cloud formed by a cloud data center server, leaf nodes are gateway nodes which are accessed into the system by a terminal and have no computing power, and internal nodes of the tree are computing points positioned at the edge of the network.
The terminal request of the cloud edge computing network infrastructure satisfies the path structure as shown in fig. 3. The computing request provided by the terminal is processed at the computing point on the path from the terminal, the edge node to the back-end cloud server, and the highest service quality is provided for the terminal user.
The terminal indicates in the service request the function to be performed, inputs data and defines the deadline for completion. The cloud computing network system then supplies the preconfigured execution container resources to process the incoming request. The container resource provisioning decision adjusts the amount of resources allocated for each container at each computing point based on changes in cloud computing network system state, including changes in functional requirements, success of requests when meeting deadline requirements, and the like.
As shown in fig. 4, at one point of computation, since the only container instance capable of handling request F 1 is currently busy, occupying the core 1 of the CPU running, request F 1 at the beginning of the queue will be skipped and request F 2 will be dispatched instead. Request F 1 will remain at the beginning of the queue and request F 1 will be scheduled when the instance of F 1 container running in core 1 completes the current request. In this way, requests waiting in the queue can be scheduled immediately.
Simulation experiments are carried out through a cloud edge computing network system provided with a plurality of edge cloud nodes, and each node is provided with different numbers of CPU cores and container instances. The experimental aim is to compare the difference in processing efficiency and response time between the traditional resource allocation method and the method proposed by the invention.
In the experimental process, a global controller is firstly constructed and is responsible for monitoring and managing the resource allocation of the whole network. The global controller collects and analyzes state information of all edge clouds in real time, including input queue size, active pool instance state, resource usage, network bandwidth and delay, service response time, and energy consumption.
Next, a series of user requests with different deadlines and resource requirements are simulated. And the global controller adopts an edge cloud priority strategy according to the request information, and selects the most suitable edge cloud to perform resource allocation. When the edge cloud cannot process more requests due to overload, the global controller can schedule part of tasks to the back-end cloud.
At the computation point, an admission controller is constructed to receive and place request information into a request queue. The admission controller judges according to the deadline of the request and the container instance condition of the calculation point, and ensures that the request is processed within the deadline.
Finally, the scheduling controller decides the processing sequence of the requests in the request queue by adopting the earliest deadline priority scheduling strategy. Each container instance only processed one request at a time to ensure high efficiency and accuracy, experimental data is shown in table 1.
Table 1 comparative test data table
It can be seen that the process of the present invention provides a significant improvement in both processing efficiency and response time over conventional processes. With the same total number of requests, the number of successful processing requests of the inventive method increases from 857 to 943, indicating higher processing power and lower request rejection rate. Furthermore, the average response time decreases from 2.47 seconds to 1.15 seconds, highlighting the advantages of the edge cloud priority strategy and the earliest deadline priority scheduling strategy in terms of reduced latency.
The increase in CPU utilization increased from 72.3% to 86.7%, indicating a more efficient utilization of resources. Meanwhile, the reduction of energy consumption (from 505.5 units to 448.2 units) shows that the method can reduce the energy cost while improving the efficiency, which is important for constructing a sustainable cloud edge computing network system.
From these data, the innovations and advantages of the present invention over the prior art in the examples can be clearly seen. The invention not only improves the processing efficiency of the request and the response speed of the system, but also optimizes the resource utilization rate and reduces the energy consumption through the intelligent global controller and the efficient resource scheduling strategy. These improvements are particularly important for real-time data processing and application in high-load environments, showing the innovation and practicality of the present invention in cloud computing network system design.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein. The scheme in the embodiment of the application can be realized by adopting various computer languages, such as object-oriented programming language Java, an transliteration script language JavaScript and the like.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present application have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the application.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present application without departing from the spirit or scope of the application. Thus, it is intended that the present application also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
It should be noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention may be modified or substituted without departing from the spirit and scope of the technical solution of the present invention, which is intended to be covered in the scope of the claims of the present invention.
Claims (10)
1. The method for supplying and distributing the containerized service resources for the cloud edge computing network system is characterized by comprising the following steps of:
The cloud edge computing network system receives user request information, constructs a global controller and selects an edge cloud for providing resources according to the user request information;
An admission controller is constructed on the computing point of the edge cloud, and the admission request information is admitted and put into a request queue;
And processing the request information in the request queue by adopting an earliest deadline priority scheduling strategy to realize service resource supply allocation.
2. The method for distributing the containerized service resource supply of the cloud computing network system according to claim 1, wherein the method comprises the following steps: the global controller monitors the latest states of all edge clouds, including the input queue size of the edge clouds, the state of an active pool instance, the resource utilization rate, the network bandwidth and delay, the service response time and the energy consumption;
The computing point comprises a container activity pool for providing a plurality of containers with different services, when the request information reaches the computing point, the admission controller judges the deadline of the request information and the calculated container instance, and when the time used by the computing point for processing the request information is less than or equal to the deadline of the request information and the computing point has the container instance for processing the request information, the request information is admitted and put into a request queue;
When the time for processing the request information by the computing point is longer than the deadline of the request information or the computing point does not have the container instance to process the request information, the computing point refuses the request information, and the refused request information is transferred to the next computing point along the path from the computing point to the back-end cloud, and is judged by an admission controller of the next computing point.
3. The method for distributing the containerized service resource supply of the cloud computing network system according to claim 2, wherein the method comprises the following steps: the container instance comprises the steps that the container instance is arranged on a CPU core, the container instance is started to occupy tasks in a period of time slot of the CPU core, and the number of cores occupied by loading a container image on an edge cloud is expressed as:
at the same time, each container in the active pool can only launch one container instance at a time, denoted as:
The containers within the time slot define a set of time-evolving over the active pool instance on the edge cloud, expressed as:
wherein, Representing the number of cores taken to load a container image; f represents a container; /(I)Representing a collection of containers; v represents a container on the edge cloud; /(I)A collection of containers on edge cloud h; t represents a time slot; /(I)Representing a set of time slots; i represents a remaining time slot; /(I)A time slot overhead representing the launch of a container instance; /(I)A binary decision variable representing running a mirror image of container f as container instance v over time slot i; h represents an edge cloud; /(I)Representing an edge cloud set; /(I)Representing a set of container f instances on an edge cloud h within a time slot t; f' represents other containers than f;
The edge cloud processes the requests for pairs by scheduling the available instances of the container on the idle core, i.e. performing the association of the container to the core, the container being in a running state, the number of running instances being defined as:
the number of running instances is limited by the total number of cores of the edge cloud and the number of cores occupied when the mirror edge cloud is loaded, and the constraint is expressed as:
wherein, A binary variable representing the association of container v with the core at time slot t; /(I)The number of thermal instances on edge cloud h for container f for time slot t is represented.
4. The method for distributing the containerized service resource supply of the cloud computing network system according to claim 3, wherein the method comprises the following steps of: for an edge cloud within a given time slot, the change in queue length of the request queue is expressed as,
For users with the same network connection point, the requests that the container needs to forward to the edge within the time slot t are expressed as:
The requests that arrive at the container of the queue class of the edge cloud at time slot t are expressed as:
wherein, Representing the number of requests for container f on a k-class queue on edge cloud h within time slot t; /(I)Representing a total number of requests for a k-class queue for container f to reach edge cloud h at time slot t; /(I)The total request number of k-type queues of the edge cloud h is served by the container f in the time slot t; /(I)Representing the total number of queue classes; /(I)Representing the number of requests that group g needs to forward to edge cloud h in time slot t with respect to container f; /(I)Representing the demand of the user group g for the container f at time slot t; /(I)Representing a set of user groups defined in terms of the location of the user's connection point to the network; /(I)The request representing container f on group g of edge cloud h arrives at the queue class; deltat gh represents the delay forwarding overhead from group g onto edge cloud h in time slots.
5. The method for distributing the containerized service resource supply of the cloud computing network system according to claim 4, wherein the method comprises the following steps: the admission controller comprises that admission acceptance of request information by the edge cloud is associated with ensuring that it gets served within its remaining deadline, the number of requests served from the k=1 queue class must always satisfy the arriving requests and the existing requests in the queue, expressed as:
The service requests of the k queue class must adhere to the queue class priority and the number of active pool instances of the container in the edge cloud and time, expressed as:
wherein, The total number of requests for the k-class queue of the service edge cloud h for container f at time slot t is represented.
6. The method for distributing the containerized service resource supply of the cloud computing network system according to claim 5, wherein the method comprises the following steps: the earliest deadline first scheduling policy includes marking the container instance as busy state when the container instance is processing the request, not accepting the request information, and if the computing point has no container instance in non-busy state, the scheduling controller puts the request information into the queuing queue until the container instance in non-busy state appears.
7. A containerized service resource supply distribution system for cloud computing network system is characterized by comprising,
The cloud edge computing network system receives user request information, constructs a global controller to monitor the latest states of all edge clouds, and selects the edge clouds providing resources according to the user request information;
the admission control module is used for constructing an admission controller on a computing point of the edge cloud, judging the deadline of the request information and the computed container instance by the admission controller, admitting the request information and putting the request information into a request queue;
and the scheduling control module processes the request information in the request queue by adopting an earliest deadline priority scheduling strategy to realize service resource supply allocation.
8. The cloud computing network system-oriented containerized service resource supply distribution system of claim 7, wherein: the global controller monitors the latest states of all edge clouds, including the input queue size of the edge clouds, the state of an active pool instance, the resource utilization rate, the network bandwidth and delay, the service response time and the energy consumption;
The computing point comprises a container activity pool for providing a plurality of containers with different services, when the request information reaches the computing point, the admission controller judges the deadline of the request information and the calculated container instance, and when the time used by the computing point for processing the request information is less than or equal to the deadline of the request information and the computing point has the container instance for processing the request information, the request information is admitted and put into a request queue;
When the time for processing the request information by the computing point is longer than the deadline of the request information or the computing point does not have the container instance to process the request information, the computing point refuses the request information, and the refused request information is transferred to the next computing point along the path from the computing point to the back-end cloud, and is judged by an admission controller of the next computing point.
9. A computing device, comprising: a memory and a processor;
The memory is for storing computer executable instructions, the processor being for executing the computer executable instructions which when executed by the processor implement the steps of the method of any one of claims 1 to 7.
10. A computer readable storage medium storing computer executable instructions which when executed by a processor implement the steps of the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410162543.7A CN117938855A (en) | 2024-02-05 | 2024-02-05 | Method and system for supplying and distributing containerized service resources for cloud edge computing network system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410162543.7A CN117938855A (en) | 2024-02-05 | 2024-02-05 | Method and system for supplying and distributing containerized service resources for cloud edge computing network system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117938855A true CN117938855A (en) | 2024-04-26 |
Family
ID=90755778
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410162543.7A Pending CN117938855A (en) | 2024-02-05 | 2024-02-05 | Method and system for supplying and distributing containerized service resources for cloud edge computing network system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117938855A (en) |
-
2024
- 2024-02-05 CN CN202410162543.7A patent/CN117938855A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ge et al. | GA-based task scheduler for the cloud computing systems | |
| US7823154B2 (en) | System and method for providing, by a plurality of schedulers, differentiated service to consumers of distributed resources | |
| Patel et al. | Priority based job scheduling techniques in cloud computing: a systematic review | |
| CN112995023A (en) | Multi-access edge computing network computing unloading system and computing unloading method thereof | |
| CN109542608B (en) | Cloud simulation task scheduling method based on hybrid queuing network | |
| CN107579926A (en) | The QoS methods to set up of Ceph cloud storage systems based on token bucket algorithm | |
| JP3306705B2 (en) | Packet transfer control device and scheduling method thereof | |
| CN107370799B (en) | A kind of online computation migration method of multi-user mixing high energy efficiency in mobile cloud environment | |
| CN110069341A (en) | What binding function configured on demand has the dispatching method of dependence task in edge calculations | |
| CN112506634B (en) | Fairness operation scheduling method based on reservation mechanism | |
| WO2020134133A1 (en) | Resource allocation method, substation, and computer-readable storage medium | |
| CN112799823A (en) | Online dispatching and scheduling method and system for edge computing tasks | |
| CN113190342B (en) | Method and system architecture for multi-application fine-grained offloading of cloud-edge collaborative networks | |
| Li et al. | Co-Scheduler: A coflow-aware data-parallel job scheduler in hybrid electrical/optical datacenter networks | |
| Dong et al. | TINA: A fair inter-datacenter transmission mechanism with deadline guarantee | |
| CN109298932B (en) | OpenFlow-based resource scheduling method, scheduler and system | |
| CN110677352A (en) | Single-switch single-controller transmission control method in software defined network | |
| Dong et al. | Task-aware flow scheduling with heterogeneous utility characteristics for data center networks | |
| CN117938855A (en) | Method and system for supplying and distributing containerized service resources for cloud edge computing network system | |
| CN117056064A (en) | Resource allocation method, device, server, storage medium and program product | |
| Hasnain et al. | Coflow scheduling with performance guarantees for data center applications | |
| Wu et al. | Weakly hard real-time scheduling algorithm for multimedia embedded system on multiprocessor platform | |
| Guan et al. | Maximize potential reserved task scheduling for URLLC transmission and edge computing | |
| Huang et al. | A dynamic and complexity aware cloud scheduling algorithm for video transcoding | |
| Jia | A dynamic scheduling framework for multi-tenancy clouds |
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 |