CN104572272A - Task scheduling method, device and system - Google Patents
Task scheduling method, device and system Download PDFInfo
- Publication number
- CN104572272A CN104572272A CN201310476104.5A CN201310476104A CN104572272A CN 104572272 A CN104572272 A CN 104572272A CN 201310476104 A CN201310476104 A CN 201310476104A CN 104572272 A CN104572272 A CN 104572272A
- Authority
- CN
- China
- Prior art keywords
- efficiency ratio
- energy efficiency
- nucleoid
- ratio information
- node
- 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.)
- Granted
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]
- G06F9/5094—Allocation of resources, e.g. of the central processing unit [CPU] where the allocation takes into account power or heat criteria
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Power Sources (AREA)
Abstract
The invention belongs to the technical field of computers and provides a task scheduling method, device and system. The method comprises the following steps: obtaining energy efficiency ratio information of all first kinds of nucleuses in each slave node, sorting all the energy efficiency ratio information according to the size order of the values of energy efficiency ratios, and thus generating a sorted energy efficiency ratio information result set; obtaining all processes in a task running on the second kinds of nucleuses; by starting with the most idle process, sequentially searching the first kinds of nucleuses suitable for running each process by starting with the minimum energy efficiency ratio information in the energy efficiency ratio information result set; if the first kinds of nucleuses suitable for running the specified process in the energy efficiency ratio information result set is searched, migrating the specified process onto the first kinds of nucleuses for running. The method fully utilizes the resource of the first kinds of nucleuses. As the first kinds of nucleuses are relatively high in energy efficiency ratio, the energy consumption cost of the whole task data center can be reduced.
Description
Technical field
The invention belongs to field of computer technology, particularly relate to a kind of method for scheduling task, Apparatus and system.
Background technology
The development of data center roughly experienced by several stage.In early days, the computer house that computer realm is huge is the blank of data center.The nineties in 20th century, the computation schema of client/server is progressively popularized, and server starts to be placed separately, and data center comes into vogue.At the beginning of 21 century, along with the development of internet, Internet data center (Internet Data Center, IDC) rapid rising, the service that it can provide constantly is upgraded.In recent years, the problem such as climate warming, energy shortage, data center is faced with and reduces energy consumption, cost-saving severe challenge, data center is energy-conservation receives unprecedented concern, the concept of " green data center " is arisen at the historic moment, and will the new developing direction of data center becoming, marketable value is unprecedentedly huge here.
ARM company ARMv8 is proposed large small nut (A57 macronucleus, A53 small nut) scheme especially, to adapt to the change of data center, particularly to the process of large data.Some time period can be found out from the access-hours figure shown in Fig. 1, data traffic is very little and not urgent, if this time only to process data with macronucleus and too wastes, if only processed data with small nut, there will be again some time period small nut to data processing situation not in time.
Given this, prior art adopts in a Socket, data traffic very little and urgent time, to execution be gone above the task scheduling of macronucleus to small nut energy-conservation to realize, as shown in Figure 2, be the task scheduling schematic diagram in single Socket situation, wherein, Socket refers to " physical cpu ", and general terminal device is provided with several fan, and this terminal device just has several Socket.As can see from Figure 2, in cluster, from node-node 1, there is a macronucleus, a small nut, can the course allocation performed be needed the macronucleus in node 1 to small nut, replace macronucleus to execute the task by small nut.Although by this large small nut changing method, the energy consumption of system can be made to reduce, but the switching of this switching just between the large small nut of intra-node, that is be carry out large small nut switching based on terminal device or at a server internal, do not consider cluster internal from the switching between the large small nut node, such as, switching between small nut in macronucleus in node 1 and node 2, so there is no the Energy Efficiency Ratio considering each small nut, really cannot play the effect of the high small nut of Energy Efficiency Ratio.
Summary of the invention
Embodiments provide a kind of method for scheduling task, Apparatus and system, being intended to solve prior art when carrying out task scheduling, the problem of the effect of the high core of Energy Efficiency Ratio can not be fully played.
First aspect, provides a kind of method for scheduling task, and described method comprises:
Obtain the Energy Efficiency Ratio information of each all first nucleoids from node, and according to the size order of the value of Energy Efficiency Ratio, described Energy Efficiency Ratio information is sorted, generate the Energy Efficiency Ratio information result collection after sequence, the value of the Energy Efficiency Ratio of described first nucleoid is higher than the first efficiency ratio preset;
Obtain all processes in the task that current second nucleoid runs, the value of the Energy Efficiency Ratio of described second nucleoid is lower than the second efficiency ratio preset, and described first efficiency ratio is greater than described second efficiency ratio;
From the notst busy process, concentrate in described Energy Efficiency Ratio information result, from minimum efficiency ratio information, search the first nucleoid being applicable to running each process successively;
Wherein, the first nucleoid being applicable to running appointment process refers to from described Energy Efficiency Ratio information result concentrates the core found, and the processing power of this core is greater than the load of described appointment process;
If concentrate the first nucleoid finding and be applicable to running described appointment process in described Energy Efficiency Ratio information result, then run on described appointment process migration to described first nucleoid.
In the implementation that the first is possible, in conjunction with first aspect, from the notst busy process, concentrate in described Energy Efficiency Ratio information result, search the first nucleoid being applicable to running each process successively from minimum efficiency ratio information after, also comprise:
If concentrate the first nucleoid not finding and be applicable to running described appointment process from described Energy Efficiency Ratio information result, then reduce the cpu frequency of described current second nucleoid.
In the implementation that the second is possible, in conjunction with first aspect or the implementation possible according to the first, in the described acquisition respectively Energy Efficiency Ratio information of all first nucleoids from node, and according to the size order of the value of Energy Efficiency Ratio, described Energy Efficiency Ratio information is sorted, before generating the Energy Efficiency Ratio information result collection after sequence, also comprise:
Obtain respectively from the cpu frequency of node;
Each cpu frequency from node is weighted, obtains the service condition value of CPU;
If described service condition value is less than the first default processing ability value, then perform the described acquisition respectively Energy Efficiency Ratio information of all first nucleoids from node, and according to the size order of the value of Energy Efficiency Ratio, described Energy Efficiency Ratio information is sorted, generate the step of the Energy Efficiency Ratio information result collection after sequence;
Described first processing ability value is according to respectively presetting from the processing power of all first nucleoids node.
In the implementation that the third is possible, the implementation possible according to the second, is weighted each cpu frequency from node described, after obtaining the service condition value of CPU, also comprises:
If described service condition value is greater than the second default processing ability value, then continue all processes run on described second nucleoid in described task;
If described service condition value is greater than the first default processing ability value, be less than the second default processing ability value, then the task immigration that described second nucleoid runs run to the first nucleoid on the node at described second nucleoid place;
Described second processing ability value is according to respectively presetting from the processing power of all second nucleoids node.
Second aspect, provides a kind of task scheduling apparatus, it is characterized in that, described task scheduling apparatus comprises:
Energy Efficiency Ratio acquiring unit, for obtaining the Energy Efficiency Ratio information of each all first nucleoids from node, and according to the size order of the value of Energy Efficiency Ratio, described Energy Efficiency Ratio information is sorted, generate the Energy Efficiency Ratio information result collection after sequence, the value of the Energy Efficiency Ratio of described first nucleoid is higher than the first efficiency ratio preset;
Process acquiring unit, for obtaining all processes in the task that current second nucleoid runs, the value of the Energy Efficiency Ratio of described second nucleoid is lower than the second efficiency ratio preset, and described first efficiency ratio is greater than described second efficiency ratio;
Unit is looked in verification, for from the notst busy process, concentrate in described Energy Efficiency Ratio information result, the first nucleoid being applicable to running each process is searched successively from minimum efficiency ratio information, wherein, the first nucleoid being applicable to running appointment process refers to from described Energy Efficiency Ratio information result concentrates the core found, and the processing power of this core is greater than the load of described appointment process;
Core migration units, if for concentrating the first nucleoid finding and be applicable to running described appointment process in described Energy Efficiency Ratio information result, then run on described appointment process migration to described first nucleoid.
In the implementation that the first is possible, in conjunction with second aspect, described task scheduling apparatus also comprises:
Frequency adjustment unit, if for concentrating the first nucleoid not finding and be applicable to running described appointment process from described Energy Efficiency Ratio information result, then reduce the cpu frequency of described current second nucleoid.
In the implementation that the second is possible, in conjunction with second aspect or the implementation possible according to the first, described task scheduling apparatus also comprises:
Frequency acquisition unit, for obtaining respectively from the cpu frequency of node;
Service condition value acquiring unit, for being weighted each cpu frequency from node, obtains the service condition value of CPU;
First processing unit, if be less than the first default processing ability value for described service condition value, call Energy Efficiency Ratio acquiring unit, the Energy Efficiency Ratio information obtaining each all first nucleoids from node is performed by described Energy Efficiency Ratio acquiring unit, and according to the size order of the value of Energy Efficiency Ratio, described Energy Efficiency Ratio information is sorted, generate the step of the Energy Efficiency Ratio information result collection after sequence;
Described first processing ability value is according to respectively presetting from the processing power of all first nucleoids node.
In the implementation that the third is possible, the implementation possible according to the second, described task scheduling apparatus also comprises:
Second processing unit, if be greater than the second default processing ability value for described service condition value, then continues all processes run on described second nucleoid in described task;
3rd processing unit, if be greater than the first default processing ability value for described service condition value, be less than the second default processing ability value, then the task immigration that described second nucleoid runs run to the first nucleoid on the node at described second nucleoid place;
Described second processing ability value is according to respectively presetting from the processing power of all second nucleoids node.
The third aspect, provides a kind of task scheduling system, described task scheduling system comprise host node and be connected with described host node at least one from node, described host node comprises task scheduling apparatus as above.
As can be seen from such scheme, the embodiment of the present invention is by running the more idle process migration that the second nucleoid performs to the first nucleoid be applicable to, utilize first kind nuclear resource fully, because the first nucleoid has relatively high Energy Efficiency Ratio, so the energy consumption cost at whole task data center can be reduced, further, also more hardware resource can be used, such as hard disk, internal memory etc.
Accompanying drawing explanation
Fig. 1 is the realization flow figure of the method for scheduling task that the embodiment of the present invention one provides;
Fig. 2 is the realization flow figure of the method for scheduling task that the embodiment of the present invention two provides;
Fig. 3 is the structured flowchart of the task scheduling apparatus that the embodiment of the present invention three provides;
Fig. 4 is the structured flowchart of the task scheduling apparatus that the embodiment of the present invention four provides.
Embodiment
In order to make object of the present invention, technical scheme and advantage clearly understand, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein only in order to explain the present invention, be not intended to limit the present invention.
In embodiments of the present invention, the more idle process migration that second nucleoid performs is run to the first nucleoid be applicable to, utilize first kind nuclear resource fully, because the first nucleoid has relatively high Energy Efficiency Ratio, so the energy consumption cost at whole task data center can be reduced, further, also more hardware resource can be used, such as hard disk, internal memory etc.
Below in conjunction with specific embodiment, realization of the present invention is described in detail:
Embodiment one
Fig. 1 shows the realization flow of the method for scheduling task that the embodiment of the present invention one provides, and be described for the host node side in task scheduling system, details are as follows:
In step S101, obtain the Energy Efficiency Ratio information of each all first nucleoids from node, and according to the size order of the value of Energy Efficiency Ratio, described Energy Efficiency Ratio information is sorted, generate the Energy Efficiency Ratio information result collection after sequence, the value of the Energy Efficiency Ratio of described first nucleoid is higher than the first efficiency ratio preset.
In the embodiment of the present invention, Energy Efficiency Ratio information comprises the mark of each the first nucleoid and the value with the mark of each the first nucleoid Energy Efficiency Ratio one to one.Task scheduling system is by a host node and multiplely forming from node of being connected with this host node, and wherein, for ARM framework, each have macronucleus and small nut from node.Wherein, in ARM framework, core high for Energy Efficiency Ratio is called small nut, the core that Energy Efficiency Ratio is low is called macronucleus, and in Intel Architecture, only distinguishes the type of core by the height of Energy Efficiency Ratio, the not title of large small nut, so, in the present embodiment, the core in task scheduling system is divided into the first nucleoid and the second nucleoid, wherein the first nucleoid refers to the core that the value of Energy Efficiency Ratio is higher, and the second nucleoid refers to the core that the value of Energy Efficiency Ratio is lower.
Be described for ARM framework below, wherein the first nucleoid refers to small nut, and the second nucleoid refers to macronucleus.
Each can be macronucleus under ARMv7 framework and small nut from the macronucleus node and small nut, also can be the macronucleus under ARMv8 framework and small nut.Eachly belong to same framework from the macronucleus node and small nut, but different two can to have from node and belong to macronucleus two kinds of different frameworks and small nut, such as, belong to macronucleus under ARMv7 framework and small nut from the macronucleus node 1 and small nut, belong to macronucleus under ARMv8 framework and small nut from the macronucleus node 2 and small nut, that is, from in node, core under ARMv7 and ARMv8 two kinds of frameworks is also deposited, and the ratio of the Energy Efficiency Ratio between the large small nut under two kinds of frameworks is different.
Table 1 shows the Energy Efficiency Ratio information of the large small nut under ARMv7 framework, and the ratio of the Energy Efficiency Ratio of large small nut is " macronucleus: small nut ≈ 11:16 " as can be seen from Table 1.The ratio difference of A57 and the A53 Energy Efficiency Ratio of ARMv8 is larger, and the ratio of their Energy Efficiency Ratio is " macronucleus: small nut ≈ 1:4 ".Parameters in concrete table 1, and the acquisition methods of parameters value is prior art, does not repeat them here.
Table 1
After getting the Energy Efficiency Ratio information of each small nut from node, can sort according to the order of size to this Energy Efficiency Ratio information, specifically can according to order sequence from big to small, also can according to order sequence from small to large, this is not restricted, finally the result set obtained after this sequence can be stored, concrete storage mode is not also restricted at this, in the present embodiment, conveniently, with the Energy Efficiency Ratio information result collection after the form memory sequencing of chained list, and Energy Efficiency Ratio information is in chained list according to sequential storage from small to large.
In step s 102, all processes in the task that current second nucleoid runs, the value of the Energy Efficiency Ratio of described second nucleoid is lower than the second efficiency ratio preset, and described first efficiency ratio is greater than described second efficiency ratio.
In the embodiment of the present invention, the current task that macronucleus runs can comprise one or more process, host node can obtain all processes in the task that current macronucleus runs, and then sorts according to idle routine to these processes got, and generates process list.During concrete sequence, the notst busy process can be placed on the foremost of list, also the notst busy process can be placed on the end of process list, this is not restricted.
In step s 103, from the notst busy process, concentrate in described Energy Efficiency Ratio information result, the first nucleoid being applicable to running each process is searched successively from minimum efficiency ratio information, wherein, the first nucleoid being applicable to running appointment process refers to from described Energy Efficiency Ratio information result concentrates the core found, and the processing power of this core is greater than the load of described appointment process, if concentrate the first nucleoid finding and be applicable to running described appointment process in described Energy Efficiency Ratio information result, then perform step S104; Otherwise perform step S105.
In the embodiment of the present invention, the small nut being applicable to running appointment process refers to that the processing power of the small nut concentrating the Energy Efficiency Ratio information that finds corresponding from Energy Efficiency Ratio information result is greater than the load of this appointment process, if host node finds the small nut being applicable to running described appointment process, then perform step S104, wherein, described process of specifying refers to from one of them process macronucleus, need to perform to each process in macronucleus and concentrate in Energy Efficiency Ratio information result, from minimum efficiency ratio information, search the operation being applicable to the first nucleoid running this process successively; If searched all over whole Energy Efficiency Ratio information result collection, still do not find the small nut being applicable to running appointment process, then perform step S105.
In step S104, run on described appointment process migration to described first nucleoid.
In step S105, reduce the cpu frequency of described current second nucleoid.
In the embodiment of the present invention, host node judges whether the cpu frequency of current macronucleus is less than predeterminated frequency value, such as 50%, if, then judge whether the cpu frequency of described current macronucleus can reduce, if dropped to minimum, then do not reduce again, otherwise continued the cpu frequency reducing described current macronucleus.
As a preferred embodiment of the present invention, before the cpu frequency reducing current second nucleoid, efficiency ratio can also be selected from the second nucleoid higher than the core of described current second nucleoid, then by task scheduling on the CPU of this core, if do not find efficiency ratio lower than the core of described current second nucleoid, then just perform the step of the cpu frequency reducing described current second nucleoid.
The present embodiment, the more idle process migration that second nucleoid performs is run to the first nucleoid be applicable to, utilize first kind nuclear resource fully, because the first nucleoid has relatively high Energy Efficiency Ratio, so the energy consumption cost at whole task data center can be reduced, further, also more hardware resource can be used, such as hard disk, internal memory etc.
One of ordinary skill in the art will appreciate that all or part of step realized in the various embodiments described above method is that the hardware that can carry out instruction relevant by program has come, corresponding program can be stored in a computer read/write memory medium, described storage medium, as ROM/RAM, disk or CD etc.
Embodiment two
Fig. 2 shows the realization flow of the method for scheduling task that the embodiment of the present invention two provides, and is described for host node side, and the present embodiment performs on the basis of embodiment one, and details are as follows:
In step s 201, obtain respectively from the cpu frequency node.
In embodiments of the present invention, host node sends the request of the inquiry cpu frequency of preset times to each from node in T time, after respectively receiving this request from node, respective cpu frequency is sent to host node.Host node is to preset times, and such as the Query Result of 3 times is averaged, using this mean value as final cpu frequency.
In addition, each also initiatively can send respective cpu frequency to host node from node, and specifically, how host node obtains respectively from the cpu frequency of node, and this is not restricted.
In step S202, each cpu frequency is weighted, obtains cpu usage value.
In step S203, described service condition value is judged, if described service condition value is less than the first default processing ability value, then perform step S204 to S208 successively; If described service condition value is greater than the second default processing ability value, then perform step S209; If described service condition value is greater than the first default processing ability value, be less than the second default processing ability value, then perform step S210.
In embodiments of the present invention, step control can be realized according to the frequency of each CPU from node to the switching between large small nut.
Wherein, the first processing ability value is that in the present embodiment, the first processing ability value is after being weighted the processing power of all small nuts, then is averaging, 80% of the mean value finally obtained according to presetting from the processing power of all small nuts node.
Second processing ability value is that in the present embodiment, the second processing ability value is after being weighted the processing power of all macronucleus, then is averaging, 80% of the mean value finally obtained according to presetting from the processing power of all macronucleus node.
In step S204, obtain the Energy Efficiency Ratio information of each all first nucleoids from node, and according to the size order of the value of Energy Efficiency Ratio, described Energy Efficiency Ratio information is sorted, generate the Energy Efficiency Ratio information result collection after sequence, the value of the Energy Efficiency Ratio of described first nucleoid is higher than the first efficiency ratio preset.
In step S205, obtain all processes in the task that current second nucleoid runs, the value of the Energy Efficiency Ratio of described second nucleoid is lower than the second efficiency ratio preset, and described first efficiency ratio is greater than described second efficiency ratio.
In step S206, from the notst busy process, concentrate in described Energy Efficiency Ratio information result, the first nucleoid being applicable to running each process is searched successively from minimum efficiency ratio information, wherein, the first nucleoid being applicable to running appointment process refers to from described Energy Efficiency Ratio information result concentrates the core found, and the processing power of this core is greater than the load of described appointment process, if concentrate the first nucleoid finding and be applicable to running described appointment process in described Energy Efficiency Ratio information result, then perform step S207; Otherwise perform step S208.
In step S207, run on described appointment process migration to described first nucleoid.
In step S208, reduce the cpu frequency of described second nucleoid.
In step S209, continue all processes run on described current second nucleoid in described task.
In step S210, the task immigration that described current second nucleoid runs is run to the first nucleoid of local node.
The present embodiment, host node obtains respectively from the cpu frequency of node, the migration model of the task in the second nucleoid is determined according to this frequency, achieve the step control of the first nucleoid and the switching of the second nucleoid, the switching frequency between the first nucleoid and the second nucleoid can be controlled further, improve the available rate of task scheduling system.
Embodiment three
Fig. 3 shows the concrete structure block diagram of the task scheduling apparatus that the embodiment of the present invention three provides, and for convenience of explanation, illustrate only the part relevant to the embodiment of the present invention.This task scheduling apparatus 3 can be the unit of the software unit be built in host node, hardware cell or software and hardware combining, wherein, host node is a node in task scheduling system, this node is connected from node with at least one, and described task scheduling apparatus 3 comprises: Energy Efficiency Ratio acquiring unit 31, process acquiring unit 32, verification look for unit 33 and core migration units 34.
Energy Efficiency Ratio acquiring unit 31, for obtaining the Energy Efficiency Ratio information of each all first nucleoids from node, and according to the size order of the value of Energy Efficiency Ratio, described Energy Efficiency Ratio information is sorted, generate the Energy Efficiency Ratio information result collection after sequence, the value of the Energy Efficiency Ratio of described first nucleoid is higher than the first efficiency ratio preset;
Process acquiring unit 32, for obtaining all processes in the task that current second nucleoid runs, the value of the Energy Efficiency Ratio of described second nucleoid is lower than the second efficiency ratio preset, and described first efficiency ratio is greater than described second efficiency ratio;
Unit 33 is looked in verification, for from the notst busy process, concentrate in described Energy Efficiency Ratio information result, the first nucleoid being applicable to running each process is searched successively from minimum efficiency ratio information, wherein, the first nucleoid being applicable to running appointment process refers to from described Energy Efficiency Ratio information result concentrates the core found, and the processing power of this core is greater than the load of described appointment process;
Core migration units 34, if for concentrating the first nucleoid finding and be applicable to running described appointment process in described Energy Efficiency Ratio information result, then run on described appointment process migration to described first nucleoid.
As a preferred embodiment of the present invention, this task scheduling apparatus 3 also comprises:
Frequency adjustment unit 35, if for concentrating the first nucleoid not finding and be applicable to running described appointment process from described Energy Efficiency Ratio information result, then reduce the cpu frequency of described current second nucleoid.
The task scheduling apparatus 3 that the embodiment of the present invention provides can be applied in the method for scheduling task embodiment one of aforementioned correspondence, and details, see the description of above-described embodiment one, do not repeat them here.
Embodiment four
Fig. 4 shows the concrete structure block diagram of the task scheduling apparatus that the embodiment of the present invention four provides, and for convenience of explanation, illustrate only the part relevant to the embodiment of the present invention.This task scheduling apparatus 4 can be the software unit be built in host node, the unit of hardware cell or software and hardware combining, wherein, host node is a node in task scheduling system, this node is connected from node with at least one, described task scheduling apparatus 4 is except comprising the Energy Efficiency Ratio acquiring unit 31 described in embodiment three, process acquiring unit 32, unit 33 is looked in verification, core migration units 34, frequency adjustment unit 35, also comprise frequency acquisition unit 41, service condition value acquiring unit 42, first processing unit 43, second processing unit 44 and the 3rd processing unit 45.
Wherein, frequency acquisition unit 41, for obtaining respectively from the cpu frequency of node;
Service condition value acquiring unit 42, for being weighted each cpu frequency from node, obtains the service condition value of CPU;
First processing unit 43, if be less than the first default processing ability value for described service condition value, call Energy Efficiency Ratio acquiring unit 31, the Energy Efficiency Ratio information obtaining each all first nucleoids from node is performed by described Energy Efficiency Ratio acquiring unit, and according to the size order of the value of Energy Efficiency Ratio, described Energy Efficiency Ratio information is sorted, generate the step of the Energy Efficiency Ratio information result collection after sequence;
Described first processing ability value is according to respectively presetting from the processing power of all first nucleoids node;
Second processing unit 44, if be greater than the second default processing ability value for described service condition value, then continues all processes run on described second nucleoid in described task;
3rd processing unit 45, if be greater than the first default processing ability value for described service condition value, be less than the second default processing ability value, then the task immigration that described second nucleoid runs run to the first nucleoid on the node at described second nucleoid place;
Described second processing ability value is according to respectively presetting from the processing power of all second nucleoids node.
The task scheduling apparatus 4 that the embodiment of the present invention provides can be applied in the method for scheduling task embodiment two of aforementioned correspondence, and details, see the description of above-described embodiment two, do not repeat them here.
It should be noted that in said system embodiment, included unit is carry out dividing according to function logic, but is not limited to above-mentioned division, as long as can realize corresponding function; In addition, the concrete title of each functional unit, also just for the ease of mutual differentiation, is not limited to protection scope of the present invention.
The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, all any amendments done within the spirit and principles in the present invention, equivalent replacement and improvement etc., all should be included within protection scope of the present invention.
Claims (9)
1. a method for scheduling task, is characterized in that, described method comprises:
Obtain the Energy Efficiency Ratio information of each all first nucleoids from node, and according to the size order of the value of Energy Efficiency Ratio, described Energy Efficiency Ratio information is sorted, generate the Energy Efficiency Ratio information result collection after sequence, the value of the Energy Efficiency Ratio of described first nucleoid is higher than the first efficiency ratio preset;
Obtain all processes in the task that current second nucleoid runs, the value of the Energy Efficiency Ratio of described second nucleoid is lower than the second efficiency ratio preset, and described first efficiency ratio is greater than described second efficiency ratio;
From the notst busy process, concentrate in described Energy Efficiency Ratio information result, from minimum efficiency ratio information, search the first nucleoid being applicable to running each process successively;
Wherein, the first nucleoid being applicable to running appointment process refers to from described Energy Efficiency Ratio information result concentrates the core found, and the processing power of this core is greater than the load of described appointment process;
If concentrate the first nucleoid finding and be applicable to running described appointment process in described Energy Efficiency Ratio information result, then run on described appointment process migration to described first nucleoid.
2. the method for claim 1, is characterized in that, from the notst busy process, concentrates in described Energy Efficiency Ratio information result, after searching the first nucleoid being applicable to running each process successively, also comprises from minimum efficiency ratio information:
If concentrate the first nucleoid not finding and be applicable to running described appointment process from described Energy Efficiency Ratio information result, then reduce the cpu frequency of described current second nucleoid.
3. method as claimed in claim 1 or 2, it is characterized in that, in the described acquisition respectively Energy Efficiency Ratio information of all first nucleoids from node, and according to the size order of the value of Energy Efficiency Ratio, described Energy Efficiency Ratio information is sorted, before generating the Energy Efficiency Ratio information result collection after sequence, also comprise:
Obtain respectively from the cpu frequency of node;
Each cpu frequency from node is weighted, obtains the service condition value of CPU;
If described service condition value is less than the first default processing ability value, then perform the described acquisition respectively Energy Efficiency Ratio information of all first nucleoids from node, and according to the size order of the value of Energy Efficiency Ratio, described Energy Efficiency Ratio information is sorted, generate the step of the Energy Efficiency Ratio information result collection after sequence;
Described first processing ability value is according to respectively presetting from the processing power of all first nucleoids node.
4. method as claimed in claim 3, is characterized in that, be weighted, after obtaining the service condition value of CPU, also comprise described to each cpu frequency from node:
If described service condition value is greater than the second default processing ability value, then continue all processes run on described second nucleoid in described task;
If described service condition value is greater than the first default processing ability value, be less than the second default processing ability value, then the task immigration that described second nucleoid runs run to the first nucleoid on the node at described second nucleoid place;
Described second processing ability value is according to respectively presetting from the processing power of all second nucleoids node.
5. a task scheduling apparatus, is characterized in that, described task scheduling apparatus comprises:
Energy Efficiency Ratio acquiring unit, for obtaining the Energy Efficiency Ratio information of each all first nucleoids from node, and according to the size order of the value of Energy Efficiency Ratio, described Energy Efficiency Ratio information is sorted, generate the Energy Efficiency Ratio information result collection after sequence, the value of the Energy Efficiency Ratio of described first nucleoid is higher than the first efficiency ratio preset;
Process acquiring unit, for obtaining all processes in the task that current second nucleoid runs, the value of the Energy Efficiency Ratio of described second nucleoid is lower than the second efficiency ratio preset, and described first efficiency ratio is greater than described second efficiency ratio;
Unit is looked in verification, for from the notst busy process, concentrate in described Energy Efficiency Ratio information result, the first nucleoid being applicable to running each process is searched successively from minimum efficiency ratio information, wherein, the first nucleoid being applicable to running appointment process refers to from described Energy Efficiency Ratio information result concentrates the core found, and the processing power of this core is greater than the load of described appointment process;
Core migration units, if for concentrating the first nucleoid finding and be applicable to running described appointment process in described Energy Efficiency Ratio information result, then run on described appointment process migration to described first nucleoid.
6. task scheduling apparatus as claimed in claim 5, it is characterized in that, described task scheduling apparatus also comprises:
Frequency adjustment unit, if for concentrating the first nucleoid not finding and be applicable to running described appointment process from described Energy Efficiency Ratio information result, then reduce the cpu frequency of described current second nucleoid.
7. the task scheduling apparatus as described in claim 5 or 6, is characterized in that, described task scheduling apparatus also comprises:
Frequency acquisition unit, for obtaining respectively from the cpu frequency of node;
Service condition value acquiring unit, for being weighted each cpu frequency from node, obtains the service condition value of CPU;
First processing unit, if be less than the first default processing ability value for described service condition value, call Energy Efficiency Ratio acquiring unit, the Energy Efficiency Ratio information obtaining each all first nucleoids from node is performed by described Energy Efficiency Ratio acquiring unit, and according to the size order of the value of Energy Efficiency Ratio, described Energy Efficiency Ratio information is sorted, generate the step of the Energy Efficiency Ratio information result collection after sequence;
Described first processing ability value is according to respectively presetting from the processing power of all first nucleoids node.
8. task scheduling apparatus as claimed in claim 7, it is characterized in that, described task scheduling apparatus also comprises:
Second processing unit, if be greater than the second default processing ability value for described service condition value, then continues all processes run on described second nucleoid in described task;
3rd processing unit, if be greater than the first default processing ability value for described service condition value, be less than the second default processing ability value, then the task immigration that described second nucleoid runs run to the first nucleoid on the node at described second nucleoid place;
Described second processing ability value is according to respectively presetting from the processing power of all second nucleoids node.
9. a task scheduling system, is characterized in that, described task scheduling system comprise host node and be connected with described host node at least one from node, described host node comprises the task scheduling apparatus as described in any one of claim 5 to 8.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310476104.5A CN104572272B (en) | 2013-10-12 | 2013-10-12 | A kind of method for scheduling task, apparatus and system |
| PCT/CN2014/086124 WO2015051685A1 (en) | 2013-10-12 | 2014-09-09 | Task scheduling method, device and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310476104.5A CN104572272B (en) | 2013-10-12 | 2013-10-12 | A kind of method for scheduling task, apparatus and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104572272A true CN104572272A (en) | 2015-04-29 |
| CN104572272B CN104572272B (en) | 2018-02-09 |
Family
ID=52812510
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310476104.5A Active CN104572272B (en) | 2013-10-12 | 2013-10-12 | A kind of method for scheduling task, apparatus and system |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN104572272B (en) |
| WO (1) | WO2015051685A1 (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105354084A (en) * | 2015-10-30 | 2016-02-24 | 浪潮(北京)电子信息产业有限公司 | CPU task immigration method and system based on bandwidth scheduling |
| CN108021223A (en) * | 2017-11-30 | 2018-05-11 | 奇酷互联网络科技（深圳）有限公司 | Optimize method, equipment, mobile terminal and the computer-readable storage medium of stand-by power consumption |
| CN109947569A (en) * | 2019-03-15 | 2019-06-28 | Oppo广东移动通信有限公司 | Bind method, apparatus, terminal and the storage medium of core |
| WO2020063041A1 (en) * | 2018-09-29 | 2020-04-02 | Oppo广东移动通信有限公司 | Method and apparatus for scheduling multi-core processor, and terminal and storage medium |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107066326A (en) * | 2017-03-27 | 2017-08-18 | 深圳市金立通信设备有限公司 | The method and terminal of a kind of scheduler task |
| CN110290215B (en) * | 2019-06-28 | 2021-09-28 | 深圳前海微众银行股份有限公司 | Signal transmission method and device |
| CN115016885B (en) * | 2021-11-19 | 2023-11-24 | 荣耀终端有限公司 | Virtual machine garbage recycling operation method and electronic equipment |
| CN114461404B (en) * | 2022-04-01 | 2022-07-08 | 统信软件技术有限公司 | Process migration method, computing device and readable storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101604264A (en) * | 2009-07-08 | 2009-12-16 | 深圳先进技术研究院 | The method for scheduling task of supercomputer and system |
| US20100100886A1 (en) * | 2007-03-02 | 2010-04-22 | Masamichi Takagi | Task group allocating method, task group allocating device, task group allocating program, processor and computer |
| JP2010160565A (en) * | 2009-01-06 | 2010-07-22 | Ricoh Co Ltd | Task scheduling apparatus, control method for task scheduling, and control program for task scheduling |
| CN102520785A (en) * | 2011-12-27 | 2012-06-27 | 东软集团股份有限公司 | Energy consumption management method and system for cloud data center |
| CN102707996A (en) * | 2012-05-15 | 2012-10-03 | 江苏中科梦兰电子科技有限公司 | Task scheduling method for heterogeneous multi-core processor |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101065436B1 (en) * | 2010-12-07 | 2011-09-19 | 경상대학교산학협력단 | Stochastic scheduling of a real-time parallel task with uncertain computation amount on mulit-core processors |
| CN102208986B (en) * | 2011-03-22 | 2014-01-15 | 曙光信息产业(北京)有限公司 | Cluster power consumption distribution and control method |
-
2013
- 2013-10-12 CN CN201310476104.5A patent/CN104572272B/en active Active
-
2014
- 2014-09-09 WO PCT/CN2014/086124 patent/WO2015051685A1/en active Application Filing
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100100886A1 (en) * | 2007-03-02 | 2010-04-22 | Masamichi Takagi | Task group allocating method, task group allocating device, task group allocating program, processor and computer |
| JP2010160565A (en) * | 2009-01-06 | 2010-07-22 | Ricoh Co Ltd | Task scheduling apparatus, control method for task scheduling, and control program for task scheduling |
| CN101604264A (en) * | 2009-07-08 | 2009-12-16 | 深圳先进技术研究院 | The method for scheduling task of supercomputer and system |
| CN102520785A (en) * | 2011-12-27 | 2012-06-27 | 东软集团股份有限公司 | Energy consumption management method and system for cloud data center |
| CN102707996A (en) * | 2012-05-15 | 2012-10-03 | 江苏中科梦兰电子科技有限公司 | Task scheduling method for heterogeneous multi-core processor |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105354084A (en) * | 2015-10-30 | 2016-02-24 | 浪潮(北京)电子信息产业有限公司 | CPU task immigration method and system based on bandwidth scheduling |
| CN105354084B (en) * | 2015-10-30 | 2018-12-18 | 浪潮(北京)电子信息产业有限公司 | A kind of CPU task immigration method and system based on bandwidth scheduling |
| CN108021223A (en) * | 2017-11-30 | 2018-05-11 | 奇酷互联网络科技（深圳）有限公司 | Optimize method, equipment, mobile terminal and the computer-readable storage medium of stand-by power consumption |
| WO2020063041A1 (en) * | 2018-09-29 | 2020-04-02 | Oppo广东移动通信有限公司 | Method and apparatus for scheduling multi-core processor, and terminal and storage medium |
| CN110968415A (en) * | 2018-09-29 | 2020-04-07 | Oppo广东移动通信有限公司 | Scheduling method and device of multi-core processor and terminal |
| CN110968415B (en) * | 2018-09-29 | 2022-08-05 | Oppo广东移动通信有限公司 | Scheduling method and device of multi-core processor and terminal |
| CN109947569A (en) * | 2019-03-15 | 2019-06-28 | Oppo广东移动通信有限公司 | Bind method, apparatus, terminal and the storage medium of core |
| CN109947569B (en) * | 2019-03-15 | 2021-04-06 | Oppo广东移动通信有限公司 | Method, device, terminal and storage medium for binding core |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104572272B (en) | 2018-02-09 |
| WO2015051685A1 (en) | 2015-04-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104572272A (en) | Task scheduling method, device and system | |
| CN106708989B (en) | Skyline query method based on space time sequence data stream application | |
| CN106777093B (en) | Skyline inquiry system based on space time sequence data flow application | |
| CN102207891B (en) | Method for achieving dynamic partitioning and load balancing of data-partitioning distributed environment | |
| CN103077082A (en) | Method and system for distributing data center load and saving energy during virtual machine migration | |
| Liao et al. | An improved parallel K-means clustering algorithm with MapReduce | |
| CN104102543A (en) | Load regulation method and load regulation device in cloud computing environment | |
| CN101593202A (en) | Based on the hash connecting method for database of sharing the Cache polycaryon processor | |
| CN112395289A (en) | Distributed photovoltaic data layered storage method and system | |
| CN101719148A (en) | Three-dimensional spatial information saving method, device, system and dispatching system | |
| CN101819459B (en) | Heterogeneous object memory system-based power consumption control method | |
| Song et al. | An efficient scheduling algorithm for energy consumption constrained parallel applications on heterogeneous distributed systems | |
| CN106326005B (en) | Parameter automatic tuning method for iterative MapReduce operation | |
| CN102855157A (en) | Method for comprehensively scheduling load of servers | |
| CN108319604B (en) | Optimization method for association of large and small tables in hive | |
| Zhang et al. | A locally distributed mobile computing framework for dnn based android applications | |
| CN112100893B (en) | Discontinuous domain optimization method for finite element calculation | |
| CN116431281A (en) | Virtual machine migration method based on whale optimization algorithm | |
| CN114138102A (en) | Cloud computing data center energy consumption optimization method based on multi-region division | |
| CN103425787B (en) | The triangle gridding of a kind of asymptotic optimization repeats the quick minimizing technology in summit | |
| CN109508354A (en) | A kind of parallel processing system (PPS) | |
| CN111813512B (en) | High-energy-efficiency Spark task scheduling method based on dynamic partition | |
| CN103678545A (en) | Network resource clustering method and device | |
| CN111881086B (en) | Big data storage method, query method, electronic device and storage medium | |
| CN104866611B (en) | A kind of practical method for fast searching of power system Active Splitting section |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20200415 Address after: 518129 Bantian HUAWEI headquarters office building, Longgang District, Guangdong, Shenzhen Patentee after: HUAWEI TECHNOLOGIES Co.,Ltd. Address before: 301, A building, room 3, building 301, foreshore Road, No. 310000, Binjiang District, Zhejiang, Hangzhou Patentee before: Huawei Technologies Co.,Ltd. |