CN109343947A - A kind of resource regulating method and device - Google Patents
A kind of resource regulating method and device Download PDFInfo
- Publication number
- CN109343947A CN109343947A CN201811124755.7A CN201811124755A CN109343947A CN 109343947 A CN109343947 A CN 109343947A CN 201811124755 A CN201811124755 A CN 201811124755A CN 109343947 A CN109343947 A CN 109343947A
- Authority
- CN
- China
- Prior art keywords
- server
- grouping
- container
- dispatching requirement
- kubernetes system
- 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
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
- G06F9/48—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
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
Disclosed herein is a kind of resource regulating method and devices.The resource regulating method includes: the dispatching requirement that container is obtained in Kubernetes system, is grouped according to the dispatching requirement of the container server that selection meets the dispatching requirement from the grouping of multiple servers；Container is scheduled according to the server grouping selected in Kubernetes system.The technical solution of this paper can increase the flexibility of scheduling of resource, to meet the dispatching requirement of different business scene.
Description
Technical field
The present invention relates to field of computer technology more particularly to a kind of resource regulating methods and device.
Background technique
Along with the development of artificial intelligence technology, deep learning is gradually risen.The realization of deep learning, needs multiple technologies
It is supported, such as server, GPU (Graphics Processing Unit, graphics processing unit), cluster, cluster management
Dispatcher software, deep learning frame, concrete application of deep learning etc..
Kubernetes has complete cluster management ability, multi-level security protection and mechanism of permitting the entrance, multi-tenant application
The resource quota managerial ability of enabling capabilities, powerful fault discovery and self-repairing capability and more granularities, thus it is extensive
It applies in artificial intelligence field as the scheduling of deep learning and monitoring support platform on ground.
Current Kubernetes uses resources left scheduling strategy, this scheduling plan for the scheduling strategy of resource
New container is preferentially slightly dispatched to the high server of surplus to run.But the mode of artificial intelligence field deep learning is more
Kind multiplicity, unified resources left scheduling strategy are not able to satisfy the training mission of deep learning under certain business scenarios
It is required that.
Summary of the invention
The embodiment of the present invention can increase money the technical problem to be solved is that a kind of resource regulating method and device is provided
The flexibility of source scheduling, to meet the dispatching requirement of different business scene.
The embodiment of the present invention provides a kind of resource regulating method, comprising:
The dispatching requirement that container is obtained in Kubernetes system, according to the dispatching requirement of the container from multiple services
Selection meets the server grouping of the dispatching requirement in device grouping；
Container is scheduled according to the server grouping selected in Kubernetes system.
The embodiment of the present invention provides a kind of resource scheduling device, comprising:
Matching module, for obtaining the dispatching requirement of container in Kubernetes system, according to the scheduling of the container
Demand selection from the grouping of multiple servers meets the server grouping of the dispatching requirement；
Scheduler module, for being scheduled according to the server grouping selected to container in Kubernetes system.
The embodiment of the present invention provides a kind of resource scheduling device, comprising:
Memory, processor and it is stored in the scheduling of resource journey that can be run on the memory and on the processor
The step of sequence, the resource scheduler realizes above-mentioned resource regulating method when being executed by the processor.
Compared with the relevant technologies, the embodiment of the present invention provides a kind of resource regulating method and device, in Kubernetes system
The dispatching requirement that container is obtained in system, according to the dispatching requirement of the container, selection meets the tune from the grouping of multiple servers
The server of degree demand is grouped, and is scheduled according to the server grouping selected to container in Kubernetes system.This hair
The technical solution of bright embodiment can increase the flexibility of scheduling of resource, to meet the dispatching requirement of different business scene.
Detailed description of the invention
Fig. 1 is a kind of resource regulating method flow chart of the embodiment of the present invention 1；
Fig. 2 is a kind of resource scheduling device schematic diagram of the embodiment of the present invention 2；
Fig. 3 is the schematic diagram that a kind of GPU resource of example 1 of the present invention is grouped；
Fig. 4 is a kind of resource regulating method flow chart of example 1 of the present invention.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention clearer, below in conjunction with attached drawing to the present invention
Embodiment be described in detail.It should be noted that in the absence of conflict, in the embodiment and embodiment in the application
Feature can mutual any combination.
Embodiment 1
As shown in Figure 1, the embodiment of the invention provides a kind of resource regulating methods, comprising:
Step S110 obtains the dispatching requirement of container, according to the dispatching requirement of the container in Kubernetes system
Selection meets the server grouping of the dispatching requirement from the grouping of multiple servers；
Step S120 is scheduled container according to the server grouping selected in Kubernetes system；
In one embodiment, the method also includes:
Server is grouped according to the grouping condition of setting in Kubernetes system, generates server grouping letter
Breath；
In one embodiment, the grouping condition of the setting comprises at least one of the following: the model of physical resource, clothes
The purposes and business rule of business device；
Wherein, physical resource comprises at least one of the following: CPU (Central Processing Unit, central processing
Device), GPU (Graphics Processing Unit, graphics processing unit) and memory；
In one embodiment, the generation server grouping information, comprising:
It is server grouping setting packet label in Kubernetes system；
In one embodiment, the dispatching requirement that container is obtained in Kubernetes system, according to the container
Dispatching requirement selection from the grouping of multiple servers meets the server grouping of the dispatching requirement, comprising:
Filter condition is set according to the dispatching requirement of container in Kubernetes system；
Server packet label is screened using the filter condition of the setting, selection meets the dispatching requirement
Server grouping；
In one embodiment, described that container is carried out according to the server grouping selected in Kubernetes system
Scheduling, comprising:
The surplus yield of each server in the server grouping selected in Kubernetes system is compared；
Container is dispatched in the server grouping on the maximum server of surplus yield.
In one embodiment, the server includes: PC cluster node server；
Above-described embodiment by disposing Kubernetes system on the server, according to setting in Kubernetes system
Grouping condition server is grouped, for server grouping setting packet label, according to appearance in Kubernetes system
The dispatching requirement of device sets filter condition, is screened using the filter condition of the setting to server packet label, selects
The server packet label for meeting the dispatching requirement determines that server is grouped according to the server packet label selected, will hold
Device is dispatched on the server in the server grouping.It is above-mentioned be first grouped dispatch again by way of, resource tune can be increased
The flexibility of degree, to meet the dispatching requirement of different business scene.
Embodiment 2
As shown in Fig. 2, the embodiment of the invention provides a kind of resource scheduling devices, comprising:
Matching module 201, for obtaining the dispatching requirement of container in Kubernetes system, according to the tune of the container
Degree demand selects the server grouping for meeting the dispatching requirement from server grouping information；
Scheduler module 202, for being scheduled according to the server grouping selected to container in Kubernetes system.
In one embodiment, the resource scheduling device further includes grouping module 203；
Grouping module 203 is grouped server for the grouping condition in Kubernetes system according to setting,
Generate server grouping information.
In one embodiment, grouping module 203, for generating server grouping information in the following ways:
It is server grouping setting packet label in Kubernetes system.
In one embodiment, the grouping condition of the setting comprises at least one of the following: the model of physical resource, clothes
The purposes and business rule of business device；
Wherein, physical resource comprises at least one of the following: CPU (Central Processing Unit, central processing
Device), GPU (Graphics Processing Unit, graphics processing unit) and memory；
In one embodiment, matching module 201 obtain appearance for sampling following manner in Kubernetes system
The dispatching requirement of device, according to the dispatching requirement of the container, selection meets the clothes of the dispatching requirement from the grouping of multiple servers
Business device grouping:
Filter condition is set according to the dispatching requirement of container in Kubernetes system；Utilize the filtering rod of the setting
Part screens server packet label, and selection meets the server grouping of the dispatching requirement；
In one embodiment, scheduler module 202, in the following ways in Kubernetes system according to choosing
Server grouping out is scheduled container:
The surplus yield of each server in the server grouping selected in Kubernetes system is compared；
Container is dispatched in the server grouping on the maximum server of surplus yield.
In one embodiment, the server includes: PC cluster node server.
Embodiment 3
The embodiment of the invention provides a kind of resource scheduling devices, comprising:
Memory, processor and it is stored in the scheduling of resource journey that can be run on the memory and on the processor
The step of sequence, the resource scheduler realizes above-mentioned resource regulating method when being executed by the processor.
Wherein, memory can be various by any kind of volatibility or non-volatile memory device or their group
It closes and realizes, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM) is erasable to compile
Journey read-only memory (EPROM), programmable read only memory (PROM), read-only memory (ROM), magnetic memory, flash
Device, disk or CD.
Processor can be central processing unit (CPU) either Field Programmable Logic Array (FPGA) or single-chip microcontroller
(MCU) either digital signal processor (DSP) or specific integrated circuit (ASIC) etc. have data-handling capacity and/or program
The logical operation device of executive capability.
Below by illustrating a kind of resource regulating method of the application.
Example 1
This example is directed to server cluster system, proposes a kind of resource regulating method, can increase the flexible of scheduling of resource
Property, to meet the dispatching requirement of artificial intelligence field deep learning.
As shown in figure 3, in this example, a kind of training business of deep learning has particular/special requirement to GPU performance, therefore,
Kubernetes system is in advance grouped the GPU resource on server according to performance models, for example is divided into GPU model
The second packet of the first grouping and GPU model P40 of P100.Kubernetes system is the service with different model GPU
Corresponding packet label, such as " GPU-P100 " and " GPU-P40 " is respectively set in device.
Filter condition is set according to the dispatching requirement of container in Kubernetes system, utilizes the filtering rod of the setting
Part screens server packet label, and selection meets the server packet label of the dispatching requirement, according to the clothes selected
Business device packet label determines that server is grouped, and container is dispatched on the server in the server grouping.
As shown in figure 4, a kind of resource regulating method, may comprise steps of:
Step S101 is grouped server according to GPU performance models in Kubernetes system, is different clothes
Corresponding packet label is respectively set in business device；
Step S102 obtains user to the dispatching requirement of container in Kubernetes system, will meet user demand
Server packet label is as filter condition；
Step S103 screens server packet label using the filter condition of the setting, described in selection satisfaction
The server of dispatching requirement is grouped；
Step S104, to the surplus yield of each server in the server grouping selected in Kubernetes system
It is compared；
Container is dispatched in the server grouping on the maximum server of surplus yield by step S105.
It should be noted that the invention may also have other embodiments, without departing substantially from spirit of that invention and its essence
In the case of, those skilled in the art can make various corresponding changes and modifications according to the present invention, but these are corresponding
Change and modification all should fall within the scope of protection of the appended claims of the present invention.
Claims (10)
1. a kind of resource regulating method, comprising:
The dispatching requirement that container is obtained in Kubernetes system is divided according to the dispatching requirement of the container from multiple servers
Selection meets the server grouping of the dispatching requirement in group；
Container is scheduled according to the server grouping selected in Kubernetes system.
2. the method as described in claim 1, which is characterized in that the method also includes:
Server is grouped according to the grouping condition of setting in Kubernetes system, generates server grouping information.
3. method according to claim 2, it is characterised in that:
The grouping condition of the setting comprises at least one of the following: the model of physical resource, the purposes of server and business rule.
4. method according to claim 2, it is characterised in that:
The generation server grouping information, comprising: packet label is set for server grouping in Kubernetes system.
5. method as claimed in claim 4, it is characterised in that:
The dispatching requirement that container is obtained in Kubernetes system, according to the dispatching requirement of the container from multiple services
Selection meets the server grouping of the dispatching requirement in device grouping, comprising:
Filter condition is set according to the dispatching requirement of container in Kubernetes system；
Server packet label is screened using the filter condition of the setting, selection meets the service of the dispatching requirement
Device grouping.
6. the method as described in claim 1, it is characterised in that:
It is described that container is scheduled according to the server grouping selected in Kubernetes system, comprising:
The surplus yield of each server in the server grouping selected in Kubernetes system is compared；
Container is dispatched in the server grouping on the maximum server of surplus yield.
7. a kind of resource scheduling device, comprising:
Matching module, for obtaining the dispatching requirement of container in Kubernetes system, according to the dispatching requirement of the container
Selection meets the server grouping of the dispatching requirement from the grouping of multiple servers；
Scheduler module, for being scheduled according to the server grouping selected to container in Kubernetes system.
8. resource scheduling device as claimed in claim 7, which is characterized in that the resource scheduling device further include:
Grouping module is grouped server for the grouping condition in Kubernetes system according to setting, generates clothes
Business device grouping information.
9. resource scheduling device as claimed in claim 8, it is characterised in that:
Grouping module, for generating server grouping information in the following ways: being server point in Kubernetes system
Group setting packet label.
10. a kind of resource scheduling device, comprising:
Memory, processor and it is stored in the resource scheduler that can be run on the memory and on the processor, institute
It states and realizes resource regulating method described in any one of the claims 1-6 when resource scheduler is executed by the processor
The step of.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811124755.7A CN109343947A (en) | 2018-09-26 | 2018-09-26 | A kind of resource regulating method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811124755.7A CN109343947A (en) | 2018-09-26 | 2018-09-26 | A kind of resource regulating method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109343947A true CN109343947A (en) | 2019-02-15 |
Family
ID=65307014
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811124755.7A Pending CN109343947A (en) | 2018-09-26 | 2018-09-26 | A kind of resource regulating method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109343947A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110333939A (en) * | 2019-06-17 | 2019-10-15 | 腾讯科技(成都)有限公司 | Task mixed scheduling method, device, dispatch server and Resource Server |
| CN112052133A (en) * | 2019-06-06 | 2020-12-08 | 北京京东尚科信息技术有限公司 | Service system monitoring method and device based on Kubernetes |
| WO2021093783A1 (en) * | 2019-11-11 | 2021-05-20 | 星环信息科技(上海)股份有限公司 | Real-time resource scheduling method and apparatus, computer device, and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170249186A1 (en) * | 2013-08-26 | 2017-08-31 | Vmware, Inc. | Cpu scheduler configured to support latency sensitive virtual machines |
| CN108062246A (en) * | 2018-01-25 | 2018-05-22 | 北京百度网讯科技有限公司 | For the resource regulating method and device of deep learning frame |
| CN108228354A (en) * | 2017-12-29 | 2018-06-29 | 杭州朗和科技有限公司 | Dispatching method, system, computer equipment and medium |
| CN108509256A (en) * | 2017-02-28 | 2018-09-07 | 华为技术有限公司 | Method, equipment and the running equipment of management and running equipment |
-
2018
- 2018-09-26 CN CN201811124755.7A patent/CN109343947A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170249186A1 (en) * | 2013-08-26 | 2017-08-31 | Vmware, Inc. | Cpu scheduler configured to support latency sensitive virtual machines |
| CN108509256A (en) * | 2017-02-28 | 2018-09-07 | 华为技术有限公司 | Method, equipment and the running equipment of management and running equipment |
| CN108228354A (en) * | 2017-12-29 | 2018-06-29 | 杭州朗和科技有限公司 | Dispatching method, system, computer equipment and medium |
| CN108062246A (en) * | 2018-01-25 | 2018-05-22 | 北京百度网讯科技有限公司 | For the resource regulating method and device of deep learning frame |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112052133A (en) * | 2019-06-06 | 2020-12-08 | 北京京东尚科信息技术有限公司 | Service system monitoring method and device based on Kubernetes |
| CN112052133B (en) * | 2019-06-06 | 2024-09-20 | 北京京东尚科信息技术有限公司 | Method and device for monitoring service system based on Kubernetes |
| CN110333939A (en) * | 2019-06-17 | 2019-10-15 | 腾讯科技(成都)有限公司 | Task mixed scheduling method, device, dispatch server and Resource Server |
| CN110333939B (en) * | 2019-06-17 | 2023-11-14 | 腾讯科技(成都)有限公司 | Task mixed scheduling method and device, scheduling server and resource server |
| WO2021093783A1 (en) * | 2019-11-11 | 2021-05-20 | 星环信息科技(上海)股份有限公司 | Real-time resource scheduling method and apparatus, computer device, and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109376009A (en) | A kind of method and device of shared resource | |
| CN111414233A (en) | Online model reasoning system | |
| CN110704186A (en) | Computing resource allocation method and device based on hybrid distribution architecture and storage medium | |
| US10104187B2 (en) | System, computer program, and method for dividing services into subsets based on interdependencies | |
| US10445209B2 (en) | Prescriptive analytics based activation timetable stack for cloud computing resource scheduling | |
| US20100287283A1 (en) | Optimized Multi-Component Co-Allocation Scheduling With Advanced Reservations for Data Transfers and Distributed Jobs | |
| CN109343947A (en) | A kind of resource regulating method and device | |
| CN110213780A (en) | Management method, management and the layout entity and storage medium of network slice | |
| CA2859500A1 (en) | Cloud-edge topologies | |
| WO2019055871A1 (en) | Systems and methods for a policy-driven orchestration of deployment of distributed applications | |
| Pascual et al. | Run-time adaptation of mobile applications using genetic algorithms | |
| CN111427677B (en) | Artificial intelligence product generation method, device and server | |
| CN114818446B (en) | Power service decomposition method and system facing 5G cloud edge terminal cooperation | |
| US12026536B2 (en) | Rightsizing virtual machine deployments in a cloud computing environment | |
| WO2022267724A1 (en) | Cognitive scheduler for kubernetes | |
| CN113553140B (en) | Resource scheduling method, equipment and system | |
| CN110162344A (en) | A kind of method, apparatus, computer equipment and readable storage medium storing program for executing that current limliting is isolated | |
| CN109542593A (en) | A kind of flow chart of data processing design method based on NIFI | |
| CN117931454A (en) | Computing power resource scheduling method, computing power resource scheduling device, computing power resource scheduling equipment, storage medium and program product | |
| CN115480785A (en) | Container-based service deployment method and device and server | |
| CN117056018A (en) | Resource scheduling method, apparatus, device, program product and storage medium | |
| CN111831362A (en) | Method for automatic discovery, classification, deployment and management of K8s environment monitoring | |
| CN109271236A (en) | A kind of method, apparatus of traffic scheduling, computer storage medium and terminal | |
| Marin et al. | Reaching for the clouds: contextually enhancing smartphones for energy efficiency | |
| CN115665157B (en) | Balanced scheduling method and system based on application resource types |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190215 |
|
| RJ01 | Rejection of invention patent application after publication |