CN106897132A - The method and device of a kind of server task scheduling - Google Patents
The method and device of a kind of server task scheduling Download PDFInfo
- Publication number
- CN106897132A CN106897132A CN201710107912.2A CN201710107912A CN106897132A CN 106897132 A CN106897132 A CN 106897132A CN 201710107912 A CN201710107912 A CN 201710107912A CN 106897132 A CN106897132 A CN 106897132A
- Authority
- CN
- China
- Prior art keywords
- task
- cpu
- task groups
- groups
- binding
- 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
- G06F9/5044—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 hardware capabilities
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multi Processors (AREA)
Abstract
The invention discloses the method and device of a kind of scheduling of server task, by the type according to task, task is distributed to pre-binding task groups；Using preset task dispatching algorithm corresponding with pre-binding task groups, task is scheduled；Wherein, pre-binding task groups comprising the first predetermined number CPU, and with the internal memory group binding of the memory node comprising the second predetermined number.Different types of task is assigned to different task groups, and have the CPU and memory node of respective numbers in task groups, will CPU and memory node be grouped, to perform different types of task, namely in task scheduling, not only allow for the characteristic of task, it is also considered that the CPU of server and the ardware feature of memory node.In task scheduling, it is contemplated that the ardware feature of server, it is possible to reduce the performance loss that task switching brings, and then the resource utilization of server is improve, improve server performance.It can be seen that, the application is conducive to improving server performance.
Description
Technical field
The present invention relates to server RAS technical fields, the method and dress of more particularly to a kind of server task scheduling
Put.
Background technology
With the progress of computer technology, the performance to server proposes new requirement.
In the hardware aspect of server, multiple-path multiple-core CPU is the sustainable development direction of CPU technologies, and internal memory is then with non-
Consistent storage access structure (NUMA) technology is main flow.Server can perform various tasks, and task can be divided into meter
Calculate intensive task and I/O intensive tasks.Computation-intensive task can refer to the CPU usages such as calculating, logic judgment
Task high, the frequent switching of computation-intensive task can reduce cpu busy percentage.I/O intensive tasks can refer to frequent net
Network transmission, read the type such as hard disk and other I/O equipment task, due to I/O intensive tasks perform during, the big portions of CPU
All it is that, in wait state, therefore the switching of I/O intensive tasks can improve the efficiency of server between timesharing.
Server only considers the characteristic of task, such as task scheduling of linux operating systems when being scheduled to task
Be task based access control priority scheduling, i.e. task priority formulate formulated using different scheduling strategies.In the task of carrying out tune
Task characteristic is only considered when spending, the hardware structure of server is not considered, may result in the degraded performance of server, for example, appointing
Business frequent switching can take CPU, or due to the wait of memory source and I/O resources so that CPU is constantly in idle shape
State.Therefore how when server task is dispatched, the performance for improving server is this area problem demanding prompt solution.
The content of the invention
It is an object of the invention to provide the method and device of a kind of scheduling of server task, it is therefore intended that solve existing skill
Server relatively low problem of performance in task scheduling in art.
In order to solve the above technical problems, the present invention provides a kind of method of server task scheduling, the method includes：
According to the type of task, the task is distributed to pre-binding task groups；
Using preset task dispatching algorithm corresponding with the pre-binding task groups, the task is scheduled；
Wherein, the pre-binding task groups comprising the first predetermined number CPU, and with the internal memory comprising the second predetermined number
The internal memory group binding of node.
Alternatively, preset task dispatching algorithm corresponding with the pre-binding task groups is utilized described, to the task
Also include after being scheduled：
The CPU usage and memory node occupancy in the pre-binding task groups are obtained using monitor task；
According to the CPU usage and memory node occupancy, the CPU of the dynamic increase and decrease pre-binding task groups
And the quantity of the memory node.
Alternatively, it is described according to the CPU usage and memory node occupancy, the dynamic increase and decrease pre-binding task
The CPU of group and the quantity of the memory node include：
When the CPU usage is more than or equal to the first predetermined threshold value, the 3rd predetermined number is called from other task groups
CPU；
When the memory node occupancy is more than or equal to the second predetermined threshold value, the 4th is called to preset from other task groups
The memory node of quantity；
When the CPU usage is less than three predetermined threshold values, the CPU that will be called returns to corresponding task groups；
When the memory node occupancy is less than four predetermined threshold values, the memory node that will be called returns to correspondence
Task groups.
Alternatively, the binding procedure of the task and the pre-binding task groups is specially：
According to the task type of required execution, multiple tasks group is created；
According to needed for the hardware configuration of server and each described task groups, by the CPU of respective numbers and each institute
Task groups are stated to be bound；
Needed for internal memory according to each task groups, the memory node of respective numbers is carried out with task groups each described
Binding；
According to the characteristic of each task groups, corresponding task scheduling algorithm is chosen；
When the task is created, according to the type of the task, the task is bound with corresponding task groups,
Obtain the pre-binding task groups.
Alternatively, the task groups include：Task more than CPU intensive type task groups, I/O intensities and committed memory resource
Group, I/O intensities and the few task groups of committed memory resource and operating system internal task group.
Additionally, present invention also offers a kind of device of server task scheduling, the device includes：
Distribute module, for the type according to task, the task is distributed to pre-binding task groups；
Scheduler module, for utilizing preset task dispatching algorithm corresponding with the pre-binding task groups, to the task
It is scheduled；
Wherein, the pre-binding task groups comprising the first predetermined number CPU, and with the internal memory comprising the second predetermined number
The internal memory group binding of node.
Alternatively, also include：
Monitoring module, for obtaining CPU usage and internal memory section in the pre-binding task groups using monitor task
Point occupancy；
Dynamic swap modules, for according to the CPU usage and memory node occupancy, dynamic increase and decrease is described pre- to tie up
Determine the CPU of task groups and the quantity of the memory node.
Alternatively, the dynamic swap modules include：
CPU adding units, for when the CPU usage is more than or equal to the first predetermined threshold value, from other task groups
Call the CPU of the 3rd predetermined number；
Memory node adding unit, for when the memory node occupancy be more than or equal to the second predetermined threshold value when, from
Other task groups call the memory node of the 4th predetermined number；
CPU reduces unit, for when the CPU usage is less than three predetermined threshold values, the CPU that will be called to be returned
To corresponding task groups；
Memory node adding unit, for when the memory node occupancy is less than four predetermined threshold values, by what is called
The memory node returns to corresponding task groups.
Alternatively, also include：
Creation module, for the task type performed according to needed for, creates multiple tasks group；
CPU binding modules, for according to needed for the hardware configuration of server and each described task groups, by respective counts
The CPU of amount is bound with task groups each described；
Internal memory binding module, needed for for the internal memory according to each task groups, by the memory node of respective numbers with
Each described task groups is bound；
Dispatching algorithm chosen module, for the characteristic according to each task groups, chooses corresponding task scheduling algorithm；
Task binding module, for when the task is created, according to the type of the task, by the task with it is corresponding
Task groups bound, obtain the pre-binding task groups.
Alternatively, the task groups include：Task more than CPU intensive type task groups, I/O intensities and committed memory resource
Group, I/O intensities and the few task groups of committed memory resource and operating system internal task group.
The method and device of a kind of server task scheduling provided by the present invention, by the type according to task, will
Above-mentioned task is distributed to pre-binding task groups；Using preset task dispatching algorithm corresponding with above-mentioned pre-binding task groups, to upper
The task of stating is scheduled；Wherein, above-mentioned pre-binding task groups comprising the first predetermined number CPU, and with comprising the second present count
The internal memory group binding of the memory node of amount.Different types of task is assigned to different task groups, and has corresponding in task groups
The CPU and memory node of quantity, will CPU and memory node be grouped, to perform different types of task, namely in task
During scheduling, the characteristic of task is not only allowed for, it is also considered that the CPU of server and the ardware feature of memory node.Adjusted in task
When spending, it is contemplated that the ardware feature of server, it is possible to reduce the performance loss that task switching brings, and then improve server
Resource utilization, improves server performance.It can be seen that, the application is conducive to improving server performance.
Brief description of the drawings
For the clearer explanation embodiment of the present invention or the technical scheme of prior art, below will be to embodiment or existing
The accompanying drawing to be used needed for technology description is briefly described, it should be apparent that, drawings in the following description are only this hair
Some bright embodiments, for those of ordinary skill in the art, on the premise of not paying creative work, can be with root
Other accompanying drawings are obtained according to these accompanying drawings.
A kind of flow of specific embodiment of the server task dispatching method that Fig. 1 is provided by the embodiment of the present invention is shown
It is intended to；
The CPU and internal memory of the task groups that Fig. 2 is provided by the embodiment of the present invention divide schematic diagram；
Fig. 3 is tied to the schematic diagram of different task group by different types of task that the embodiment of the present invention is provided；
Fig. 4 is provided by the embodiment of the present invention in adjustment task groups CPU and the schematic diagram of memory node；
The structured flowchart of the server task dispatching device that Fig. 5 is provided by the embodiment of the present invention.
Specific embodiment
In order that those skilled in the art more fully understand the present invention program, with reference to the accompanying drawings and detailed description
The present invention is described in further detail.Obviously, described embodiment is only a part of embodiment of the invention, rather than
Whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art are not making creative work premise
Lower obtained every other embodiment, belongs to the scope of protection of the invention.
Refer to Fig. 1, a kind of specific embodiment party of the server task dispatching method that Fig. 1 is provided by the embodiment of the present invention
The schematic flow sheet of formula, the method is comprised the following steps：
Step 101：According to the type of task, the task is distributed to pre-binding task groups, wherein, the pre-binding is appointed
Business group comprising the first predetermined number CPU, and with the internal memory group binding of the memory node comprising the second predetermined number；
It should be noted that the type of above-mentioned task can refer to the task class divided according to task characteristic in server
Not.Computation-intensive task is one kind of server task, i.e. CPU intensive type task, the CPU usage of the task of the type compared with
Height, for example, the task such as calculating, logic judgment；I/O intensive tasks are also one kind of server task, are appointed the type is performed
During business, CPU is essentially in wait state, for example, frequently network transmission, the reading task such as hard disk and other I/O equipment.
More specifically, being divided into the task of server, CPU intensive type task, I/O are intensive and committed memory is provided
Task, I/O intensity and the few task of committed memory resource and operating system internal task more than source.Certainly, service is had a high regard for
Business can also have other types, be not limited thereto.
Above-mentioned pre-binding task groups can refer to the task groups with the binding in advance of above-mentioned task, i.e., according to the class of above-mentioned task
Type, is distributed to corresponding task groups and goes to perform.
Because the type of above-mentioned task can have various, thus above-mentioned pre-binding task groups be likely to it is various, for example, when clothes
Business have a high regard for business for CPU intensive type task, I/O are intensive and committed memory resource more than task, I/O is intensive and committed memory is provided
When the few task in source and operating system internal task, correspondingly, above-mentioned pre-binding task groups can be CPU intensive type task
Task groups, I/O intensities and the few task groups of committed memory resource and behaviour more than group, I/O intensities and committed memory resource
Make internal system task groups.
It is understood that according to the hardware resource of the required by task for performing each type, in above-mentioned pre-binding task groups
There is CPU, and binding has respective numbers memory node.CPU quantity and memory node in each task groups, can be by task groups
The number of performed required by task hardware resource is determined.For example, for CPU intensive type task, corresponding CPU intensive
The quantity of the CPU of type task groups is more；And for the task more than I/O intensities and committed memory resource, corresponding I/O is close
The internal memory required for task groups more than collection type and committed memory resource is more, therefore the quantity of its memory node can correspondingly increase
Plus.
Therefore the numerical value of above-mentioned first predetermined number and the second predetermined number can be determined by the type of its task groups, example
Such as, when above-mentioned pre-binding task groups are CPU intensive type task groups, the numerical value of the first predetermined number can be 4, the second present count
The numerical value of amount can be 2.The type of task is different, and the value of the first predetermined number and the second predetermined number also has accordingly not
Together, i.e. the type of pre-binding task groups is different, and the value of the first predetermined number and the second predetermined number also has corresponding difference.
In order to preferably introduce the CPU numbers and memory node of each task groups, it is introduced below in conjunction with Fig. 2, Fig. 2
By the CPU and internal memory division schematic diagram of the task groups that the embodiment of the present invention is provided.
As shown in Fig. 2 task groups can include that CPU intensive type task groups, I/O are intensive and committed memory resource more than appoint
Business group, I/O intensities and the few task groups of committed memory resource and operating system internal task group.CPU intensive type task groups
Inside there are 4 CPU, respectively CPU0, CPU1, CPU2 and CPU3, there is the memory group 0 including memory node node0 and node1；
I/O is intensive and committed memory resource more than task groups in have 2 CPU, a respectively CPU4 and CPU5, and with including 4 internal memories
The memory group 1 of node is bound, respectively node2, node3, node4 and node5；I/O is intensive and committed memory resource is few
Task groups have 2 CPU, respectively CPU6 and CPU7, and bound with including 2 memory groups of memory node 2, respectively
Node6 and node7；Operating system internal task group, there is 2 CPU, is not CPU8 and CPU9, and with including 2 memory nodes
Memory group 3 is bound, respectively node8 and node9.It should be evident that the division of the CPU and memory node in task groups is not limited
Situation proposed in above.
It should be noted that the internal storage structure of server can have in multiple with NUMA internal storage structures, NUMA internal storage structures
Node (Node) is deposited, node can be divided into local node (Local Node), neighbor node (Neighbour Node) and distal end section
Point (Remote Node) three types, each node can possess multiple CPU (each CPU can have multiple cores or thread),
Total Memory Controller Hub is used in node.The speed that CPU accesses different type node memory is differed, usually, this
Ground node>Neighbor node>Distant-end node.
It is understood that need to carry out the binding of task and task groups before being scheduled task, therefore in the present invention
Some embodiments in, the binding flow of task and task groups can be specially：According to the task type of required execution, create many
Individual task groups；According to needed for the hardware configuration of server and each described task groups, by the CPU of respective numbers and each institute
Task groups are stated to be bound；Needed for internal memory according to each task groups, by described in the memory node of respective numbers and each
Task groups are bound；According to the characteristic of each task groups, corresponding task scheduling algorithm is chosen；Creating the task
When, according to the type of the task, the task is bound with corresponding task groups, obtain the pre-binding task groups.
It should be noted that being that task groups are divided, it is necessary to created to task groups before task and task group binding
The internal memory and CPU quantity of respective numbers.
According to the type of task, the task groups for performing corresponding task type are created, after task groups are created, can be according to clothes
The hardware configuration of business device, corresponding hardware resource is distributed to different task groups, as each task group assignment respective numbers
CPU and memory node.
The binding procedure of task and task groups can occur in task creation, you can with the class according to being created for task
Type difference, the task type being tied to performed by corresponding task groups, i.e. task groups and the task type for being created
Unanimously.
In order to preferably introduce the task of each type and the binding procedure of different task groups, enter below in conjunction with Fig. 3
Row is introduced, and Fig. 3 is tied to the schematic diagram of different task group by different types of task that the embodiment of the present invention is provided.
It should be noted that Fig. 3 is based on Fig. 2, therefore the CPU numbers included by each task groups are no longer discussed in detail
And memory node, specifically may refer to the detailed description of figure 2 above.
As shown in figure 3, Task a, Task b, Task c ... and CPU intensive type task groups are bound, Task a,
Task b, Task c ... are CPU intensive type task；By Task m, Task n, Task l ... and I/O is intensive and takes interior
Deposit the task groups more than resource to be bound, Task m, Task n, Task l ... are I/O intensities and committed memory resource is more
Task；By Task x, Task y, Task z ... and I/O is intensive and the few task groups of committed memory resource are bound,
Task x, Task y, Task z ... be that I/O is intensive and committed memory resource more than task；By Task p, Task q,
Task r ... are bound with operating system internal task group, and Task p, Task q, Task r ... are operating system inside and appoint
Business.
Step 102：Using preset task dispatching algorithm corresponding with the pre-binding task groups, the task is adjusted
Degree.
It should be noted that above-mentioned preset task dispatching algorithm can refer to the task type according to performed by task groups and
The dispatching algorithm of selection.Task scheduling algorithm has a lot, such as SCHED_FIFO dispatching algorithms and SCHED_RR dispatching algorithms.
According to the difference of above-mentioned pre-binding task groups, its task scheduling algorithm is also different, correspondingly task scheduling process
It is different.
For example, when task groups are CPU intensive type task groups, its corresponding task scheduling algorithm can be adjusted for SCHED_FIFO
Degree algorithm.The scheduling process of its task can be specially：Number of tasks less than in task groups total check figure (assuming that there is 4 CPU,
24 cores of each CPU, total check figure 4*24=96) when, each task binds a core, and task can always run to task and terminate,
Or task actively concedes CPU, i.e., task switching is not carried out during task；And work as number of tasks more than the total check figure in task groups
When, priority task high can seize the task of low priority.
When task groups be that I/O is intensive and committed memory resource more than task groups, its corresponding task scheduling algorithm can be with
It is SCHED_RR dispatching algorithms.Task in the task groups is performed in turn, i.e., one time of timeslice of each tasks carrying is just
Switch, meanwhile, priority task high can be with the low task of preempting priority.Can also be calculated according to task priority
During timeslice, the task relatively long for some timeslices, it is not necessary to too frequent switching for example reads substantial amounts of data from disk
Database data, is difficult to complete because hard disk speed is slow, in the short time, and this subtask is switched out, it is possible to which the time waited slightly length is again
Switch back.
When task groups are that I/O is intensive and the few task groups of committed memory resource, its corresponding task scheduling algorithm can be with
It is SCHED_RR dispatching algorithms.Task in the task groups is performed in turn, i.e., one time of timeslice of each tasks carrying is just
Switch.Meanwhile, priority task high can be with the low task of preempting priority.Because the corresponding task of the task groups may
It is the operation with user mutual, such as operates terminal, therefore need frequency swap-in row task to switch, strengthens Consumer's Experience.
When task groups are for and for operating system internal task group, its corresponding task scheduling algorithm can be SCHED_
OTHER timesharing dispatching algorithms.The task of the task groups is usually the task that maintenance system is normally run, and these task majorities are
Periodically performing for task, such as refresh disk cache and safeguard network linking etc..Therefore can be appointed using corresponding
Business dispatching algorithm, respective timeslice is calculated according to task priority.
It should be evident that the task priority in each task groups can carry out real-time adjustment according to the state of task,
Task priority can be adjusted, so that task can be performed in time, it is to avoid some priorities of task according to the state of current task
Level is low cannot to be performed forever.
In order to ensure the timely execution of the task in each task groups, can in the heavier loads in certain task groups,
Corresponding hardware resource is called from other task groups, to treat many tasks, after task excessive accordingly has been processed, also
The hardware resource that will can be called before is returned to corresponding task groups.
Therefore in some embodiments of the invention, the process of the hardware resource in dynamic increase and decrease task groups can be specially：
The CPU usage and memory node occupancy in the pre-binding task groups are obtained using monitor task；Accounted for according to the CPU
With rate and memory node occupancy, the number of the dynamic CPU and the memory node for increasing and decreasing the pre-binding task groups
Amount.
It should be noted that the task type of above-mentioned monitor task can be operating system internal task, i.e. monitor task
Can distribute to operating system internal task group and perform.
When may be appreciated, when CPU usage is too high, show the heavier loads of the CPU of current task group, can be corresponding
Ground increases a number of CPU.And when memory node occupancy is too high, show current task group memory node load compared with
Weight, can correspondingly increase a number of memory node.Correspondingly, when CPU usage or memory node occupancy are relatively low
When, it can correspondingly reduce a number of CPU or memory node.
The CPU or memory node that are called from other task groups are returned to phase by the number for reducing CPU or memory node
The heavier loads of the task groups answered, i.e. task groups from other task groups, it is necessary to call a number of CPU or internal memory section
Point, after a period of time, the load of task groups declines, then the CPU or memory node that will be called before, returns to and is called
CPU or the corresponding task groups of memory node.
The number for reducing CPU or memory node may also mean that itself CPU or memory node to other task groups
Call, i.e. the light load of current task group, and the heavier loads of other task groups, other task groups can be from current task group
Call a number of CPU and memory node.
In order to preferably introduce the increase process of explanation CPU or memory node, explanation is introduced below in conjunction with Fig. 4,
Fig. 4 is provided by the embodiment of the present invention in adjustment task groups CPU and the schematic diagram of memory node.
It should be noted that Fig. 4 is based on Fig. 3, thus be no longer discussed in detail each task groups and corresponding CPU and
Memory node, specifically may refer to corresponding contents above.
As shown in figure 4, the present load of CPU intensive type task groups is heavier, i.e., its CPU usage is higher, therefore it is close from I/O
Collection type and the few task groups of committed memory resource call CPU6 to process current task, and now, I/O is intensive and committed memory resource
Then there was only CPU7 in few task groups.And I/O is intensive and committed memory resource more than task groups memory node occupancy
It is higher, therefore it calls a memory node node8 from operating system internal task group, now, operating system internal task is formed a team
The memory group 3 of member then only has memory node node9.
In some embodiments of the invention, according to the CPU usage and memory node occupancy, dynamic increase and decrease institute
The quantity of the CPU and the memory node that state pre-binding task groups can be specially：When the CPU usage be more than or
During equal to the first predetermined threshold value, the CPU of the 3rd predetermined number is called from other task groups；When the memory node occupancy is more than
Or during equal to the second predetermined threshold value, the memory node of the 4th predetermined number is called from other task groups；When the CPU usage is small
When three predetermined threshold values, the CPU that will be called returns to corresponding task groups；When the memory node occupancy is less than the
During four predetermined threshold values, the memory node that will be called returns to corresponding task groups.
It should be noted that the default threshold of above-mentioned first predetermined threshold value, the second predetermined threshold value, the 3rd predetermined threshold value and the 4th
The numerical value of value can arbitrarily set.For example, the first predetermined threshold value can be set as into 90%, the second predetermined threshold value is set as
90%, the 3rd predetermined threshold value is set as 50%, and the 4th predetermined threshold value is set as 50%.
And the numerical values recited of above-mentioned 3rd predetermined number and the 4th predetermined number can also task setting.For example, can be with
3rd predetermined number is set as 1, i.e., calls 1 CPU, the 4th predetermined number to be set as 2 from other task groups, i.e., from other
Business group calls 2 memory nodes.
Number is preset by the way that whether the CPU usage and memory node occupancy of task groups exceed or fall below certain
Value, so that each task groups can reasonably using the hardware resource of server.
The method of the server task scheduling that the embodiment of the present invention is provided, by the type according to task, by above-mentioned
Business is distributed to pre-binding task groups；Using preset task dispatching algorithm corresponding with above-mentioned pre-binding task groups, to above-mentioned task
It is scheduled；Wherein, above-mentioned pre-binding task groups comprising the first predetermined number CPU, and with comprising in the second predetermined number
Deposit the internal memory group binding of node.Different types of task is assigned to different task groups, and there are respective numbers in task groups
CPU and memory node, will CPU and memory node be grouped, to perform different types of task, namely in task scheduling
When, not only allow for the characteristic of task, it is also considered that the CPU of server and the ardware feature of memory node.In task scheduling,
In view of the ardware feature of server, it is possible to reduce the performance loss that task switching brings, and then improve the resource of server
Utilization rate, improves server performance.
Server task dispatching device provided in an embodiment of the present invention is introduced below, service described below is had a high regard for
Business dispatching device can be mutually to should refer to above-described server task dispatching method.
The structured flowchart of the server task dispatching device that Fig. 5 is provided by the embodiment of the present invention, the service of reference picture 5 is had a high regard for
The device for being engaged in dispatching can include：
Distribute module 501, for the type according to task, the task is distributed to pre-binding task groups；
Scheduler module 502, for utilizing preset task dispatching algorithm corresponding with the pre-binding task groups, to described
Business is scheduled；
Wherein, the pre-binding task groups comprising the first predetermined number CPU, and with the internal memory comprising the second predetermined number
The internal memory group binding of node.
Alternatively, also include：
Monitoring module, for obtaining CPU usage and internal memory section in the pre-binding task groups using monitor task
Point occupancy；
Dynamic swap modules, for according to the CPU usage and memory node occupancy, dynamic increase and decrease is described pre- to tie up
Determine the CPU of task groups and the quantity of the memory node.
Alternatively, the dynamic swap modules include：
CPU adding units, for when the CPU usage is more than or equal to the first predetermined threshold value, from other task groups
Call the CPU of the 3rd predetermined number；
Memory node adding unit, for when the memory node occupancy be more than or equal to the second predetermined threshold value when, from
Other task groups call the memory node of the 4th predetermined number；
CPU reduces unit, for when the CPU usage is less than three predetermined threshold values, the CPU that will be called to be returned
To corresponding task groups；
Memory node adding unit, for when the memory node occupancy is less than four predetermined threshold values, by what is called
The memory node returns to corresponding task groups.
Alternatively, also include：
Creation module, for the task type performed according to needed for, creates multiple tasks group；
CPU binding modules, for according to needed for the hardware configuration of server and each described task groups, by respective counts
The CPU of amount is bound with task groups each described；
Internal memory binding module, needed for for the internal memory according to each task groups, by the memory node of respective numbers with
Each described task groups is bound；
Dispatching algorithm chosen module, for the characteristic according to each task groups, chooses corresponding task scheduling algorithm；
Task binding module, for when the task is created, according to the type of the task, by the task with it is corresponding
Task groups bound, obtain the pre-binding task groups.
Alternatively, the task groups include：Task more than CPU intensive type task groups, I/O intensities and committed memory resource
Group, I/O intensities and the few task groups of committed memory resource and operating system internal task group.
The device of the server task scheduling that the embodiment of the present invention is provided, different types of task is assigned to different
Task groups, and have the CPU and memory node of respective numbers in task groups, will CPU and memory node be grouped, to perform not
The task of same type, namely in task scheduling, not only allow for the characteristic of task, it is also considered that the CPU and internal memory of server
The ardware feature of node.In task scheduling, it is contemplated that the ardware feature of server, it is possible to reduce the performance that task switching brings
Loss, and then the resource utilization of server is improve, improve server performance.
Each embodiment is described by the way of progressive in this specification, and what each embodiment was stressed is and other
The difference of embodiment, between each embodiment same or similar part mutually referring to.For being filled disclosed in embodiment
For putting, because it is corresponded to the method disclosed in Example, so description is fairly simple, related part is referring to method part
Illustrate.
Professional further appreciates that, with reference to the unit of each example of the embodiments described herein description
And algorithm steps, can be realized with electronic hardware, computer software or the combination of the two, in order to clearly demonstrate hardware and
The interchangeability of software, generally describes the composition and step of each example according to function in the above description.These
Function is performed with hardware or software mode actually, depending on the application-specific and design constraint of technical scheme.Specialty
Technical staff can realize described function to each specific application using distinct methods, but this realization should not
Think beyond the scope of this invention.
The step of method or algorithm for being described with reference to the embodiments described herein, directly can be held with hardware, processor
Capable software module, or the two combination is implemented.Software module can be placed in random access memory (RAM), internal memory, read-only deposit
Reservoir (ROM), electrically programmable ROM, electrically erasable ROM, register, hard disk, moveable magnetic disc, CD-ROM or technology
In field in known any other form of storage medium.
The method and device to server task provided by the present invention scheduling are described in detail above.Herein
Apply specific case to be set forth principle of the invention and implementation method, the explanation of above example is only intended to help
Understand the method for the present invention and its core concept.It should be pointed out that for those skilled in the art, not taking off
On the premise of the principle of the invention, some improvement and modification can also be carried out to the present invention, these are improved and modification also falls into this
In invention scope of the claims.
Claims (10)
1. a kind of method that server task is dispatched, it is characterised in that including：
According to the type of task, the task is distributed to pre-binding task groups；
Using preset task dispatching algorithm corresponding with the pre-binding task groups, the task is scheduled；
Wherein, the pre-binding task groups comprising the first predetermined number CPU, and with the memory node comprising the second predetermined number
Internal memory group binding.
2. the method for claim 1, it is characterised in that described using corresponding with the pre-binding task groups default
Task scheduling algorithm, also includes after being scheduled to the task：
The CPU usage and memory node occupancy in the pre-binding task groups are obtained using monitor task；
According to the CPU usage and memory node occupancy, the CPU of the dynamic increase and decrease pre-binding task groups and
The quantity of the memory node.
3. method as claimed in claim 2, it is characterised in that described to be taken according to the CPU usage and memory node
Rate, the CPU of the dynamic increase and decrease pre-binding task groups and the quantity of the memory node include：
When the CPU usage is more than or equal to the first predetermined threshold value, the 3rd predetermined number is called from other task groups
CPU；
When the memory node occupancy is more than or equal to the second predetermined threshold value, the 4th predetermined number is called from other task groups
Memory node；
When the CPU usage is less than three predetermined threshold values, the CPU that will be called returns to corresponding task groups；
When the memory node occupancy is less than four predetermined threshold values, the memory node that will be called returns to corresponding
Business group.
4. the method as described in any one of claims 1 to 3, it is characterised in that the task and the pre-binding task groups
Binding procedure is specially：
According to the task type of required execution, multiple tasks group is created；
According to needed for the hardware configuration of server and each described task groups, by the CPU of respective numbers and each described
Business group is bound；
Needed for internal memory according to each task groups, the memory node of respective numbers is tied up with task groups each described
It is fixed；
According to the characteristic of each task groups, corresponding task scheduling algorithm is chosen；
When the task is created, according to the type of the task, the task is bound with corresponding task groups, obtained
The pre-binding task groups.
5. method as claimed in claim 4, it is characterised in that the task groups include：CPU intensive type task groups, I/O are intensive
Inside task groups, I/O intensity and the few task groups of committed memory resource and operating system more than type and committed memory resource
Task groups.
6. the device that a kind of server task is dispatched, it is characterised in that including：
Distribute module, for the type according to task, the task is distributed to pre-binding task groups；
Scheduler module, for utilizing preset task dispatching algorithm corresponding with the pre-binding task groups, is carried out to the task
Scheduling；
Wherein, the pre-binding task groups comprising the first predetermined number CPU, and with the memory node comprising the second predetermined number
Internal memory group binding.
7. device as claimed in claim 6, it is characterised in that also include：
Monitoring module, is accounted for for being obtained CPU usage and memory node in the pre-binding task groups using monitor task
With rate；
Dynamic swap modules, for according to the CPU usage and memory node occupancy, the dynamic increase and decrease pre-binding to be appointed
The CPU of business group and the quantity of the memory node.
8. device as claimed in claim 7, it is characterised in that the dynamic swap modules include：
CPU adding units, for when the CPU usage is more than or equal to the first predetermined threshold value, being called from other task groups
The CPU of the 3rd predetermined number；
Memory node adding unit, for when the memory node occupancy is more than or equal to the second predetermined threshold value, from other
Task groups call the memory node of the 4th predetermined number；
CPU reduces unit, for when the CPU usage is less than three predetermined threshold values, it is right that the CPU that will be called is returned to
The task groups answered；
Memory node adding unit, for when the memory node occupancy is less than four predetermined threshold values, described in calling
Memory node returns to corresponding task groups.
9. the device as described in any one of claim 6 to 8, it is characterised in that also include：
Creation module, for the task type performed according to needed for, creates multiple tasks group；
CPU binding modules, for according to needed for the hardware configuration of server and each described task groups, by respective numbers
CPU is bound with task groups each described；
Internal memory binding module, needed for for the internal memory according to each task groups, by the memory node of respective numbers and each
The task groups are bound；
Dispatching algorithm chosen module, for the characteristic according to each task groups, chooses corresponding task scheduling algorithm；
Task binding module, for when the task is created, according to the type of the task, the task being appointed with corresponding
Business group is bound, and obtains the pre-binding task groups.
10. device as claimed in claim 9, it is characterised in that the task groups include：CPU intensive type task groups, I/O are close
In task groups, I/O intensity and the few task groups of committed memory resource and operating system more than collection type and committed memory resource
Portion's task groups.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710107912.2A CN106897132A (en) | 2017-02-27 | 2017-02-27 | The method and device of a kind of server task scheduling |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710107912.2A CN106897132A (en) | 2017-02-27 | 2017-02-27 | The method and device of a kind of server task scheduling |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106897132A true CN106897132A (en) | 2017-06-27 |
Family
ID=59184214
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710107912.2A Pending CN106897132A (en) | 2017-02-27 | 2017-02-27 | The method and device of a kind of server task scheduling |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106897132A (en) |
Cited By (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107273202A (en) * | 2017-07-14 | 2017-10-20 | 郑州云海信息技术有限公司 | A kind of method and apparatus of control task scheduling |
| CN107656805A (en) * | 2017-10-20 | 2018-02-02 | 广东电网有限责任公司电力调度控制中心 | A kind of electric power data job scheduling method based on Hadoop platform |
| CN107678860A (en) * | 2017-10-13 | 2018-02-09 | 郑州云海信息技术有限公司 | A kind of optimization method and system of KVM virtual machines CPU scheduling strategies |
| CN108021453A (en) * | 2017-12-22 | 2018-05-11 | 联想(北京)有限公司 | A kind of computing resource optimization method, device and server cluster |
| 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 |
| CN108227901A (en) * | 2017-12-30 | 2018-06-29 | 广东技术师范学院 | It is a kind of based on the disk power consumption state switching method sporadicly asked |
| CN109144718A (en) * | 2018-07-06 | 2019-01-04 | 北京比特大陆科技有限公司 | A kind of memory allocation method, memory release method and relevant device |
| CN109144716A (en) * | 2017-06-28 | 2019-01-04 | 中兴通讯股份有限公司 | Operating system dispatching method and device, equipment based on machine learning |
| CN109408215A (en) * | 2018-11-07 | 2019-03-01 | 郑州云海信息技术有限公司 | A kind of method for scheduling task and device of calculate node |
| CN109471724A (en) * | 2018-10-23 | 2019-03-15 | 郑州云海信息技术有限公司 | A kind of method and system of switch operating amount of memory |
| CN109471707A (en) * | 2018-10-12 | 2019-03-15 | 传化智联股份有限公司 | The dispositions method and device of scheduler task |
| CN109558245A (en) * | 2018-12-03 | 2019-04-02 | 群蜂信息技术(上海)有限公司 | A kind of method for processing business based on microserver framework, device and server |
| CN109600396A (en) * | 2019-01-23 | 2019-04-09 | 浙江安点科技有限责任公司 | Realize the firewall system of highly reliable safety |
| CN109725996A (en) * | 2018-06-15 | 2019-05-07 | 中国平安人寿保险股份有限公司 | Mission mode selection and task executing method, device, equipment and storage medium |
| CN109814985A (en) * | 2017-11-20 | 2019-05-28 | 杭州华为数字技术有限公司 | A kind of method for scheduling task and scheduler calculate equipment, system |
| CN110111458A (en) * | 2019-04-16 | 2019-08-09 | 深圳市多度科技有限公司 | Realize method, system and door access machine that face is deployed to ensure effective monitoring and control of illegal activities |
| CN110750341A (en) * | 2018-07-24 | 2020-02-04 | 深圳市优必选科技有限公司 | Task scheduling method, device, system, terminal equipment and storage medium |
| CN110764981A (en) * | 2019-09-09 | 2020-02-07 | 北京新数科技有限公司 | Integrated operation and maintenance management method and device for IT machine room |
| WO2020063040A1 (en) * | 2018-09-26 | 2020-04-02 | Oppo广东移动通信有限公司 | Processor core configuration method and apparatus, terminal, and storage medium |
| CN112513814A (en) * | 2020-04-01 | 2021-03-16 | 华为技术有限公司 | Task scheduling method and device |
| CN112667380A (en) * | 2020-12-30 | 2021-04-16 | 珠海亿智电子科技有限公司 | Multiprocessor task scheduling method, device and storage medium |
| CN112817728A (en) * | 2021-02-20 | 2021-05-18 | 咪咕音乐有限公司 | Task scheduling method, network device and storage medium |
| CN113032137A (en) * | 2019-12-25 | 2021-06-25 | 中科寒武纪科技股份有限公司 | Task allocation method and device, computer equipment and readable storage medium |
| CN113312166A (en) * | 2021-07-29 | 2021-08-27 | 阿里云计算有限公司 | Resource processing method and device |
| CN113641482A (en) * | 2021-08-31 | 2021-11-12 | 联通(广东)产业互联网有限公司 | AI algorithm off-line scheduling method, system, computer equipment and storage medium |
| CN117707738A (en) * | 2024-01-10 | 2024-03-15 | 苏州市欧冶半导体有限公司 | Task processing method, system, device, computer equipment and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104181510A (en) * | 2014-09-05 | 2014-12-03 | 电子科技大学 | Parallel radar pulse compression method |
| CN105141541A (en) * | 2015-09-23 | 2015-12-09 | 浪潮(北京)电子信息产业有限公司 | Task-based dynamic load balancing scheduling method and device |
| CN105302638A (en) * | 2015-11-04 | 2016-02-03 | 国家计算机网络与信息安全管理中心 | MPP (Massively Parallel Processing) cluster task scheduling method based on system load |
| CN105808328A (en) * | 2014-12-31 | 2016-07-27 | 杭州华为数字技术有限公司 | Task scheduling method, device and system |
-
2017
- 2017-02-27 CN CN201710107912.2A patent/CN106897132A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104181510A (en) * | 2014-09-05 | 2014-12-03 | 电子科技大学 | Parallel radar pulse compression method |
| CN105808328A (en) * | 2014-12-31 | 2016-07-27 | 杭州华为数字技术有限公司 | Task scheduling method, device and system |
| CN105141541A (en) * | 2015-09-23 | 2015-12-09 | 浪潮(北京)电子信息产业有限公司 | Task-based dynamic load balancing scheduling method and device |
| CN105302638A (en) * | 2015-11-04 | 2016-02-03 | 国家计算机网络与信息安全管理中心 | MPP (Massively Parallel Processing) cluster task scheduling method based on system load |
Cited By (34)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109144716A (en) * | 2017-06-28 | 2019-01-04 | 中兴通讯股份有限公司 | Operating system dispatching method and device, equipment based on machine learning |
| CN107273202A (en) * | 2017-07-14 | 2017-10-20 | 郑州云海信息技术有限公司 | A kind of method and apparatus of control task scheduling |
| CN107678860A (en) * | 2017-10-13 | 2018-02-09 | 郑州云海信息技术有限公司 | A kind of optimization method and system of KVM virtual machines CPU scheduling strategies |
| CN107656805A (en) * | 2017-10-20 | 2018-02-02 | 广东电网有限责任公司电力调度控制中心 | A kind of electric power data job scheduling method based on Hadoop platform |
| CN109814985A (en) * | 2017-11-20 | 2019-05-28 | 杭州华为数字技术有限公司 | A kind of method for scheduling task and scheduler calculate equipment, system |
| CN109814985B (en) * | 2017-11-20 | 2021-07-16 | 华为技术有限公司 | Task scheduling method and scheduler, computing device and system |
| CN108021453A (en) * | 2017-12-22 | 2018-05-11 | 联想(北京)有限公司 | A kind of computing resource optimization method, device and server cluster |
| 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 |
| CN108227901A (en) * | 2017-12-30 | 2018-06-29 | 广东技术师范学院 | It is a kind of based on the disk power consumption state switching method sporadicly asked |
| CN108227901B (en) * | 2017-12-30 | 2020-07-14 | 广东技术师范大学 | Disk power consumption state switching method based on sporadic request |
| CN109725996B (en) * | 2018-06-15 | 2023-07-21 | 中国平安人寿保险股份有限公司 | Task mode selection and task execution method, device, equipment and storage medium |
| CN109725996A (en) * | 2018-06-15 | 2019-05-07 | 中国平安人寿保险股份有限公司 | Mission mode selection and task executing method, device, equipment and storage medium |
| CN109144718A (en) * | 2018-07-06 | 2019-01-04 | 北京比特大陆科技有限公司 | A kind of memory allocation method, memory release method and relevant device |
| CN110750341A (en) * | 2018-07-24 | 2020-02-04 | 深圳市优必选科技有限公司 | Task scheduling method, device, system, terminal equipment and storage medium |
| CN110750341B (en) * | 2018-07-24 | 2022-08-02 | 深圳市优必选科技有限公司 | Task scheduling method, device, system, terminal equipment and storage medium |
| WO2020063040A1 (en) * | 2018-09-26 | 2020-04-02 | Oppo广东移动通信有限公司 | Processor core configuration method and apparatus, terminal, and storage medium |
| CN109471707A (en) * | 2018-10-12 | 2019-03-15 | 传化智联股份有限公司 | The dispositions method and device of scheduler task |
| CN109471724A (en) * | 2018-10-23 | 2019-03-15 | 郑州云海信息技术有限公司 | A kind of method and system of switch operating amount of memory |
| CN109408215A (en) * | 2018-11-07 | 2019-03-01 | 郑州云海信息技术有限公司 | A kind of method for scheduling task and device of calculate node |
| CN109408215B (en) * | 2018-11-07 | 2021-10-01 | 郑州云海信息技术有限公司 | Task scheduling method and device for computing node |
| CN109558245A (en) * | 2018-12-03 | 2019-04-02 | 群蜂信息技术(上海)有限公司 | A kind of method for processing business based on microserver framework, device and server |
| CN109600396A (en) * | 2019-01-23 | 2019-04-09 | 浙江安点科技有限责任公司 | Realize the firewall system of highly reliable safety |
| CN110111458A (en) * | 2019-04-16 | 2019-08-09 | 深圳市多度科技有限公司 | Realize method, system and door access machine that face is deployed to ensure effective monitoring and control of illegal activities |
| CN110764981A (en) * | 2019-09-09 | 2020-02-07 | 北京新数科技有限公司 | Integrated operation and maintenance management method and device for IT machine room |
| CN113032137A (en) * | 2019-12-25 | 2021-06-25 | 中科寒武纪科技股份有限公司 | Task allocation method and device, computer equipment and readable storage medium |
| WO2021196074A1 (en) * | 2020-04-01 | 2021-10-07 | 华为技术有限公司 | Task scheduling method and apparatus |
| CN112513814A (en) * | 2020-04-01 | 2021-03-16 | 华为技术有限公司 | Task scheduling method and device |
| CN112667380A (en) * | 2020-12-30 | 2021-04-16 | 珠海亿智电子科技有限公司 | Multiprocessor task scheduling method, device and storage medium |
| CN112817728A (en) * | 2021-02-20 | 2021-05-18 | 咪咕音乐有限公司 | Task scheduling method, network device and storage medium |
| CN112817728B (en) * | 2021-02-20 | 2024-05-24 | 咪咕音乐有限公司 | Task scheduling method, network device and storage medium |
| CN113312166A (en) * | 2021-07-29 | 2021-08-27 | 阿里云计算有限公司 | Resource processing method and device |
| CN113641482A (en) * | 2021-08-31 | 2021-11-12 | 联通(广东)产业互联网有限公司 | AI algorithm off-line scheduling method, system, computer equipment and storage medium |
| CN113641482B (en) * | 2021-08-31 | 2024-03-22 | 联通(广东)产业互联网有限公司 | AI algorithm offline scheduling method, system, computer equipment and storage medium |
| CN117707738A (en) * | 2024-01-10 | 2024-03-15 | 苏州市欧冶半导体有限公司 | Task processing method, system, device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106897132A (en) | The method and device of a kind of server task scheduling | |
| CN104580396B (en) | A kind of method for scheduling task, node and system | |
| Sharma et al. | Performance analysis of load balancing algorithms | |
| CN106293950B (en) | A kind of resource optimization management method towards group system | |
| CN111078363B (en) | NUMA node scheduling method, device, equipment and medium of virtual machine | |
| CN110113387A (en) | A kind of processing method based on distributed batch processing system, apparatus and system | |
| CN112272203B (en) | Cluster service node selection method, system, terminal and storage medium | |
| CN111796908B (en) | System and method for automatic elastic expansion and contraction of resources and cloud platform | |
| JP2011501268A (en) | Hierarchical reserved resource scheduling infrastructure | |
| CN102282543A (en) | Information processing device and information processing method | |
| Tantalaki et al. | Pipeline-based linear scheduling of big data streams in the cloud | |
| CN111104227B (en) | Resource control method and device of K8s platform and related components | |
| CN104239154B (en) | Job scheduling method and job scheduler in a kind of Hadoop clusters | |
| US8352950B2 (en) | Algorithm to share physical processors to maximize processor cache usage and topologies | |
| CN113032102B (en) | Resource rescheduling method, device, equipment and medium | |
| CN108376103A (en) | A kind of the equilibrium of stock control method and server of cloud platform | |
| Biswas et al. | An auto-scaling framework for controlling enterprise resources on clouds | |
| CN109151070A (en) | Service scheduling method, the electronic device of point-to-point CDN based on block chain | |
| CN113467933A (en) | Thread pool optimization method, system, terminal and storage medium for distributed file system | |
| CN111159859B (en) | Cloud container cluster deployment method and system | |
| CN113672391A (en) | Parallel computing task scheduling method and system based on Kubernetes | |
| Yazdanov et al. | EHadoop: Network I/O aware scheduler for elastic MapReduce cluster | |
| US10108351B2 (en) | Reallocate memory pending queue based on stall | |
| CN109086128B (en) | Task scheduling method and device | |
| Carrega et al. | Energy-aware consolidation scheme for data center cloud applications |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| 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: 20170627 |