CN107133106A - A kind of NUMA system optimization method and device - Google Patents
A kind of NUMA system optimization method and device Download PDFInfo
- Publication number
- CN107133106A CN107133106A CN201710379373.8A CN201710379373A CN107133106A CN 107133106 A CN107133106 A CN 107133106A CN 201710379373 A CN201710379373 A CN 201710379373A CN 107133106 A CN107133106 A CN 107133106A
- Authority
- CN
- China
- Prior art keywords
- task
- numa
- cpu frequency
- node
- numa system
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—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
- G06F9/505—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 load
Abstract
The invention discloses a kind of NUMA system optimization method, this method comprises the following steps：Within the objective optimization cycle, the running status of each task on each NUMA node of NUMA system in a upper optimization cycle is obtained；For each task, according to the running status of the task, it is determined that whether the cpu frequency resource for distributing to the task is more than the minimum cpu frequency demand of the task；If it is, the cpu frequency resource for distributing to the task is adjusted to the minimum cpu frequency demand for only meeting the task.The technical scheme provided using the embodiment of the present invention, part cpu frequency resource can be saved, the cpu frequency resource saved can distribute to other to the more tasks of cpu frequency demand, or for responding the request that performs of more multitask, improve the operating efficiency of NUMA system.Optimize device the invention also discloses a kind of NUMA system, with relevant art effect.
Description
Technical field
The present invention relates to field of computer technology, more particularly to a kind of NUMA system optimization method and device.
Background technology
With the fast development of computer technology, NUMA (Non Uniform memory Access Architecture,
Nonuniform memory access structure) technology is also developed rapidly, meanwhile, the NUMA system based on NUMA technique constructions is also each
Industry is widely applied.
NUMA system includes multiple NUMA nodes, and NUMA node is used for the execution for carrying task.Each task takes NUMA
The initial resource of system distribution, can just discharge corresponding resource after task terminates.
In NUMA system, task in the process of running its demand and NUMA system to resource performance all over time
Change and change., will if the resource for distributing to the task is far longer than the resource requirement of the task for certain task
Cause the wasting of resources, while other may be caused to have the task of more resource requirements not run normally so that NUMA system work
Make efficiency relatively low.
The content of the invention
It is an object of the invention to provide a kind of NUMA system optimization method and device, to save resource, NUMA system is improved
Operating efficiency.
In order to solve the above technical problems, the present invention provides following technical scheme：
A kind of NUMA system optimization method, including：
Within the objective optimization cycle, each task on each NUMA node of NUMA system in the upper optimization cycle of acquisition
Running status；
For each task, according to the running status of the task, it is determined that whether the cpu frequency resource for distributing to the task is big
In the minimum cpu frequency demand of the task；
If it is, the cpu frequency resource for distributing to the task is adjusted to the minimum cpu frequency need for only meeting the task
Ask.
The present invention a kind of embodiment in, in the acquisition in an optimization cycle NUMA system it is each
On NUMA node before the running status of each task, in addition to：
Determine whether the current business performance of the NUMA system is less than default first performance threshold value；
Each appoint on each NUMA node of the NUMA system if it is, being reduced according to default first adjustable strategies
The cpu frequency level of business operation.
In a kind of embodiment of the present invention, described according to the default first adjustable strategies reduction NUMA
On each NUMA node of system after the cpu frequency level of each task run, in addition to：
When the service feature for detecting the NUMA system is more than default second performance threshold value, according to default second
Adjustable strategies heighten the cpu frequency level of each task run on each NUMA node of the NUMA system, described secondary
Energy threshold value is more than or equal to the first performance threshold value.
The present invention a kind of embodiment in, in the acquisition in an optimization cycle NUMA system it is each
On NUMA node before the running status of each task, in addition to：
Within the objective optimization cycle, the total cpu frequency of all tasks currently run in the NUMA system is determined
Demand；
If the ratio that total cpu frequency demand and the NUMA system can provide cpu frequency resource is less than default ratio
Example threshold value, then according to total cpu frequency demand, determine target NUMA node to be closed in the NUMA system；
By on other NUMA nodes of the task scheduling on the target NUMA node to the NUMA system, and close institute
State target NUMA node.
In a kind of embodiment of the present invention, the task scheduling by the target NUMA node is described in
On other nodes of NUMA system, including：
Determine each task on the target NUMA node to the demand degree of cpu frequency resource and the NUMA system its
Demand degree of each task to cpu frequency resource on his NUMA node；
According to demand degree of each task to cpu frequency resource, the task on the target NUMA node is scheduled, with
The task of only setting quantity on same NUMA node is set to be more than the demand degree threshold value of setting to cpu frequency resource requirement degree.
A kind of NUMA system optimizes device, including：
Running status obtains module, within the objective optimization cycle, NUMA system to be every in the upper optimization cycle of acquisition
The running status of each task on individual NUMA node；
Frequency determining module, for for each task, according to the running status of the task, it is determined that distributing to the task
Whether cpu frequency resource is more than the minimum cpu frequency demand of the task, if it is, triggering frequency regulation block；
The frequency regulation block, for the cpu frequency resource for distributing to the task to be adjusted to only meeting the task
Minimum cpu frequency demand.
It is also inclusive to can determine that module in a kind of embodiment of the present invention, it is used for：
In the acquisition in an optimization cycle on each NUMA node of NUMA system each task running status it
Before, determine whether the current business performance of the NUMA system is less than default first performance threshold value；
Each appoint on each NUMA node of the NUMA system if it is, being reduced according to default first adjustable strategies
The cpu frequency level of business operation.
In a kind of embodiment of the present invention, in addition to frequency level heightens module, is used for：
Each task on each NUMA node according to the default first adjustable strategies reduction NUMA system
After the cpu frequency level of operation, when the service feature for detecting the NUMA system is more than default second performance threshold value,
The cpu frequency layer of each task run on each NUMA node of the NUMA system is heightened according to default second adjustable strategies
Level, the second performance threshold value is more than or equal to the first performance threshold value.
In a kind of embodiment of the present invention, in addition to：
Aggregate demand determining module, it is every on each NUMA node of NUMA system in an optimization cycle in the acquisition
Before the running status of individual task, within the objective optimization cycle, all currently run in the NUMA system are determined
Total cpu frequency demand of business；
Closed node determining module is treated, for cpu frequency can be provided in total cpu frequency demand and the NUMA system
When the ratio of resource is less than preset ratio threshold value, according to total cpu frequency demand, determine to be closed in the NUMA system
Target NUMA node；
Closed node processing module is treated, for by the task scheduling on the target NUMA node to the NUMA system
On other NUMA nodes, and close the target NUMA node.
It is described to treat closed node processing module in a kind of embodiment of the present invention, specifically for：
Determine each task on the target NUMA node to the demand degree of cpu frequency resource and the NUMA system its
Demand degree of each task to cpu frequency resource on his NUMA node；
According to demand degree of each task to cpu frequency resource, the task on the target NUMA node is scheduled, with
The task of only setting quantity on same NUMA node is set to be more than the demand degree threshold value of setting to cpu frequency resource requirement degree.
The technical scheme provided using the embodiment of the present invention, within the objective optimization cycle, can obtain an optimization week
The running status of each task on each NUMA node of interim NUMA system, for each task, according to the operation of the task
State, if it is determined that the cpu frequency resource for distributing to the task is more than the minimum cpu frequency demand of the task, then will can divide
The cpu frequency resource of the dispensing task is adjusted to the minimum cpu frequency demand for only meeting the task.Part can so be saved
Cpu frequency resource, the cpu frequency resource saved can distribute to other to the more tasks of cpu frequency demand, or be used for
The operating efficiency for performing request, improving NUMA system of response more multitask.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
There is the accompanying drawing used required in technology description to be briefly described, it should be apparent that, drawings in the following description are only this
Some embodiments of invention, for those of ordinary skill in the art, on the premise of not paying creative work, can be with
Other accompanying drawings are obtained according to these accompanying drawings.
Fig. 1 is a kind of implementing procedure figure of NUMA system optimization method in the embodiment of the present invention；
Fig. 2 is a kind of structural representation of NUMA system optimization device in the embodiment of the present invention.
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.
Shown in Figure 1, a kind of implementing procedure figure of the NUMA system optimization method provided by the embodiment of the present invention should
Method may comprise steps of：
S110：Within the objective optimization cycle, obtain each on each NUMA node of NUMA system in a upper optimization cycle
The running status of task.
In actual applications, NUMA system can be optimized according to the setting cycle, the objective optimization cycle is any one
The individual cycle.
In each optimization cycle, the running status of each task on each NUMA node of NUMA system can be entered
Row monitoring.For some task, the running status of the task can include the resource requirement of the task, distribute to the task
Resource situation, the implementation progress of the task etc..
Within the objective optimization cycle, each NUMA of NUMA system in a upper optimization cycle can be obtained by Monitoring Data
The running status of each task on node.
S120：For each task, according to the running status of the task, it is determined that distributing to the cpu frequency resource of the task
Whether the minimum cpu frequency demand of the task is more than.
In NUMA system, in the process of running, its cpu frequency demand is not each task on each NUMA node
Unalterable, time to time change.It is understood that for certain task, as long as meeting the minimum CPU frequencies of the task
Rate demand, the task just can normally be run.If the cpu frequency resource for distributing to the task is more than the minimum CPU of the task
Frequency needs, it will cause the waste of part cpu frequency resource.
For each task, according to the running status of the task, it may be determined that distributing to the cpu frequency resource of the task is
The no minimum cpu frequency demand more than the task.If it is, step S130 operation can be continued executing with, otherwise, can not
Do any operation.
S130：The cpu frequency resource for distributing to the task is adjusted to the minimum cpu frequency demand for only meeting the task.
For each task, if it is determined that the cpu frequency resource for distributing to the task is more than the minimum cpu frequency of the task
Demand, then can adjust the cpu frequency resource for distributing to the task to the minimum cpu frequency demand for only meeting the task, with
Save a part of cpu frequency resource.It is more to cpu frequency demand that the cpu frequency resource saved can distribute to other
Task, or perform request for response more multitask.
Specifically, cpu frequency resource can be adjusted by adjusting cpu frequency level, i.e., by reducing cpu frequency level
Mode adjusts the cpu frequency resource for distributing to the task to the minimum cpu frequency demand for only meeting the task.
The method provided using the embodiment of the present invention, within the objective optimization cycle, can be obtained in an optimization cycle
The running status of each task on each NUMA node of NUMA system, for each task, according to the running status of the task,
If it is determined that the cpu frequency resource for distributing to the task is more than the minimum cpu frequency demand of the task, then this can will be distributed to
The cpu frequency resource of task is adjusted to the minimum cpu frequency demand for only meeting the task.Part CPU frequencies can so be saved
Rate resource, the cpu frequency resource saved can distribute to other to the more tasks of cpu frequency demand, or for responding
The operating efficiency for performing request, improving NUMA system of more multitask.
In one embodiment of the invention, before step S110, it can also comprise the following steps：
Step one：Determine whether the current business performance of NUMA system is less than default first performance threshold value, if it is,
Perform the operation of step 2；
Step 2：Each task run on each NUMA node of NUMA system is reduced according to default first adjustable strategies
Cpu frequency level.
In embodiments of the present invention, one or more performance thresholds of NUMA system can be preset according to actual conditions
Value.When the service feature of NUMA system is less than default first performance threshold value, it is believed that NUMA system present load intensity
It is larger, to ensure that each task can normally be run in NUMA system, NUMA systems can be reduced according to the first adjustable strategies of setting
The cpu frequency level of each task run on each NUMA node of system.Such as the cpu frequency level of each task run is reduced
One-level.
The service feature time to time change of NUMA system.Carried out as above in the cpu frequency level to each task run
After degradation processing, when the service feature for detecting NUMA system is more than default second performance threshold value, it is believed that NUMA systems
Present load intensity of uniting is smaller.In such a case, it is possible to heighten each of NUMA system according to default second adjustable strategies
The cpu frequency level of each task run on NUMA node.The cpu frequency level of each task run is such as heightened into one-level.
In embodiments of the present invention, second performance threshold value is more than or equal to first performance threshold value, first performance threshold value and the
Two performance threshold can be set and be adjusted according to actual conditions, such as set first performance threshold value as 40%, be set secondary
Energy threshold value is 50%, and the embodiment of the present invention is without limitation.
In actual applications, setting second performance threshold value is more than first performance threshold value, can avoid because to cpu frequency level
Frequently adjustment, took multi-system resource.
According to the magnitude relationship of the service feature of NUMA system and default performance threshold, adjust each in NUMA system
The cpu frequency level of each task run on NUMA node, can be carried out on the basis of NUMA system has preferable service feature
Energy saving optimizing, that is, consider the service feature and resource factors of NUMA system, preferably NUMA system optimized.
In one embodiment of the invention, before step S110, this method can also comprise the following steps：
First step：Within the objective optimization cycle, the total CPU frequencies of all tasks currently run in NUMA system are determined
Rate demand；
Second step：If total cpu frequency demand and NUMA system can provide the ratio of cpu frequency resource less than default
Proportion threshold value, then according to total cpu frequency demand, determine target NUMA node to be closed in NUMA system；
3rd step：By on other NUMA nodes of the task scheduling on target NUMA node to NUMA system, and close
Close target NUMA node.
For ease of description, above three step is combined and illustrated.
In NUMA system, the running status of each task on each NUMA node can be monitored in real time.In objective optimization
In cycle, it may be determined that the total cpu frequency demand for all tasks currently run in NUMA system.If total cpu frequency demand
With NUMA system can provide cpu frequency resource ratio be less than preset ratio threshold value, such as 30%, then may indicate that NUMA system work as
Preceding intensity of load is smaller, considers from Energy Angle, can determine mesh to be closed in NUMA system according to total cpu frequency demand
Mark NUMA node, it is determined that target NUMA node to be closed can have one or more.It is determined that target NUMA sections to be closed
, can be by other NUMA nodes of the task scheduling on target NUMA node to NUMA system after point, and close target NUMA
Node.
, specifically can be by following by other NUMA nodes of the task scheduling on target NUMA node to NUMA system
Step is carried out：
Step one：Determine each task on target NUMA node to the demand degree of cpu frequency resource and NUMA system its
Demand degree of each task to cpu frequency resource on his NUMA node；
Step 2：According to demand degree of each task to cpu frequency resource, the task on target NUMA node is adjusted
Degree, so that the task of only setting quantity is more than the demand degree threshold of setting to cpu frequency resource requirement degree on same NUMA node
Value.
In embodiments of the present invention, each task shows that the demand degree of cpu frequency resource corresponding task is provided to cpu frequency
The desirability in source, if two different tasks are all higher than a certain numerical value to the demand degree of cpu frequency resource, show this two
Individual task has larger demand to cpu frequency resource, if the two tasks are put in same NUMA node, it is to CPU
The competition of frequency resource is larger.
For each task, according to demand of the task to cpu frequency resource, it may be determined that the task is provided to cpu frequency
The demand degree in source.Specifically, the corresponding relation formula of demand and demand degree can be preset, such as when demand is more than default first threshold
During value, its corresponding demand degree is set as the first value, when demand is between first threshold and Second Threshold, its is set corresponding
Demand degree is second value.For example, when demand is more than 90%, its corresponding demand degree is 5, when demand is in 70%-90%
Between when, its corresponding demand degree be 4, the value of demand degree is higher here shows that demand is bigger.Certainly, in actual applications, also
Other corresponding relations can be set, the embodiment of the present invention is without limitation.
It is determined that wait after the target NUMA node closed in NUMA system, it may be determined that each task on target NUMA node
Demand degree to cpu frequency resource and on other NUMA nodes of NUMA system each task to the demand degree of cpu frequency resource,
Task according to only setting quantity on same NUMA node is more than the demand degree threshold value of setting to cpu frequency resource requirement degree
Principle, by the task scheduling on target NUMA node to other NUMA nodes.This way it is possible to avoid producing same NUMA node
On cpu frequency resource competition.
It should be noted that demand degree threshold value can be set and be adjusted according to actual conditions, such as it is set as 3, this hair
Bright embodiment is without limitation.
Corresponding to above method embodiment, the embodiment of the present invention additionally provides a kind of NUMA system optimization device, hereafter
A kind of NUMA system optimization device of description can be mutually to should refer to a kind of above-described NUMA system optimization method.
Shown in Figure 2, the device is included with lower module：
Running status obtains module 210, within the objective optimization cycle, obtaining NUMA system in a upper optimization cycle
The running status of each task on each NUMA node；
Frequency determining module 220, for for each task, according to the running status of the task, it is determined that distributing to this
Whether the cpu frequency resource of business is more than the minimum cpu frequency demand of the task, if it is, triggering frequency regulation block；
Frequency regulation block 230, for the cpu frequency resource for distributing to the task to be adjusted to only meeting the task most
Low cpu frequency demand.
The device provided using the embodiment of the present invention, within the objective optimization cycle, can be obtained in an optimization cycle
The running status of each task on each NUMA node of NUMA system, for each task, according to the running status of the task,
If it is determined that the cpu frequency resource for distributing to the task is more than the minimum cpu frequency demand of the task, then this can will be distributed to
The cpu frequency resource of task is adjusted to the minimum cpu frequency demand for only meeting the task.Part CPU frequencies can so be saved
Rate resource, the cpu frequency resource saved can distribute to other to the more tasks of cpu frequency demand, or for responding
The operating efficiency for performing request, improving NUMA system of more multitask.
It is also inclusive to can determine that module in a kind of embodiment of the present invention, it is used for：
In acquisition in an optimization cycle on each NUMA node of NUMA system before the running status of each task, really
Whether the current business performance for determining NUMA system is less than default first performance threshold value；
Transported if it is, reducing each task on each NUMA node of NUMA system according to default first adjustable strategies
Capable cpu frequency level.
In a kind of embodiment of the present invention, in addition to frequency level heightens module, is used for：
The CPU of each task run on each NUMA node that NUMA system is reduced according to default first adjustable strategies
After frequency level, when the service feature for detecting NUMA system is more than default second performance threshold value, according to default the
Two adjustable strategies heighten the cpu frequency level of each task run on each NUMA node of NUMA system, second performance
Threshold value is more than or equal to the first performance threshold value.
In a kind of embodiment of the present invention, in addition to：
Each appoint on aggregate demand determining module, each NUMA node for NUMA system in an optimization cycle in acquisition
Before the running status of business, within the objective optimization cycle, the total cpu frequency of all tasks currently run in NUMA system is determined
Demand；
Closed node determining module is treated, for the ratio of cpu frequency resource can be provided in total cpu frequency demand and NUMA system
When example is less than preset ratio threshold value, according to total cpu frequency demand, target NUMA node to be closed in NUMA system is determined；
Closed node processing module is treated, for by other NUMA of the task scheduling on target NUMA node to NUMA system
On node, and close target NUMA node.
In a kind of embodiment of the present invention, closed node processing module is treated, specifically for：
Determine that each task is saved to the demand degree of cpu frequency resource and other NUMA of NUMA system on target NUMA node
Demand degree of each task to cpu frequency resource on point；
According to demand degree of each task to cpu frequency resource, the task on target NUMA node is scheduled, so that together
The task of only setting quantity is more than the demand degree threshold value of setting to cpu frequency resource requirement degree on one NUMA node.
The embodiment of each in this specification is described by the way of progressive, what each embodiment was stressed be with it is other
Between the difference of embodiment, 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
Explanation.
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.
Directly it can be held with reference to the step of the method or algorithm that the embodiments described herein is described with hardware, processor
Capable software module, or the two combination are 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 any other form of storage medium well known in field.
Specific case used herein is set forth to the principle and embodiment of the present invention, and above example is said
It is bright to be only intended to help and understand technical scheme and its core concept.It should be pointed out that for the common of the art
For technical staff, under the premise without departing from the principles of the invention, some improvement and modification can also be carried out to the present invention, these
Improve and modification is also fallen into the protection domain of the claims in the present invention.
Claims (10)
1. a kind of NUMA system optimization method, it is characterised in that including：
Within the objective optimization cycle, the operation of each task on each NUMA node of NUMA system in a upper optimization cycle is obtained
State；
For each task, according to the running status of the task, it is somebody's turn to do it is determined that whether the cpu frequency resource for distributing to the task is more than
The minimum cpu frequency demand of task；
If it is, the cpu frequency resource for distributing to the task is adjusted to the minimum cpu frequency demand for only meeting the task.
2. NUMA system optimization method according to claim 1, it is characterised in that in the acquisition in an optimization cycle
On each NUMA node of NUMA system before the running status of each task, in addition to：
Determine whether the current business performance of the NUMA system is less than default first performance threshold value；
Transported if it is, reducing each task on each NUMA node of the NUMA system according to default first adjustable strategies
Capable cpu frequency level.
3. NUMA system optimization method according to claim 2, it is characterised in that described according to the default first adjustment
On each NUMA node of the strategy reduction NUMA system after the cpu frequency level of each task run, in addition to：
When the service feature for detecting the NUMA system is more than default second performance threshold value, according to the default second adjustment
Strategy heightens the cpu frequency level of each task run on each NUMA node of the NUMA system, the second performance threshold
Value is more than or equal to the first performance threshold value.
4. the NUMA system optimization method according to any one of claims 1 to 3, it is characterised in that one in the acquisition
In optimization cycle on each NUMA node of NUMA system before the running status of each task, in addition to：
Within the objective optimization cycle, the total cpu frequency demand of all tasks currently run in the NUMA system is determined；
If the ratio that total cpu frequency demand and the NUMA system can provide cpu frequency resource is less than preset ratio threshold
Value, then according to total cpu frequency demand, determine target NUMA node to be closed in the NUMA system；
By on other NUMA nodes of the task scheduling on the target NUMA node to the NUMA system, and close the mesh
Mark NUMA node.
5. NUMA system optimization method according to claim 4, it is characterised in that described by the target NUMA node
Task scheduling on other nodes of the NUMA system, including：
Determine each task on the target NUMA node to the demand degree of cpu frequency resource and the NUMA system other
Demand degree of each task to cpu frequency resource on NUMA node；
According to demand degree of each task to cpu frequency resource, the task on the target NUMA node is scheduled, so that together
The task of only setting quantity is more than the demand degree threshold value of setting to cpu frequency resource requirement degree on one NUMA node.
6. a kind of NUMA system optimizes device, it is characterised in that including：
Running status obtains module, within the objective optimization cycle, NUMA system to be each in the upper optimization cycle of acquisition
The running status of each task on NUMA node；
Frequency determining module, for for each task, according to the running status of the task, it is determined that distributing to the CPU of the task
Whether frequency resource is more than the minimum cpu frequency demand of the task, if it is, triggering frequency regulation block；
The frequency regulation block, for the cpu frequency resource for distributing to the task to be adjusted to only meeting the minimum of the task
Cpu frequency demand.
7. NUMA system according to claim 6 optimizes device, it is characterised in that also inclusive to can determine that module, it is used for：
In the acquisition in an optimization cycle on each NUMA node of NUMA system before the running status of each task, really
Whether the current business performance of the fixed NUMA system is less than default first performance threshold value；
Transported if it is, reducing each task on each NUMA node of the NUMA system according to default first adjustable strategies
Capable cpu frequency level.
8. NUMA system according to claim 7 optimizes device, it is characterised in that also heighten module including frequency level,
For：
Each task run on each NUMA node according to the default first adjustable strategies reduction NUMA system
Cpu frequency level after, detect the NUMA system service feature be more than default second performance threshold value when, according to
Default second adjustable strategies heighten the cpu frequency level of each task run on each NUMA node of the NUMA system,
The second performance threshold value is more than or equal to the first performance threshold value.
9. the NUMA system optimization device according to any one of claim 6 to 8, it is characterised in that also include：
Each appoint on aggregate demand determining module, each NUMA node for NUMA system in an optimization cycle in the acquisition
Before the running status of business, within the objective optimization cycle, all tasks for determining currently to run in the NUMA system
Total cpu frequency demand；
Closed node determining module is treated, for cpu frequency resource can be provided in total cpu frequency demand and the NUMA system
Ratio be less than preset ratio threshold value when, according to total cpu frequency demand, determine target to be closed in the NUMA system
NUMA node；
Treat closed node processing module, for by the task scheduling on the target NUMA node to the NUMA system other
On NUMA node, and close the target NUMA node.
10. NUMA system according to claim 9 optimizes device, it is characterised in that described to treat closed node processing module,
Specifically for：
Determine each task on the target NUMA node to the demand degree of cpu frequency resource and the NUMA system other
Demand degree of each task to cpu frequency resource on NUMA node；
According to demand degree of each task to cpu frequency resource, the task on the target NUMA node is scheduled, so that together
The task of only setting quantity is more than the demand degree threshold value of setting to cpu frequency resource requirement degree on one NUMA node.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710379373.8A CN107133106A (en) | 2017-05-25 | 2017-05-25 | A kind of NUMA system optimization method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710379373.8A CN107133106A (en) | 2017-05-25 | 2017-05-25 | A kind of NUMA system optimization method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107133106A true CN107133106A (en) | 2017-09-05 |
Family
ID=59733000
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710379373.8A Pending CN107133106A (en) | 2017-05-25 | 2017-05-25 | A kind of NUMA system optimization method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107133106A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112559130A (en) * | 2020-12-16 | 2021-03-26 | 恒生电子股份有限公司 | Container distribution method, device, electronic equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1614530A (en) * | 2003-11-03 | 2005-05-11 | 技嘉科技股份有限公司 | Method and apparatus for adjusting CPU working frequency automatically |
| CN106033371A (en) * | 2015-03-13 | 2016-10-19 | 杭州海康威视数字技术股份有限公司 | Method and system for dispatching video analysis task |
| CN106648890A (en) * | 2016-12-06 | 2017-05-10 | 中国科学院重庆绿色智能技术研究院 | Cloud computing server resource on-line management method and system with energy consumption sensing function |
-
2017
- 2017-05-25 CN CN201710379373.8A patent/CN107133106A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1614530A (en) * | 2003-11-03 | 2005-05-11 | 技嘉科技股份有限公司 | Method and apparatus for adjusting CPU working frequency automatically |
| CN106033371A (en) * | 2015-03-13 | 2016-10-19 | 杭州海康威视数字技术股份有限公司 | Method and system for dispatching video analysis task |
| CN106648890A (en) * | 2016-12-06 | 2017-05-10 | 中国科学院重庆绿色智能技术研究院 | Cloud computing server resource on-line management method and system with energy consumption sensing function |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112559130A (en) * | 2020-12-16 | 2021-03-26 | 恒生电子股份有限公司 | Container distribution method, device, electronic equipment and storage medium |
| CN112559130B (en) * | 2020-12-16 | 2024-01-19 | 恒生电子股份有限公司 | Container distribution method, device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20150324232A1 (en) | Resource management method and apparatus for virtual machine system, and virtual machine system | |
| CN103179048B (en) | Main frame qos policy transform method and the system of cloud data center | |
| CN103810048A (en) | Automatic adjusting method and device for thread number aiming to realizing optimization of resource utilization | |
| CN107145388A (en) | Method for scheduling task and system under a kind of multitask environment | |
| CN105975049B (en) | A kind of accidental task low energy consumption dispatching method of tasks synchronization | |
| CN105630606A (en) | Method and device for adjusting capacity of thread pools | |
| CN113467933B (en) | Distributed file system thread pool optimization method, system, terminal and storage medium | |
| CN112269641A (en) | Scheduling method, scheduling device, electronic equipment and storage medium | |
| CN113032102B (en) | Resource rescheduling method, device, equipment and medium | |
| CN106294198B (en) | A kind of memory allocation method and method for releasing of (SuSE) Linux OS | |
| CN107193649A (en) | A kind of method for scheduling task and device based on NUMA system | |
| CN105005504B (en) | A kind of the single task mode implementation method and system of Android platform | |
| CN109062666A (en) | A kind of cluster virtual machine management method and relevant apparatus | |
| CN104850423B (en) | The method for identifying the application program launching stage under Android system | |
| CN106407013B (en) | Method and device for dynamically scheduling resources, resource scheduling server and system | |
| CN109992392A (en) | A kind of calculation resource disposition method, device and Resource Server | |
| CN113626190B (en) | Load balancing method in microkernel operating system facing multi-kernel environment | |
| CN105824687B (en) | A kind of method and device of Java Virtual Machine performance automated tuning | |
| CN107133106A (en) | A kind of NUMA system optimization method and device | |
| CN107180118A (en) | A kind of file system cache data managing method and device | |
| CN105094944B (en) | A kind of virtual machine migration method and device | |
| CN112114647A (en) | Power supply control method, system and device of server | |
| CN111104218B (en) | Storage system data synchronization method, device, equipment and readable storage medium | |
| CN107463430A (en) | A kind of virutal machine memory dynamic management system and method based on internal memory and Swap spaces | |
| CN109144664B (en) | Dynamic migration method of virtual machine based on user service quality demand difference |
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: 20170905 |