CN109375992A - A kind of resource regulating method and device - Google Patents
A kind of resource regulating method and device Download PDFInfo
- Publication number
- CN109375992A CN109375992A CN201810943505.XA CN201810943505A CN109375992A CN 109375992 A CN109375992 A CN 109375992A CN 201810943505 A CN201810943505 A CN 201810943505A CN 109375992 A CN109375992 A CN 109375992A
- Authority
- CN
- China
- Prior art keywords
- node
- task
- resource
- task requests
- tag
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
- G06F9/5038—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering the execution order of a plurality of tasks, e.g. taking priority or time dependency constraints into consideration
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mobile Radio Communication Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
A kind of resource regulating method and device, wherein method specifically includes resource scheduling device and receives the task requests that client is sent, the task requests include the corresponding resource tag of the task requests, which passes through the specified resource for executing required by task of resource tag that the task requests include;The resource tag that resource scheduling device has according to node each in node group determines destination node, the resource tag which there are the task requests to include from the node group;Resource scheduling device indicates that the destination node executes the specified task of the task requests.On the one hand, the resource for the required by task specified due to the corresponding resource tag of task requests, it is possible to determine that processing capacity is suitble to the destination node of the task, to help to improve the execution efficiency of the specified task of the task node.On the other hand, the destination node that the resource tag for including with the task requests is determined from node group, can help improve the resource utilization of node.
Description
Technical field
This application involves field of cloud computer technology more particularly to a kind of resource regulating methods and device.
Background technique
Under video cloud computing scene, such as traffic video, coding and decoding video is carried out, face characteristic extracts, license plate
Identification and the processing such as congestion event monitoring, need to realize node by task schedule to be treated into cloud server cluster into
Row processing.Another resource coordination person (Yet Another Resource Negotiator, YARN) is that Hadoop 2.X is general
Resource management system, be responsible for cloud server cluster in node resource management and task schedule.
In the prior art based on the resource scheduling scheme of YARN, task is submitted to some resource queue, the resource queue
Division be logical group division, that is to say, that the task can be dispatched in logical group corresponding with the resource queue
Node on run.Be divided into 3 resource queues, respectively resource queue 1,
Resource queue 2 and resource queue 3,1 counterlogic cluster A of resource queue, 2 counterlogic cluster B of resource queue, resource queue 3
Counterlogic cluster C, if a task 1 is committed to resource queue 2, which will be scheduled to logical group B
In node.Node binding in resource scheduling scheme in the prior art, in resource queue and logical group, if it is determined that go out
Resource queue can only just determine egress in the corresponding logical group of resource queue.As it can be seen that scheduling of resource in the prior art
Scheme causes task run inefficient, and cluster it is possible that task is scheduled on the very weak node of processing capacity
In node resource utilization it is not high.
To sum up, the execution efficiency of raising task and the resource utilization of node how to be realized, still needs to further further investigate.
Summary of the invention
The application provides a kind of resource regulating method and device, for improving the execution efficiency of task and the resource benefit of node
With rate.
In a first aspect, the application provides a kind of resource regulating method, this method comprises: resource scheduling device receives client
The task requests of transmission, the task requests include the corresponding resource tag of the task requests, and the task requests pass through institute
State the specified resource for executing required by task of resource tag that task requests include;Resource scheduling device is according to section each in node group
The resource tag that point has determines that destination node, the destination node include with the task requests from the node group
Resource tag;Resource scheduling device indicates that the destination node executes the specified task of the task requests.
Based on the program, resource scheduling device receives the task requests that client is sent, and resource scheduling device is according to node
The resource tag that the resource tag and task requests that each node has in group include determines target section from the node group
Point, the resource tag that there are the destination node task requests to include;Resource scheduling device indicates that the destination node is held
The specified task of the row task requests.On the one hand, since the task requests include the corresponding resource mark of the task requests
Label, the task requests specify the resource for executing required by task by the resource tag that the task requests include, so according to
The resource of the specified required by task of task requests determines destination node, can find processing capacity and is suitble to the task requests specified
Task node, to help to improve the execution efficiency of the specified task of the task node.On the other hand, execution is being determined
When the destination node of the specified task of the task requests, the resource tag being had according to node determines have from node group
The resource tag that the task requests include in this way, the resource of node group interior joint rationally can be utilized, and then helps to improve section
The resource utilization of point.
When receiving task requests in resource scheduling device, and determining destination node from node group, since some tasks exist
Required resource includes graphics processor (graphics processing unit, GPU) when execution, some tasks are when being executed
Required resource does not include GPU, therefore, if led executing in the task schedule to the node with GPU of no GPU demand
Being badly in need of GPU resource of the task is caused to can not find the suitable node with GPU resource, to cause the GPU resource on the node
Waste.Therefore, in order to avoid the waste of GPU resource in node group, the above method, several optional realization sides presented below are based on
Method.
Implementation method one, resource scheduling device can according to the corresponding resource tag of task requests whether include GPU money
Source label, selection execute the destination node of the specified task of the task requests.
As a kind of specific example, when the corresponding resource tag of the task requests does not include the resource tag of GPU, from
The node of the resource tag without GPU is selected in node group;Indicate that the node of selection executes the specified task of the task requests.
As another specific example, when the corresponding resource tag of the task requests includes the resource tag of GPU, from
The node of resource tag of the selection with GPU in node group;Indicate that the node of selection executes the specified task of the task requests.
By the implementation method, according to the corresponding resource tag of task requests whether include GPU label, selection executes institute
The resource for the required by task specified according to task requests may be implemented, on demand in the destination node for stating the specified task of task requests
GPU resource scheduling is carried out, so as to promote the resource utilization of node group interior joint, especially promotes the utilization of GPU resource
Rate helps avoid being depleted and leading because of central processing unit (central processing unit, CPU) resource in node
Cause the GPU resource waste more than needed in the node.
Implementation method two, task requests can also include the stock number for executing required by task, and resource scheduling device can root
It whether include GPU resource amount according to the stock number of the execution required by task, selection executes the target of the specified task of the task requests
Node.
As a kind of specific example, if the stock number for executing required by task does not include GPU resource, from the section
The node without the GPU resource is selected in point group;Indicate that the node of selection executes the specified task of the task requests.
As another specific example, if the stock number for executing required by task does not include GPU resource, from described
The node without the GPU resource is selected in node group;Indicate that the node of selection executes the specified task of the task requests.
Further, as a kind of possible implementation, task requests record has the stock number for executing required by task;Money
The resource tag that source dispatching device has according to node each in node group determines destination node from node group, specifically includes:
Selection meets the destination node of the specified stock number of the task requests from the node group.Specifically, resource scheduling device
The resource tag that selection meets the specified stock number of the task requests and includes with the task requests from node group
Node, as destination node.In this way, the destination node determined not only meets, resource tag that task requests include is specified to be held
The resource of row required by task also meets the stock number for executing required by task, to help to improve, the task requests are specified to appoint
The execution efficiency of business.
Based on any of the above-described embodiment, further can also include: resource scheduling device obtain it is every in the node group
The resource information of a node, the resource information record the resource for having the node identification of node and node to have;Scheduling of resource dress
The resource information according to each node in the node group is set, the resource mark that each node in the node group has is recorded
The stock number signed and had.In this way, the money that resource scheduling device can have according to the resource tag and node that node has
Source amount, and then can accurately determine the node for executing specifying in task requests for task.
It further can also include: that specifying in task requests for task has with another task based on any of the above-described embodiment
When having compatibility, resource scheduling device indicates that destination node executes another task;In the task that task requests are specified and another
When business has anti-compatibility, another node different from destination node executes another in resource scheduling device instruction node group
Business.In this way, resource scheduling device is when determining the node of execution task, it is contemplated that the compatibility between task, so as to true
Make the destination node for the task of being suitably executed.
Second aspect, the application provide a kind of resource regulating method, this method comprises: client generates task requests, appoint
Business request includes the corresponding resource tag of the task requests, and task requests pass through the specified execution of resource tag that task requests include
The resource of required by task;Then, client sends task requests to resource scheduling device, and task requests are used for resource scheduling device
According to the resource tag that node each in node group has, the destination node for executing the specified task of task requests, target are determined
The resource tag that there are node task requests to include.
Based on the program, client generates task requests, and task requests are specified by the resource tag that task requests include
The resource of required by task is executed, client sends task requests to resource scheduling device, on the one hand, so as to resource scheduling device root
Destination node is determined according to the resource of the specified required by task of task requests, can find processing capacity and the task requests is suitble to refer to
The node of fixed task, to help to improve the execution efficiency of the specified task of the task node.On the other hand, so as to resource
Dispatching device is when determining the destination node for executing the specified task of the task requests, the resource tag being had according to node, from
The resource tag for including with the task requests is determined in node group, in this way, the money of node group interior joint can be utilized rationally
Source, and then help to improve the resource utilization of node.
The third aspect, the application provide a kind of device, which has the function for each embodiment for realizing above-mentioned first aspect
Energy.The function can also execute corresponding software realization by hardware realization by hardware.The hardware or software include one
A or multiple modules corresponding with above-mentioned function.
In a kind of possible design, which includes communication unit and processing unit, in which:
Receiving unit, for receiving the task requests of client transmission, the task requests include the task requests pair
The resource tag answered, the task requests pass through the specified money for executing required by task of resource tag that the task requests include
Source;
The processing unit, the resource tag for being had according to node each in node group, from the node group really
Set the goal node, the resource tag that there are the destination node task requests to include;Indicate that the destination node executes institute
State the specified task of task requests.
Fourth aspect, the application provide a kind of device, which has the function for each embodiment for realizing above-mentioned second aspect
Energy.The function can also execute corresponding software realization by hardware realization by hardware.The hardware or software include one
A or multiple modules corresponding with above-mentioned function.
In a kind of possible design, which includes communication unit and processing unit, in which:
Processing unit, for generating task requests, the task requests include the corresponding resource tag of the task requests,
The task requests pass through the specified resource for executing required by task of resource tag that the task requests include;
Communication unit, for sending the task requests to resource scheduling device, the task requests are used for the resource
The resource tag that dispatching device has according to node each in node group determines the mesh for executing the specified task of the task requests
Mark node, the resource tag that there are the destination node task requests to include.
5th aspect, the application provide a kind of device, comprising: processor and memory;The memory is calculated for storing
Machine instruction;The processor executes the computer instruction of memory storage, so that the device executes above-mentioned first aspect or first
The resource regulating method in method that any embodiment of aspect provides.It should be noted that the memory can integrate in place
It manages in device, is also possible to independently of except processor.
6th aspect, the application provide a kind of device, comprising: processor and memory;The memory is calculated for storing
Machine instruction;The processor executes the computer instruction of memory storage, so that the device executes above-mentioned second aspect or the
The resource regulating method in method that any embodiment of two aspects provides.It should be noted that the memory can integrate in
In processor, it is also possible to independently of except processor.
7th aspect, the application also provide a kind of system, in the method that any embodiment including first aspect provides
The client in method that any embodiment of resource scheduling device and second aspect provides.
Eighth aspect, the application also provide a kind of computer readable storage medium, in the computer readable storage medium
It is stored with computer instruction, which is computer-executed to realize method described in above-mentioned various aspects.
9th aspect, the application also provides a kind of computer program product comprising instruction, when it runs on computers
When, so that computer executes method described in above-mentioned various aspects.
Detailed description of the invention
Fig. 1 a is a kind of possible system architecture schematic diagram provided by the present application;
Fig. 1 b is alternatively possible system architecture schematic diagram provided by the present application;
Fig. 2 is a kind of resource regulating method schematic diagram provided by the present application;
Fig. 3 is the mapping relations schematic diagram of node identification provided by the present application and resource tag;
Fig. 4 is another resource regulating method schematic diagram provided by the present application;
Fig. 5 is a kind of apparatus structure schematic diagram provided by the present application;
Fig. 6 is another apparatus structure schematic diagram provided by the present application.
Specific embodiment
The application is described in further detail below in conjunction with attached drawing.Concrete operation method in embodiment of the method
It can be applied in Installation practice or system embodiment.Wherein, in the description of the present application, unless otherwise indicated, " multiple "
It is meant that two or more.
Fig. 1 a illustrates a kind of possible system architecture schematic diagram provided by the embodiments of the present application.The system architecture
In include client, resource scheduling device and node group N number of node for including, in Fig. 1 a it is merely exemplary show node 1 ...,
Node N, wherein N is positive integer.
Wherein, the operation that client is submitted according to user generates task requests, and task requests is sent to scheduling of resource
Device, task requests are by the specified resource for executing required by task of resource tag that task requests include, and task requests are for providing
Source dispatching device determines the destination node for executing the specified task of task requests from node group.
Resource scheduling device receives the task requests that client is sent, and according to the N for including in task requests and node group
The resource that a node has determines the destination node for executing the task requests from node group, and indicates that destination node executes and be somebody's turn to do
The specified task of task requests.
The N number of node for including in node group, being executed for executing resource scheduling device instruction for task.
Based on Fig. 1 a, Fig. 1 b illustrates alternatively possible system architecture signal provided by the embodiments of the present application
Figure.
Referring to Fig. 1 b, task requests forwarding module, the task schedule control that resource scheduling device includes are illustrated
Module, label custom block, scheduling engine (Yarn) module and Docker five, mirror image warehouse (Docker Registry)
Module.
Wherein, task requests forwarding module receives the algorithmic dispatching task of user for being responsible for, provides load balancing and ask
Seek transfer capability.
Task schedule control module, for completing the function such as task parameters parsing, task schedule control and task queue management
Energy.As an example, task parameters parsing function, which can be, parses task priority, and task queue management function can be
The parsing task parameters of the task is put into task queue according to the time sequencing for the task that receives, task schedule control function
Can be the time that task queue is put into according to task priority and task, carried out from scheduler task in task queue to Yarn from
Reason, wherein the task priority scheduling of high priority, the identical task of priority are carried out according to the time sequencing for being put into task queue
Scheduling.
Label custom block mainly includes that information collection reports submodule and the customized submodule of multi-tag.Wherein believe
Breath acquisition reports submodule to be used for the resource information that each node has in acquisition node group, on the one hand, information collection reports submodule
The resource information of acquisition is reported to Yarn by block, is task schedule for holding so as to the resource information that Yarn has according to node
The resource of the row required by task;On the other hand, information collection reports submodule that the resource information of acquisition is sent to multi-tag certainly
Submodule is defined, so that the customized submodule of label is that each type of resource generates resource tag respectively according to resource information,
And the corresponding relationship for all resource tags that the mark of node and the node have is reported into Yarn.
Scheduling engine (Yarn) module, all resource tags for the mark of the node received and node to have
Corresponding relationship is stored in node label database, and realizes specific task tune according to all resource tags that node has
Spend function.Wherein, node label database is located in Yarn, and illustratively, node label database can be configuration file
Form is present in Yarn.
As a specific example, Yarn is when receiving task scheduled in task schedule control module, from node
What the resource tag and task requests inquiring the resource tag that each node has in label resources library, and being had according to node had
Resource tag determines the node for executing the specified task of task requests.
The privately owned mirror image storehouse management open source component that Docker Registry is Docker, main offer mirror image upload,
The abilities such as downloading.
Below with reference to system architecture shown in Fig. 1 a and Fig. 1 b, resource regulating method provided by the present application is carried out specifically
It is bright.
Fig. 2 illustrates a kind of resource regulating method flow diagram provided by the present application.This method includes following
Step:
Step 201, client generates task requests.
Wherein, which includes the corresponding resource tag of the task requests, which passes through the task requests
Including the specified resource for executing required by task of resource tag.
It herein, may include one or more resource tags in the task requests.As a kind of implementation, resource tag
It can be present in task requests in the form of list of labels (label list).
As a kind of implementation, the corresponding resource tag of the task requests is determined according to the stock number of required by task.Tool
Body, client determines the corresponding resource tag of the task requests according to the stock number of required by task, alternatively, client according to appoint
Stock number needed for business and preset rules determine the corresponding resource tag of the task requests, the preset rules and resource scheduling device
Determine that the rule of the corresponding resource tag of task requests is consistent.
As another implementation, the corresponding resource tag of the task requests can be to be obtained from resource scheduling device
's.Specifically, client sends query messages to resource scheduling device, which, should for obtaining resource tag set
It include the resource tag that each node has in node group, as an example, the resource mark that node has in resource tag set
The resource that label instruction node has.Then, client receives the response message from resource scheduling device, which includes
The resource tag set.It, can be according to the resource of required by task, from resource mark after client gets the resource tag set
Resource selection one or more resource tag is signed, the resource tag of selection is carried in task requests.
As a specific example, above-mentioned task requests can also be including information some or all of in the following contents:
Task names (name), job identification (job identity), algorithm image banner (image identity), task priority
(priority), number of copies (copy number), the resource of required by task (such as the resources such as CPU, memory and GPU), Yi Jiren
Stock number needed for business.
Illustratively, an operation may include multiple tasks, and submitting an operation with user is to check in traffic video
Act of violating regulations example, client can generate one or more task requests according to the operation, and a task requests can correspond to one
A task can also correspond to multiple tasks.It is divided into two by taking the corresponding task of a task requests as an example, such as by the operation
Task, the corresponding task 1 of task requests 1 are to convert video to image, and the corresponding task 2 of task requests 2 is image procossing, than
Such as recognition of face.In the example, the task 1 for belonging to same operation is identical with the job identification of task 2, that is, task requests 1 include
Job identification and task requests 2 job identification that includes it is identical.
In above-mentioned example, task 1 and task 2 belong to same operation, due to being to have dependence between the task in same operation
Relationship, it is possible to which the task priority setting that will belong to the multiple tasks of same operation is identical, i.e., task requests 1 include
Task priority is identical with the task priority that task requests 2 include.
Before any one task in task 1 and task 2 executes on being dispatched to node, needs first to generate in node and hold
Device includes meeting the resource for executing required by task in container.By taking task 1 is dispatched to node 1 as an example, instruction should in task requests 1
1 corresponding algorithm image banner (image identity) of task, Docker Registry push algorithm image banner are identified
Mirror image to node 1, to generate container in node 1.
Step 202, client sends the task requests to resource scheduling device.Correspondingly, resource scheduling device can connect
Receive the task requests of client transmission.
Step 203, the resource tag that resource scheduling device has according to node each in node group is determined from node group
Destination node.
Wherein, the resource tag which there are task requests to include.
Specifically, after resource scheduling device receives task requests, according to the corresponding resource tag of task requests and
The resource tag that node has determines the destination node for the resource tag that there are task requests to include from node group.
Optionally, the resource tag that each node has in node group can be one or more.
As a kind of implementation, resource scheduling device obtains the resource information of each node in node group, resource letter
The resource that breath record has the node identification of node and node to have;Resource scheduling device is according to the money of each node in node group
Source information records each node resource tag having in node group and the stock number having.In this way, resource scheduling device
The stock number that can be had according to the resource tag and node that node has, and then can accurately determine for executing task
The node of being specified in request for task.
Based on above-mentioned Fig. 1 b, the resource information of each node in label custom block acquisition node group can be used,
And generate the resource tag that each node has.The node identification wherein recorded in resource information can be the IP address of node,
The resource that the node of record has can be the resources such as CPU, GPU, memory, disk.Optionally, the resource tag which has
It can store in node label database.When resource scheduling device executes above-mentioned steps 203, from node label database
The resource tag of each node is traversed, to be the specified task choosing target section of task requests according to the resource tag of node
Point.
As an example, as shown in figure 3, being the mapping relations schematic diagram of node identification and resource tag.Wherein, one
Node can correspond to one or more resource tags, and each resource tag can also correspond to one or more nodes.Referring to Fig. 3,
For the address agreement (internet protocol, IP) interconnected between network using node identification, node 1 (Node 1)
IP address: 192.168.0.10, corresponding resource label 1 (Label 1) and Label 3;The IP address of node 2 (Node 2):
192.168.0.11, corresponding resource label 2 (Label 2);The IP address of node 3 (Node 3): 192.168.0.12, it is corresponding
Label 2 and resource tag M (Label M);The IP address of node N (Node N): 192.168.0.xx, corresponding Label1, money
Source label 3 (Label 3) and Label M, wherein M is positive integer.
Single node is supported in such a way that the unique identification (such as IP address) of node maps resource tag by the example
Customized multiple resource tags, compared to the prior art in can only be given indirectly using primary Yarn interface capability by task queue
Node beats resource tag and node does not support the scheme of multiple resource tags, and it is customized multiple that the application can be implemented as node
Resource tag, and direct appointed task when mission dispatching is combined to request corresponding resource tag, optimization scheduling of resource may be implemented
Process, and then help to improve the utilization rate of node group resource.
Further, the resource tag that there are above-mentioned destination node task requests to include can specifically have at least two realities
Existing mode.
As a kind of implementation, the resource tag that destination node has includes the corresponding resource tag of task requests.
Specifically, determining the resource mark having from node group when resource scheduling device receives the task requests
Label include the node of the corresponding resource tag of task requests, as destination node.
In the implementation, node can have the corresponding resource tag of one or more following resource types: CPU master
Frequently, the model of GPU, GPU block number, memory bar model, bandwidth, computer room information.Such as the model of GPU may include P4, P40,
P100 and V100, such as memory bar model may include Double Data Rate (double data rate, DDR) 1, DDR2, DDR4 etc..
Illustratively, for example node 1 has 1 resource tag, which indicates that CPU frequency is high;For another example, node 2 has 2 moneys
Source label indicates respectively that CPU frequency is low and 1 piece of GPUP100.The resource that the corresponding resource tag of task requests and node have
Label is similar, no longer illustrates one by one herein.
Below with reference to Tables 1 and 2, the resource tag that the resource tag and task requests have to node includes is illustrated
Explanation.
Assuming that including 6 nodes, respectively Node 1, Node 2, Node 3, Node 4,5 and of Node in node group
Node 6, according to the resource information of each node, the resource tag for determining that each node has is as shown in table 1 below.
The resource tag that 1 node of table has
Based on the resource tag that each node in table 1 has, resource tag set as shown in table 2 below is formed.
2 resource tag set of table
The resource that required by task is executed specified by resource tag set and task requests of the client according to the table 2, from
The corresponding resource tag of task requests is selected in resource tag set.
For example, including 3 resource tags in task requests 1, respectively CPU frequency is high, bandwidth is high, 2 pieces of GPU P100,
It include that CPU frequency is high, bandwidth is high, 2 pieces of GPU P100 based on above-mentioned table 1, in 4 resource tags that only Node 3 has, that
The Node 3 is the destination node that can execute the specified task of the task requests 1.
For another example, task requests 2 include 2 resource tags, respectively 4 pieces of GPU P40 and memory DDR1, based on above-mentioned
Table 1 includes at least 4 pieces of GPU P40 and memory DDR1 in the resource tag that Node 2 and Node4 have, then can be from Node
Destination node is determined in 2 and Node4, specifically, destination node can be determined at random from Node 2 and Node4, it can also
To determine destination node according to available resources in Node 2 and Node4, for example select the more nodes of available resources as mesh
Mark node.
By the implementation, after resource scheduling device receives task requests, can be determined according in node group
It out include the node of the corresponding resource tag of task requests with the presence or absence of the resource tag having, to determine the target of execution task
Node, it is this to determine that the mode of destination node is simple according to two resource tag matchings, help to save the time.
As another implementation, the resource that node indicated by the resource tag that destination node has has meets
The requirement of resource indicated by the corresponding resource tag of the task requests.
Specifically, determining the resource mark having from node group when resource scheduling device receives the task requests
The resource that the indicated node of label has, meets the node of resource requirement indicated by the corresponding resource tag of the task requests,
As destination node.
In the implementation, node can have the corresponding resource tag of one or more following resource types: CPU core
Number, memory size, disk size, bandwidth.Illustratively, 2 resource tags that node 4 has indicate respectively that disk can use appearance
Measure 201-300GB and memory size 1000-1500MB.The resource mark that the corresponding resource tag of task requests and node have
It signs similar, no longer illustrates one by one herein.
Below with reference to table 3 and table 4, the resource tag that the resource tag and task requests have to node includes is illustrated
Explanation.
Assuming that including 6 nodes, respectively Node 1, Node 2, Node 3, Node 4,5 and of Node in node group
Node 6, according to the resource information of each node, the resource tag for determining that each node has is as shown in table 3 below.
The resource tag that 3 node of table has
Based on the resource tag that each node in table 3 has, resource tag set as shown in table 4 below is formed.
4 resource tag set of table
The resource that required by task is executed specified by resource tag set and task requests of the client according to the table 4, from
The corresponding resource tag of task requests is selected in resource tag set.
In the implementation, the resource tag that node has resource tag corresponding with task requests can be identical, can also
With not identical.
For example, task requests 1 include 1 resource tag, memory size 500-700MB is indicated, be based on above-mentioned table 3, Node
Resource indicated by the resource tag that 2 and Node 5 has includes memory indicated by the corresponding resource tag of task requests 1
Capacity 500-1000MB, so Node 2 and Node 5 can be used as destination node.Although in addition, 3 moneys that Node 4 has
It does not include the memory size 500-1000MB of the corresponding resource tag of task requests 1 instruction in resource indicated by the label of source, but
It is memory size 1000-1500MB indicated by resource tag that Node 4 has, higher than the memory size requirement of required by task
500-1000MB, that is to say, that Node 4 meets the requirement of resource indicated by the corresponding resource tag of task requests 1, so
Node 4 can also be used as destination node.It therefore, can be from so selecting a node to make in Node 2, Node 5 and Node 4
For destination node, to execute the specified task of the task requests 1.
For another example, task requests 2 include 2 resource tags, indicate respectively memory size 500-900MB, bandwidth 21-50M,
It is signified that the resource as indicated by 2 resource tags for only having Node 2 to have meets the corresponding resource tag of the task requests 2
The requirement for the resource shown, it is possible to regard the Node 2 as destination node.
By this way, the resource that node indicated by the resource tag having in node has, meets the task requests
The requirement of resource indicated by corresponding resource tag, that is, can determine that destination node, without the money being had according to node
Label resource tag corresponding with task requests in source is identical, and to determine destination node, thus realization that can be more flexible is appointed
The scheduling on demand of business.
Further, in the implementation, if available resources change in node, section shown in table 3 is influenced
The resource tag that point has also changes, then needing when the available resources of node change, for table 3 and table 4
In resource tag be updated, specifically, can be updated according to the available resources of node to the resource tag of node.
In conjunction with above two implementation, as another implementation, the resource tag that node has can be above-mentioned
Any or multiple resources label combination in any or multiple resources label and table 3 in table 1, correspondingly, in resource tag set
It is also possible to the combination of above-mentioned table 2 and the resource tag in table 4.Such as 3 resource tags that Node 1 has, it indicates respectively
CPU frequency height, CPU core number 32, bandwidth 21-50M.
Further, as a kind of possible implementation, task requests record has the stock number for executing required by task, real
In the concrete mode of existing above-mentioned steps 203, resource scheduling device selects to meet what the task requests were specified from the node group
The destination node of stock number.
Specifically, resource scheduling device selects to have the corresponding resource tag of the task requests and be met from node group
The node of the specified stock number of the task requests, as destination node.It is asked in this way, the destination node determined not only meets task
Ask including the specified execution required by task of resource tag resource, also meet the stock number for executing required by task, to help
In the execution efficiency for improving the specified task of the task requests.
Step 204, resource scheduling device indicates that the destination node executes the specified task of the task requests.
201- step 204 through the above steps, client generate task requests, and task requests include by task requests
The specified resource for executing required by task of resource tag, client send task requests, resource scheduling device to resource scheduling device
The task requests that client is sent are received, the resource tag and task that resource scheduling device has according to node each in node group
The resource tag that request includes determines that destination node, the destination node have the task requests packet from the node group
The resource tag included;Resource scheduling device indicates that the destination node executes the specified task of the task requests.On the one hand, by
It include the corresponding resource tag of the task requests in the task requests, the task requests include by the task requests
The resource tag specified resource for executing required by task, so the resource for the required by task specified according to task requests determines mesh
Node is marked, the node for the task that processing capacity is suitble to the task requests specified can be found, to help to improve the task section
The execution efficiency of the specified task of point.On the other hand, when determining the destination node for executing the specified task of the task requests, root
The resource tag having according to node determines the resource tag for including with the task requests, in this way, can close from node group
Reason utilizes the resource of node group interior joint, and then helps to improve the resource utilization of node.
It can also include task compatibility resource tag, task compatibility in task requests based on any of the above-described embodiment
Resource tag is used to indicate the compatibility or anti-compatibility of the task requests specified task and another task, so as to basis
Between task whether compatibility, further realize the scheduling on demand of task.
As a kind of possible implementation, when the task and another task that task requests are specified are affinity, money
Source dispatching device indicates that destination node executes another task;There is anti-compatibility in the task that task requests are specified and another task
When, resource scheduling device indicates that another node different from destination node in node group executes another task.In this way, scheduling of resource
Device is when determining the node of execution task, it is contemplated that the compatibility between task may thereby determine that out the task of being suitably executed
Destination node.
As an example, for example the specified task of task requests is task 1, and another task is task 2, scheduling of resource dress
Being set to the specified destination node of task 1 is Node 1.If task 1 and task 2 are affinity, can be executed using Node 1
Task 2.
As another example, for example the specified task of task requests is task 1, and another task is task 3, scheduling of resource
Device is that the specified destination node of task 1 is Node 1.It, can be using except Node if task 1 and task 3 have anti-compatibility
Node except 1 executes task 2, executes on the same node to avoid task 1 and task 2, and then improves task 1 and task 2
Treatment effeciency.
Based on any of the above-described embodiment, flag bit can also be increased in task requests, flag bit is asked for identifying task
Ask whether corresponding resource tag is necessary resource tag.Specifically, if the resource tag of mark bit identification is necessary money
Source label, then destination node must have the resource tag of the mark bit identification;If indicate bit identification resource tag be it is non-must
The resource tag wanted, then destination node can not have the resource tag of the mark bit identification.
As an example, the value of flag bit is 0, indicates that the resource tag of the mark bit identification is non-essential resource mark
Label;The value of flag bit is 1, indicates that the resource tag of the mark bit identification is necessary resource tag.As another example, mark
The value of will position is 1, indicates that the resource tag of the mark bit identification is non-essential resource tag;The value of flag bit is 0, and indicating should
The resource tag for indicating bit identification is necessary resource tag.
As an example, with the value of flag bit be 1 indicate necessary resource tag, the value of flag bit be 0 indicate it is non-must
For the resource tag wanted, it is assumed that the corresponding resource tag of task requests includes GPU P40 and memory DDR, and wherein GPU P40 is
Necessary resource tag, memory DDR is non-essential resource tag, then can indicate in the flag bit in task requests
It is 1,0, referring to table 1, selection has the node of the corresponding resource tag of task requests from node shown in table 1, at this point, root
According to the corresponding resource tag of task requests and flag bit, the Node 2 with necessary resource tag GPU P40 can be determined
With Node 4, and determine from Node 2 and Node 4 for executing specifying in the task requests for task.
In the application, when receiving task requests in resource scheduling device, and determining destination node from node group, due to having
A little tasks when being executed needed for resource include GPU, some tasks when being executed needed for resource do not include GPU, therefore, if
It will be executed in the task schedule to the node with GPU of no GPU demand, cause being badly in need of GPU resource of the task that cannot dispatch again
Onto the node, to cause the waste of the GPU resource on the node.Therefore, in order to avoid the wave of GPU resource in node group
Take, is based on the above method, several optional implementation methods presented below.
Implementation method one, resource scheduling device can according to the corresponding resource tag of task requests whether include GPU money
Source label, selection execute the destination node of the specified task of the task requests.
As a kind of specific example, resource scheduling device does not include GPU's in the corresponding resource tag of the task requests
When resource tag, the node of the resource tag without GPU is selected from node group, and indicates that the node of selection executes the task
The specified task of request.
Specifically, resource scheduling device selects the node of the resource tag without GPU from node group, and from not having
Selection has the destination node of the corresponding resource tag of task requests in the node of the resource tag of GPU, and dispatches the target section
Point executes specifying in task requests for task.
As another specific example, resource scheduling device includes GPU's in the corresponding resource tag of the task requests
When resource tag, the node of resource tag of the selection with GPU from node group, and indicate that the node of selection executes the task and asks
Seek specified task.
Specifically, resource scheduling device selects the node of the resource tag with GPU from node group, and from GPU
Resource tag node in selection there is the destination node of the corresponding resource tag of task requests, and dispatch the destination node and hold
Being specified in row task requests for task.
By the implementation method, according to the corresponding resource tag of task requests whether include GPU resource tag, selection holds
The destination node of the specified task of the row task requests, may be implemented the resource for the required by task specified according to task requests,
GPU resource scheduling is carried out, on demand so as to promote the resource utilization of node group interior joint, especially promotion GPU resource
Utilization rate helps avoid causing the GPU resource more than needed in the node to waste due to cpu resource is depleted in node.
Implementation method two, task requests can also include the stock number for executing required by task, and resource scheduling device can root
It whether include GPU resource amount according to the stock number of the execution required by task, selection executes the target of the specified task of the task requests
Node.
As a kind of specific example, if the stock number for executing required by task does not include GPU resource, from the section
The node without the GPU resource is selected in point group, and indicates that the node of selection executes the specified task of the task requests.
As another specific example, if the stock number for executing required by task includes GPU resource, from the section
Selection has the node of the GPU resource in point group, and indicates that the node of selection executes the specified task of the task requests.
A kind of possible implementation of above-mentioned resource regulating method is discussed in detail below with reference to Fig. 4.
Fig. 4 example shows another resource regulating method flow diagram provided by the embodiments of the present application.This method packet
Include following steps:
Step 401, resource scheduling device receives task requests.
Step 402, resource scheduling device determines whether executing and specifying in task requests for task needs GPU;If so, holding
Row step 403;If it is not, thening follow the steps 406.
Step 403, resource scheduling device traverse node tag database searches the node of the resource tag with GPU, if
It is to then follow the steps 404;If it is not, thening follow the steps 409.Wherein, node identification and resource are stored in node label database
The corresponding relationship of label.
Step 404, resource scheduling device is determined from the node of the resource tag with GPU with task requests pair
The destination node for the resource tag answered.
Step 405, resource scheduling device regulation goal node executes the specified task of task requests.
Step 406, resource scheduling device traverse node tag database searches the node of the resource tag without GPU,
If so, thening follow the steps 407;If it is not, thening follow the steps 409.Wherein, node identification and money are stored in node label database
The corresponding relationship of source label.
Step 407, resource scheduling device is determined from the node of the resource tag without GPU with task requests
The destination node of corresponding resource tag.
Step 408, resource scheduling device regulation goal node executes the specified task of task requests.
Step 409, resource scheduling device sends task schedule failed message to client, and abandons dispatching the task requests
Specified task.
Pass through the example, the optimum organization tune of the GPU resource based on the customized multiple resource tags of node and sensing node
Degree, compared with the existing technology in scheduling when all nodes put on an equal footing, the scheme of the not no GPU resource of sensing node, the application
In example may be implemented task GPU resource scheduling on demand, help to improve the resource utilization of each node in node group,
Especially improve the utilization rate of the GPU resource of node.
It is understood that each equipment in above-described embodiment is in order to realize corresponding function, it may include executing
The corresponding hardware configuration of each function and/or software module.Those skilled in the art should be readily appreciated that, in conjunction with herein
Each exemplary unit and algorithm steps of the disclosed embodiments description, the application can be with hardware or hardware and computer software
Combining form realize.Some functions is executed in a manner of hardware or computer software driving hardware actually, is depended on
The specific application and design constraint of technical solution.Professional technician can use not Tongfang to each specific application
Method realizes described function, but this realizes it is not considered that exceeding scope of the present application.
Using integrated unit, Fig. 5 shows the possible of device involved in the embodiment of the present application
Exemplary block diagram, the device 500 can exist in the form of software.Device 500 may include: that communication unit 501 and processing are single
Member 502.As a kind of implementation, which may include receiving unit and transmission unit.Processing unit 502 is used
Control management is carried out in the movement to device 500.Communication unit 501 is for supporting device 500 and other devices (such as client
End) communication.Device 500 can also include storage unit 503, for the program code and data of storage device 500, such as on
The resource tag that the method interior joint of stating has.
Wherein, processing unit 502 can be processor or controller, such as can be universal cpu, general processor, number
Word signal processing (digital signal processing, DSP), specific integrated circuit (application specific
Integrated circuits, ASIC), field programmable gate array (field programmable gate array,
FPGA) either other programmable logic device, transistor logic, hardware component or any combination thereof.It may be implemented
Or it executes and combines various illustrative logic blocks, module and circuit described in present disclosure.The processor
It can be the combination for realizing computing function, combined for example including one or more microprocessors, the combination of DSP and microprocessor
Deng.Communication unit 501 can be communication interface, transceiver or transmission circuit etc., wherein the communication interface is to be referred to as, specific
In realization, which may include multiple interfaces.Storage unit 503 can be memory.
The device 500 can also be that resource scheduling device involved in the application, processing unit 502 can support device
500 execute the movement of resource scheduling device in above each method example, such as processing unit 502 can execute the step in Fig. 2
Rapid 203 and step 204.Communication unit 501 can support the communication between device 500 and client, for example, communication unit 501
For supporting device 500 to execute the step 202 in Fig. 2.
Specifically, communication unit 501 can be used for receiving the task requests of client transmission, and the task requests include described
The corresponding resource tag of task requests, the task requests specify execution task by the resource tag that the task requests include
Required resource;
The processing unit 502 can be used for the resource tag having according to node each in node group, from the node group
Middle determining destination node, the resource tag that there are the destination node task requests to include;Indicate that the destination node is held
The specified task of the row task requests.
In one possible implementation, the resource tag includes the resource tag for identifying GPU;The processing
Unit 502, is also used to: when the corresponding resource tag of the task requests does not include the resource tag of the GPU, from the section
The node of the resource tag without the GPU is selected in point group;Indicate that the node of selection executes what the task requests were specified
Task.
In one possible implementation, the task requests record has the stock number for executing the required by task;Institute
State processing unit 502, be used for: selection meets the destination node of the specified stock number of the task requests from the node group.
In one possible implementation, processing unit 502 are also used to: obtaining each node in the node group
Resource information, the resource information records the resource for having the node identification of node and node to have;According in the node group
Each node resource information, record each node resource tag having and the resource having in the node group
Amount.
In one possible implementation, the processing unit 502, is also used to: appointing what the task requests were specified
When business is affinity with another task, indicate that the destination node executes another task;It is specified in the task requests
Task and another task have anti-compatibility when, indicate that another node different from the destination node in the node group is held
Row another task.
The device 500 can also be client involved in the application.Processing unit 502 can support device 500 to execute
Above in each method example client movement, such as processing unit 502 be used for support device 500 execute Fig. 2 in step
201.Communication unit 501 can support the communication between device 500 and other devices (such as resource scheduling device), for example, logical
The step 202 that letter unit 501 is used to that device 500 to be supported to execute in Fig. 2.
Specifically, processing unit 502, can be used for generating task requests, and the task requests include the task requests pair
The resource tag answered, the task requests pass through the specified money for executing required by task of resource tag that the task requests include
Source;
Communication unit 501 can be used for sending the task requests to resource scheduling device, and the task requests are for described
The resource tag that resource scheduling device has according to node each in node group determines and executes the specified task of the task requests
Destination node, the destination node has the task requests resource tag that includes.
As shown in fig.6, being a kind of schematic device provided by the present application, which can be above-mentioned resource scheduling device
Or client.The device 600 includes: memory 601, processor 602 and communication interface 603.Optionally, device 600 can be with
Including bus 604.Wherein, communication interface 603, processor 602 and memory 601 can be connected with each other by bus 604;Always
Line 604 can be Peripheral Component Interconnect standard (peripheral component interconnect, abbreviation PCI) bus or
Expanding the industrial standard structure (extended industry standard architecture, abbreviation EISA) bus etc..It is described
Bus 604 can be divided into address bus, data/address bus, control bus etc..Only to be indicated with a thick line in Fig. 6 convenient for indicating,
It is not intended that only a bus or a type of bus.
Processor 602 can be a CPU, microprocessor, ASIC, or one or more for controlling application scheme journey
The integrated circuit that sequence executes.
Communication interface 603, using the device of any transceiver one kind, for other devices or communication, such as
Ethernet, wireless access network (radio access network, RAN), WLAN (wireless local area
Networks, WLAN), wired access network etc..
Memory 601 can be read-only memory (read-only memory, ROM) or can store static information and instruction
Other kinds of static storage device, random access memory (random access memory, RAM) or letter can be stored
The other kinds of dynamic memory of breath and instruction, is also possible to Electrically Erasable Programmable Read-Only Memory (Electrically
Erasable Programmable read only memory), CD-ROM (compact disc read-only
Memory, CD-ROM) or other optical disc storages, optical disc storage (including compression optical disc, laser disc, optical disc, Digital Versatile Disc,
Blu-ray Disc etc.), magnetic disk storage medium or other magnetic storage apparatus or can be used in carrying or storing that there is instruction or number
According to structure type desired program code and can by any other medium of computer access, but not limited to this.Memory
It can be and be individually present, be connected by bus 604 with processor.Memory can also be integrated with processor.
Wherein, memory 601 be used for store execution application scheme computer executed instructions, and by processor 602
Control executes.Processor 602 is for executing the computer executed instructions stored in memory 601, to realize that the application is above-mentioned
The method that embodiment provides.
Optionally, the computer executed instructions in the embodiment of the present application can also be referred to as application code, the application
Embodiment is not especially limited this.
Method provided by the embodiment of the present application can be performed in the said goods, has the corresponding functional module of execution method and has
Beneficial effect.The not technical detail of detailed description in the present embodiment, reference can be made to method provided by the embodiment of the present application.
In the above-described embodiments, it can be realized wholly or partly by software, hardware or a combination thereof, when using soft
When part program is realized, can entirely or partly it realize in the form of a computer program product.Computer program product includes one
A or multiple instruction.When loading on computers and executing computer program instructions, entirely or partly generate according to the application
The process or function of embodiment.Computer can be general purpose computer, special purpose computer, computer network or other can compile
Range device.Instruction can store in computer storage medium, or from a computer storage medium to another computer
Storage medium transmission, for example, instruction can from a web-site, computer, server or data center by it is wired (such as
Coaxial cable, optical fiber, twisted pair) or wireless (such as infrared, wireless, microwave etc.) mode to another web-site, computer,
Server or data center are transmitted.Computer storage medium can be any medium that computer can access and either wrap
The data storage devices such as server, the data center integrated containing one or more media.Medium can be magnetic medium, (for example,
Floppy disk, hard disk, tape, magneto-optic disk (MO) etc.), optical medium (such as CD) or semiconductor medium (such as ROM, EPROM,
EEPROM, solid state hard disk (solid state disk, SSD)) etc..
The embodiment of the present application is referring to the method, equipment (system) and computer program product according to the embodiment of the present application
Flowchart and/or the block diagram describe.It should be understood that can be by each process in instruction implementation flow chart and/or block diagram
And/or the combination of the process and/or box in box and flowchart and/or the block diagram.These instructions be can provide to general meter
Calculation machine, special purpose computer, Embedded Processor or other programmable data processing devices processor to generate a machine, make
It obtains and is generated by the instruction that computer or the processor of other programmable data processing devices execute for realizing in flow chart one
The device for the function of being specified in a process or multiple processes and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
Obviously, those skilled in the art can carry out various modification and variations without departing from the essence of the application to the application
Mind and range.In this way, if these modifications and variations of the application belong to the range of the claim of this application and its equivalent technologies
Within, then the application is also intended to include these modifications and variations.
Claims (14)
1. a kind of resource regulating method characterized by comprising
The task requests that client is sent are received, the task requests include the corresponding resource tag of the task requests, described
Task requests pass through the specified resource for executing required by task of resource tag that the task requests include;
According to the resource tag that node each in node group has, destination node, the target section are determined from the node group
The resource tag that there are point the task requests to include;
Indicate that the destination node executes the specified task of the task requests.
2. the method according to claim 1, wherein the method also includes:
When the corresponding resource tag of the task requests does not include the resource tag of the GPU, selected from the node group
The node of resource tag without the GPU;
Indicate that the node of selection executes the specified task of the task requests.
3. method according to claim 1 or 2, which is characterized in that the task requests record, which has, executes the task institute
The stock number needed;
The resource tag being had according to node each in node group determines destination node from the node group, comprising:
Selection meets the destination node of the specified stock number of the task requests from the node group.
4. method according to any one of claims 1 to 3, which is characterized in that the method also includes:
The resource information of each node in the node group is obtained, the resource information record has the node identification and section of node
The resource that point has;
According to the resource information of each node in the node group, the resource that each node in the node group has is recorded
Label and the stock number having.
5. method according to any one of claims 1 to 4, which is characterized in that the method also includes:
When the task and another task that the task requests are specified are affinity, it is described another to indicate that the destination node executes
One task;
The task requests specify task and another task have anti-compatibility when, indicate in the node group with the mesh
It marks the different another node of node and executes another task.
6. a kind of resource regulating method characterized by comprising
Task requests are generated, the task requests include the corresponding resource tag of the task requests, and the task requests pass through
The specified resource for executing required by task of the resource tag that the task requests include;
The task requests are sent to resource scheduling device, the task requests are for the resource scheduling device according to node group
In the resource tag that has of each node, determine the destination node for executing the specified task of the task requests, the target section
The resource tag that there are point the task requests to include.
7. a kind of device characterized by comprising
Communication unit, for receiving the task requests of client transmission, the task requests include that the task requests are corresponding
Resource tag, the task requests pass through the specified resource for executing required by task of resource tag that the task requests include;
Processing unit, the resource tag for being had according to node each in node group determine target section from the node group
Point, the resource tag that there are the destination node task requests to include;Indicate that the destination node executes the task and asks
Seek specified task.
8. device according to claim 7, which is characterized in that the processing unit is also used to:
When the corresponding resource tag of the task requests does not include the resource tag of the GPU, selected from the node group
The node of resource tag without the GPU;
Indicate that the node of selection executes the specified task of the task requests.
9. device according to claim 7 or 8, which is characterized in that the task requests record, which has, executes the task institute
The stock number needed;The processing unit, is specifically used for:
Selection meets the destination node of the specified stock number of the task requests from the node group.
10. device according to any one of claims 7 to 9, which is characterized in that the processing unit is also used to:
The resource information of each node in the node group is obtained, the resource information record has the node identification and section of node
The resource that point has;
According to the resource information of each node in the node group, the resource that each node in the node group has is recorded
Label and the stock number having.
11. according to the described in any item devices of claim 7 to 10, which is characterized in that the processing unit is also used to:
When the task and another task that the task requests are specified are affinity, it is described another to indicate that the destination node executes
One task;
The task requests specify task and another task have anti-compatibility when, indicate in the node group with the mesh
It marks the different another node of node and executes another task.
12. a kind of device characterized by comprising
Processing unit, for generating task requests, the task requests include the corresponding resource tag of the task requests, described
Task requests pass through the specified resource for executing required by task of resource tag that the task requests include;
Communication unit, for sending the task requests to resource scheduling device, the task requests are used for the scheduling of resource
The resource tag that device has according to node each in node group determines the target section for executing the specified task of the task requests
Point, the resource tag that there are the destination node task requests to include.
13. a kind of device, which is characterized in that including processor and memory;The processor executes the calculating of memory storage
Machine instruction, so that method described in any one of described device perform claim requirement 1 to 6.
14. a kind of computer readable storage medium, which is characterized in that store computer in the computer readable storage medium and refer to
It enables, the computer instruction is computer-executed to realize method described in any one of claims 1 to 6.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810943505.XA CN109375992A (en) | 2018-08-17 | 2018-08-17 | A kind of resource regulating method and device |
PCT/CN2019/081200 WO2020034646A1 (en) | 2018-08-17 | 2019-04-03 | Resource scheduling method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810943505.XA CN109375992A (en) | 2018-08-17 | 2018-08-17 | A kind of resource regulating method and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109375992A true CN109375992A (en) | 2019-02-22 |
Family
ID=65404025
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810943505.XA Pending CN109375992A (en) | 2018-08-17 | 2018-08-17 | A kind of resource regulating method and device |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN109375992A (en) |
WO (1) | WO2020034646A1 (en) |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109922316A (en) * | 2019-03-04 | 2019-06-21 | 北京旷视科技有限公司 | Media resource scheduling and managing medium resource method, apparatus and electronic equipment |
WO2020034646A1 (en) * | 2018-08-17 | 2020-02-20 | 华为技术有限公司 | Resource scheduling method and device |
CN110968424A (en) * | 2019-09-12 | 2020-04-07 | 广东浪潮大数据研究有限公司 | Resource scheduling method, device and storage medium based on K8s |
CN111090511A (en) * | 2019-12-24 | 2020-05-01 | 北京推想科技有限公司 | Task processing method and device and computer readable storage medium |
CN111198756A (en) * | 2019-12-28 | 2020-05-26 | 北京浪潮数据技术有限公司 | Application scheduling method and device of kubernets cluster |
CN111552550A (en) * | 2020-04-26 | 2020-08-18 | 星环信息科技(上海)有限公司 | Task scheduling method, device and medium based on GPU (graphics processing Unit) resources |
CN111556126A (en) * | 2020-04-24 | 2020-08-18 | 杭州浮云网络科技有限公司 | Model management method, system, computer device and storage medium |
CN112199200A (en) * | 2020-12-04 | 2021-01-08 | 腾讯科技(深圳)有限公司 | Resource scheduling method and device, computer equipment and storage medium |
CN112395061A (en) * | 2020-11-17 | 2021-02-23 | 广东电科院能源技术有限责任公司 | Computing task scheduling device and method |
CN112667378A (en) * | 2020-12-28 | 2021-04-16 | 紫光云技术有限公司 | Computing resource scheduling method based on resource label |
CN112689007A (en) * | 2020-12-23 | 2021-04-20 | 江苏苏宁云计算有限公司 | Resource allocation method, device, computer equipment and storage medium |
CN112861346A (en) * | 2021-02-07 | 2021-05-28 | 北京润尼尔网络科技有限公司 | Data processing system, method and electronic equipment |
WO2021143287A1 (en) * | 2020-01-16 | 2021-07-22 | 珠海格力电器股份有限公司 | Service scheduling method and apparatus, electronic device, and storage medium |
CN113344311A (en) * | 2020-03-03 | 2021-09-03 | 北京国双科技有限公司 | Task execution method and device, storage medium, processor and electronic equipment |
WO2022088659A1 (en) * | 2020-10-26 | 2022-05-05 | 北京市商汤科技开发有限公司 | Resource scheduling method and apparatus, electronic device, storage medium, and program product |
WO2022252510A1 (en) * | 2021-06-04 | 2022-12-08 | 深圳前海微众银行股份有限公司 | Resource management method, apparatus and device |
CN117579705A (en) * | 2024-01-16 | 2024-02-20 | 四川并济科技有限公司 | System and method for dynamically scheduling servers based on batch data requests |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113535348A (en) * | 2020-04-21 | 2021-10-22 | 华为技术有限公司 | Resource scheduling method and related device |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104137482A (en) * | 2014-04-14 | 2014-11-05 | 华为技术有限公司 | Disaster recovery data center configuration method and device under cloud computing framework |
US20160103699A1 (en) * | 2014-10-13 | 2016-04-14 | Vmware, Inc. | Cloud virtual machine defragmentation for hybrid cloud infrastructure |
CN105677467A (en) * | 2015-12-31 | 2016-06-15 | 中国科学院深圳先进技术研究院 | Yarn resource scheduler based on quantified labels |
CN107135257A (en) * | 2017-04-28 | 2017-09-05 | 东方网力科技股份有限公司 | Task is distributed in a kind of node cluster method, node and system |
CN107515784A (en) * | 2016-06-16 | 2017-12-26 | 阿里巴巴集团控股有限公司 | A kind of method and apparatus of computing resource in a distributed system |
CN107818013A (en) * | 2016-09-13 | 2018-03-20 | 华为技术有限公司 | A kind of application scheduling method thereof and device |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106020937A (en) * | 2016-07-07 | 2016-10-12 | 腾讯科技(深圳)有限公司 | Method, device and system for creating virtual machine |
CN107038069B (en) * | 2017-03-24 | 2020-05-08 | 北京工业大学 | Dynamic label matching DLMS scheduling method under Hadoop platform |
CN109144710B (en) * | 2017-06-16 | 2020-12-04 | 中国移动通信有限公司研究院 | Resource scheduling method, device and computer readable storage medium |
CN109375992A (en) * | 2018-08-17 | 2019-02-22 | 华为技术有限公司 | A kind of resource regulating method and device |
-
2018
- 2018-08-17 CN CN201810943505.XA patent/CN109375992A/en active Pending
-
2019
- 2019-04-03 WO PCT/CN2019/081200 patent/WO2020034646A1/en active Application Filing
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104137482A (en) * | 2014-04-14 | 2014-11-05 | 华为技术有限公司 | Disaster recovery data center configuration method and device under cloud computing framework |
US20160103699A1 (en) * | 2014-10-13 | 2016-04-14 | Vmware, Inc. | Cloud virtual machine defragmentation for hybrid cloud infrastructure |
CN105677467A (en) * | 2015-12-31 | 2016-06-15 | 中国科学院深圳先进技术研究院 | Yarn resource scheduler based on quantified labels |
CN107515784A (en) * | 2016-06-16 | 2017-12-26 | 阿里巴巴集团控股有限公司 | A kind of method and apparatus of computing resource in a distributed system |
CN107818013A (en) * | 2016-09-13 | 2018-03-20 | 华为技术有限公司 | A kind of application scheduling method thereof and device |
CN107135257A (en) * | 2017-04-28 | 2017-09-05 | 东方网力科技股份有限公司 | Task is distributed in a kind of node cluster method, node and system |
Cited By (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020034646A1 (en) * | 2018-08-17 | 2020-02-20 | 华为技术有限公司 | Resource scheduling method and device |
CN109922316A (en) * | 2019-03-04 | 2019-06-21 | 北京旷视科技有限公司 | Media resource scheduling and managing medium resource method, apparatus and electronic equipment |
CN110968424A (en) * | 2019-09-12 | 2020-04-07 | 广东浪潮大数据研究有限公司 | Resource scheduling method, device and storage medium based on K8s |
CN110968424B (en) * | 2019-09-12 | 2023-04-07 | 广东浪潮大数据研究有限公司 | Resource scheduling method, device and storage medium based on K8s |
CN111090511A (en) * | 2019-12-24 | 2020-05-01 | 北京推想科技有限公司 | Task processing method and device and computer readable storage medium |
CN111198756A (en) * | 2019-12-28 | 2020-05-26 | 北京浪潮数据技术有限公司 | Application scheduling method and device of kubernets cluster |
WO2021143287A1 (en) * | 2020-01-16 | 2021-07-22 | 珠海格力电器股份有限公司 | Service scheduling method and apparatus, electronic device, and storage medium |
CN113344311A (en) * | 2020-03-03 | 2021-09-03 | 北京国双科技有限公司 | Task execution method and device, storage medium, processor and electronic equipment |
CN111556126A (en) * | 2020-04-24 | 2020-08-18 | 杭州浮云网络科技有限公司 | Model management method, system, computer device and storage medium |
CN111552550A (en) * | 2020-04-26 | 2020-08-18 | 星环信息科技(上海)有限公司 | Task scheduling method, device and medium based on GPU (graphics processing Unit) resources |
WO2022088659A1 (en) * | 2020-10-26 | 2022-05-05 | 北京市商汤科技开发有限公司 | Resource scheduling method and apparatus, electronic device, storage medium, and program product |
CN112395061A (en) * | 2020-11-17 | 2021-02-23 | 广东电科院能源技术有限责任公司 | Computing task scheduling device and method |
CN112199200A (en) * | 2020-12-04 | 2021-01-08 | 腾讯科技(深圳)有限公司 | Resource scheduling method and device, computer equipment and storage medium |
CN112689007A (en) * | 2020-12-23 | 2021-04-20 | 江苏苏宁云计算有限公司 | Resource allocation method, device, computer equipment and storage medium |
CN112667378A (en) * | 2020-12-28 | 2021-04-16 | 紫光云技术有限公司 | Computing resource scheduling method based on resource label |
CN112861346A (en) * | 2021-02-07 | 2021-05-28 | 北京润尼尔网络科技有限公司 | Data processing system, method and electronic equipment |
WO2022252510A1 (en) * | 2021-06-04 | 2022-12-08 | 深圳前海微众银行股份有限公司 | Resource management method, apparatus and device |
CN117579705A (en) * | 2024-01-16 | 2024-02-20 | 四川并济科技有限公司 | System and method for dynamically scheduling servers based on batch data requests |
CN117579705B (en) * | 2024-01-16 | 2024-04-02 | 四川并济科技有限公司 | System and method for dynamically scheduling servers based on batch data requests |
Also Published As
Publication number | Publication date |
---|---|
WO2020034646A1 (en) | 2020-02-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109375992A (en) | A kind of resource regulating method and device | |
US10628236B2 (en) | System and method for inter-datacenter communication | |
US20190066014A1 (en) | Work task assignment method, apparatus and computing system | |
CN107590001B (en) | Load balancing method and device, storage medium and electronic equipment | |
CN105450618B (en) | A kind of operation method and its system of API server processing big data | |
CN107657532A (en) | The processing method and system of a kind of operation flow | |
US9853906B2 (en) | Network prioritization based on node-level attributes | |
EP2944072B1 (en) | System for service allocation | |
CN105282215A (en) | Reputation-based strategy for forwarding and responding to interests over a content centric network | |
US20070234369A1 (en) | Policy based message aggregation framework | |
CN106330769A (en) | Business processing method and server | |
US11949760B2 (en) | System and method for content parsing | |
US20120216203A1 (en) | Holistic task scheduling for distributed computing | |
Drezner et al. | The central warehouse location problem revisited | |
US20220116479A1 (en) | Systems and methods for managing an automotive edge computing environment | |
CN108605017A (en) | Inquiry plan and operation perception communication buffer management | |
CN103270520A (en) | Importance class based data management | |
CN109711768A (en) | The automobile-used stream tube control system of commodity, method and information data processing terminal | |
Zhong et al. | Integrated scheduling of production and distribution with release dates and capacitated deliveries | |
Arendt et al. | Intelligent control of freight services on the basis of autonomous multi-agent transport coordination | |
CN109242680B (en) | Method for dynamically adjusting block chain consensus | |
US20230333884A1 (en) | Method and system for performing domain level scheduling of an application in a distributed multi-tiered computing environment using reinforcement learning | |
US20120265801A1 (en) | Out of order assembling of data packets | |
US20230333880A1 (en) | Method and system for dynamic selection of policy priorities for provisioning an application in a distributed multi-tiered computing environment | |
US11100454B1 (en) | CDD with heuristics for automated variable use-case based constrained logistics route optimization |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190222 |