CN105718317A - Task scheduling method and task scheduling device - Google Patents

Task scheduling method and task scheduling device Download PDF

Info

Publication number
CN105718317A
CN105718317A CN201610029299.2A CN201610029299A CN105718317A CN 105718317 A CN105718317 A CN 105718317A CN 201610029299 A CN201610029299 A CN 201610029299A CN 105718317 A CN105718317 A CN 105718317A
Authority
CN
China
Prior art keywords
task
working node
queue
represent
task queue
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
Application number
CN201610029299.2A
Other languages
Chinese (zh)
Other versions
CN105718317B (en
Inventor
苏志远
亓开元
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Inspur Beijing Electronic Information Industry Co Ltd
Original Assignee
Inspur Beijing Electronic Information Industry Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Inspur Beijing Electronic Information Industry Co Ltd filed Critical Inspur Beijing Electronic Information Industry Co Ltd
Priority to CN201610029299.2A priority Critical patent/CN105718317B/en
Publication of CN105718317A publication Critical patent/CN105718317A/en
Application granted granted Critical
Publication of CN105718317B publication Critical patent/CN105718317B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46Multiprogramming arrangements
    • G06F9/50Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5005Allocation of resources, e.g. of the central processing unit [CPU] to service a request
    • G06F9/5027Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
    • G06F9/5038Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering the execution order of a plurality of tasks, e.g. taking priority or time dependency constraints into consideration
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46Multiprogramming arrangements
    • G06F9/48Program initiating; Program switching, e.g. by interrupt
    • G06F9/4806Task transfer initiation or dispatching
    • G06F9/4843Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
    • G06F9/4881Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues

Abstract

The invention discloses a task scheduling method and a task scheduling device, wherein the method comprises the following steps of obtaining request information of TaskTrackers, wherein the request information is task requesting information sent when the TaskTrackers have an idle work groove; calculating the resource consumption features and the rest task proportion of each task queue, wherein the resource construction feature of any one task queue is obtained through calculation according to the CPU (Central Processing Unit) occupation rate, the memory occupation rate and the bandwidth occupation rate of operating tasks in the task queue, and the rest task proportion of any one task queue is the ratio value of non-operating tasks to all of the tasks in the task queue; selecting the task array achieving the effect that the product of the resource consumption feature and the rest task proportion in the task queue is not smaller than that of other task queues as a target task queue; and processing the non-operating tasks in the target task queue by using the idle work groove. Therefore the non-operating tasks in the task queue in urgent need of being processed are preferably processed, and the task execution efficiency is greatly improved.

Description

A kind of method for scheduling task and device
Technical field
The present invention relates to big data Hadoop colony dispatching technical field, more particularly, it relates to a kind of method for scheduling task and device.
Background technology
Hadoop is a distributed system architecture, and along with the application of Hadoop is more and more extensive, its problems faced is also more and more many, wherein, including the problems faced when realizing the scheduling for task.
Specifically, in prior art, under Hadoop platform, the job scheduling algorithm of acquiescence is mainly FIFO (FirstInputFirstOutput, FIFO) algorithm, when namely needing task is scheduling, according to the principle of FIFO, task is scheduling, every time the task of scheduling is existing at first in not processed task of task, hereby it is achieved that about time-related fairness in the process of scheduler task.But, inventor have found that, the scheduling of task is carried out according to FIFO algorithm, consider only time-related fairness, and do not take into account the practical situation that different tasks is pending, it is possible to priority treatment cannot be badly in need of processed task, thus, cause task possibly being badly in need of being processed to be processed in time, and then reduce tasks carrying efficiency.
In sum, prior art exists the problem owing to task possibly being badly in need of being processed cannot be processed and cause tasks carrying inefficient in time.
Summary of the invention
It is an object of the invention to provide a kind of method for scheduling task and device, with the problem that cannot be processed in time and cause tasks carrying inefficient due to task possibly being badly in need of being processed solving to exist in prior art.
To achieve these goals, the present invention provides following technical scheme:
A kind of method for scheduling task, including:
Obtaining the solicited message of working node, described solicited message is that described working node exists the information of the request task of transmission during vacant working groove;
Calculate resource consumption feature and the residue task ratio of each task queue, wherein, the resource consumption of arbitrary task queue is characterized as and is calculated obtaining according to the CPU usage of being currently running in this task queue of task, memory usage and bandwidth usage, and the residue task ratio of arbitrary task queue is the ratio of off-duty task and whole tasks in this task queue;
Choosing the product of its resource consumption feature in described task queue and residue task ratio, to be not less than the task queue of other task queues be goal task queue, and utilizes described vacant working groove that the off-duty task in described goal task queue is processed.
Preferably, utilize described vacant working groove that the off-duty task in described goal task queue is processed, including:
Calculating the ability characteristics of described working node, wherein, described ability characteristics is be calculated obtaining according to the property value that the CPU of described working node, internal memory, hard disk and bandwidth are corresponding;
Calculate the resource occupation feature of the off-duty task comprised in described goal task queue, wherein, in described goal task queue, the resource occupation of arbitrary off-duty task is characterized as the ratio of all meansigma methodss of the work nest quantity that tasks take in work nest quantity that this off-duty task takies and described goal task queue;
Choosing the off-duty task that in described goal task queue, the ability characteristics of its resource occupation feature and described working node matches is goal task, and utilizes described vacant working groove that described goal task is processed.
Preferably, choosing the off-duty task that in described goal task queue, the ability characteristics of its resource occupation feature and described working node matches is goal task, including:
The ability characteristics of described working node and node capacity threshold value are compared, if the ability characteristics of described working node is more than described node capacity threshold value, then judge the off-duty task that whether there is its resource occupation feature in described object queue more than resource occupation threshold value, if it is, by its resource occupation feature more than the off-duty task of resource occupation threshold value being chosen a task as goal task;
If the ability characteristics of described working node is equal to described node capacity threshold value, then judge the off-duty task that whether there is its resource occupation feature in described object queue equal to resource occupation threshold value, a task is chosen as goal task if it is, be equal in the off-duty task of resource occupation threshold value by its resource occupation feature;
If the ability characteristics of described working node is less than described node capacity threshold value, then judge the off-duty task that whether there is its resource occupation feature in described object queue less than resource occupation threshold value, if it is, by its resource occupation feature less than the off-duty task of resource occupation threshold value being chosen a task as goal task.
Preferably, also include:
Whole tasks are divided into the first predetermined number task queue, and distribute the second predetermined number the work nest for the task in this task queue is processed for each task queue.
Preferably, calculate the resource consumption feature of each task queue, including: calculate the resource consumption feature of arbitrary task queue according to the following formula:
C Q i = r c p u × T c p u + r m e m × T m e m + r b a n d × T b a n d
Wherein,Represent the resource consumption feature of arbitrary task queue, TcpuRepresent the meansigma methods of the CPU usage of being currently running in this task queue of task, rcpuRepresent TcpuShared weight, TmemRepresent the meansigma methods of the memory usage of being currently running in this task queue of task, rmemRepresent TmemShared weight, TbandRepresent the meansigma methods of the bandwidth usage of being currently running in this task queue of task, rbandRepresent TbandShared weight.
Preferably, calculate the ability characteristics of described working node, including: calculate the ability characteristics of described working node according to the following formula:
N r e s o u r c e = w c u c p u avg c p u + w m u m e m avg m e m + w d u d i s k avg d i s k + w b u b a n d avg b a n d
Wherein, NresourceRepresent the ability characteristics of described working node, ucpuRepresent the CPU frequency of described working node, wcRepresent ucpuShared weight, umemRepresent the memory size of described working node, wmRepresent umemShared weight, udiskRepresent the hard disk size of described working node, wdRepresent udiskShared weight, ubandRepresent the amount of bandwidth of described working node, wbRepresent ubandShared weight, avgcpuRepresent described working node the meansigma methods of whole working node CPU frequencys in the cluster, avgmemRepresent described working node the meansigma methods of whole working node memory sizes in the cluster, avgdiskRepresent described working node the meansigma methods of whole working node hard disk size in the cluster, avgbandRepresent described working node the meansigma methods of whole working node amount of bandwidth in the cluster.
Preferably, also include:
Calculating the load condition of described working node, wherein, described load condition is calculated according to the CPU usage of described working node, memory usage and bandwidth usage;
Judge that whether described load condition is less than node load threshold value, if it is, calculate resource consumption feature and the residue task ratio of each task queue, if it is not, then refusal is described working node distribution task.
Preferably, calculate the load condition of described working node, including: calculate the load condition of described working node according to the following formula:
L=fcpu×Lcpu+fmem×Lmem+fband×Lband
Wherein, L represents the load condition of described working node, LcpuRepresent the CPU usage of described working node, fcpuRepresent LcpuShared weight, LmemRepresent the memory usage of described working node, fmemRepresent LmemShared weight, LbandRepresent the bandwidth usage of described working node, fbandRepresent LbandShared weight.
A kind of task scheduling apparatus, including:
Monitoring module, for obtaining the solicited message of working node, described solicited message is that described working node exists the information of the request task of transmission during vacant working groove;
Computing module, for calculating resource consumption feature and the residue task ratio of each task queue, wherein, the resource consumption of arbitrary task queue is characterized as and is calculated obtaining according to the CPU usage of being currently running in this task queue of task, memory usage and bandwidth usage, and the residue task ratio of arbitrary task queue is the ratio of off-duty task and whole tasks in this task queue;
Choose module, it is goal task queue that product for choosing its resource consumption feature in described task queue and residue task ratio is not less than the task queue of other task queues, and utilizes described vacant working groove that the off-duty task in described goal task queue is processed.
Preferably, choose module to include:
Choosing unit, be used for: calculate the ability characteristics of described working node, wherein, described ability characteristics is be calculated obtaining according to the property value that the CPU of described working node, internal memory, hard disk and bandwidth are corresponding;Calculate the resource occupation feature of the off-duty task comprised in described goal task queue, wherein, in described goal task queue, the resource occupation of arbitrary off-duty task is characterized as the ratio of all meansigma methodss of the work nest quantity that tasks take in work nest quantity that this off-duty task takies and described goal task queue;Choosing the off-duty task that in described goal task queue, the ability characteristics of its resource occupation feature and described working node matches is goal task, and utilizes described vacant working groove that described goal task is processed.
The invention provides a kind of method for scheduling task and device, wherein, the method includes: obtain the solicited message of working node, and described solicited message is that described working node exists the information of the request task of transmission during vacant working groove;Calculate resource consumption feature and the residue task ratio of each task queue, wherein, the resource consumption of arbitrary task queue is characterized as and is calculated obtaining according to the CPU usage of being currently running in this task queue of task, memory usage and bandwidth usage, and the residue task ratio of arbitrary task queue is the ratio of off-duty task and whole tasks in this task queue;Choosing the product of its resource consumption feature in described task queue and residue task ratio, to be not less than the task queue of other task queues be goal task queue, and utilizes described vacant working groove that the off-duty task in described goal task queue is processed.Compared with prior art, technique scheme disclosed in the present application considers the situation (i.e. residue task ratio) of off-duty task in the resource consumption (i.e. resource consumption feature) of being currently running in arbitrary task queue of task and this task queue and determines whether this task queue is in resource tense situation, resource consumption feature is more big with the product of residue task ratio, show that the resource tense situation of corresponding task queue is more serious, more need in time corresponding task queue to be processed, therefore, choose the goal task queue that in task queue, resource tense situation is the most serious to process, thus priority treatment is badly in need of the off-duty task in processed task queue, not only better ensure that the fairness of task scheduling, and substantially increase tasks carrying efficiency, decrease the total time of tasks carrying.
Accompanying drawing explanation
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, the accompanying drawing used required in embodiment or description of the prior art will be briefly described below, apparently, accompanying drawing in the following describes is only embodiments of the invention, for those of ordinary skill in the art, under the premise not paying creative work, it is also possible to obtain other accompanying drawing according to the accompanying drawing provided.
The flow chart of a kind of method for scheduling task that Fig. 1 provides for the embodiment of the present invention;
The flow chart utilizing described vacant working groove that the off-duty task in described goal task queue is processed in a kind of method for scheduling task that Fig. 2 provides for the embodiment of the present invention;
The structural representation of a kind of task scheduling apparatus that Fig. 3 provides for the embodiment of the present invention.
Detailed description of the invention
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is only a part of embodiment of the present invention, rather than whole embodiments.Based on the embodiment in the present invention, the every other embodiment that those of ordinary skill in the art obtain under not making creative work premise, broadly fall into the scope of protection of the invention.
Refer to Fig. 1, it illustrates the flow chart of a kind of method for scheduling task that the embodiment of the present invention provides, it is possible to comprise the following steps:
S1: obtain the solicited message of working node, solicited message is that working node exists the information of the request task of transmission during vacant working groove.
A kind of method for scheduling task and device that the embodiment of the present invention provides specifically can apply to Hadoop cluster, namely, it is applied to include in the Hadoop of multiple working node, and each working node in cluster is respectively provided with the work nest of respective amount, in cluster, the quantity of the work nest of each working node can be identical, can also be different, all can be determined according to actual needs;Working node utilizes its work nest having that task is processed, the work nest that any task is not processed by vacant working groove and current time.
Wherein, it should be noted that, the subject of the step of a kind of method for scheduling task that the embodiment of the present invention provides can be the scheduling node (JobTracker) in Hadoop cluster, and a kind of task scheduling apparatus being provided below specifically can be provided in the device in scheduling node, scheduling node can pass through heart beating connection (Heartbeat) and the whole working nodes (TaskTracker) comprised in cluster are monitored, when listening to any operative node and comprising solicited message, it is the solicited message getting this working node.
S2: calculate resource consumption feature and the residue task ratio of each task queue, wherein, the resource consumption of arbitrary task queue is characterized as and is calculated obtaining according to the CPU usage of being currently running in this task queue of task, memory usage and bandwidth usage, and the residue task ratio of arbitrary task queue is the ratio of off-duty task and whole tasks in this task queue.
Wherein, CPU usage is actually for the ratio of the cpu resource that takies of task that runs with whole cpu resources, in like manner, memory usage represents the ratio of memory source that the task of operation takies and full memory resource, bandwidth usage represents the bandwidth resources and the ratio of whole bandwidth resources that the task of operation takies, and can characterize corresponding task queue by the calculated resource consumption feature of CPU usage, memory usage and bandwidth usage and consume the situation of resource.
S3: choosing the product of its resource consumption feature in task queue and residue task ratio, to be not less than the task queue of other task queues be goal task queue, and utilizes vacant working groove that the off-duty task in goal task queue is processed.
Wherein, off-duty task such as is at the pending task, it is necessary to explanation, in the application, arbitrary task carries out process and is this task of operation.
It should be noted that, iff the resource consumption by being currently running in arbitrary task queue of task, namely resource consumption feature judges whether this task queue is in resource tense situation is incomplete, therefore, the application is additionally contemplates that in task queue the situation of off-duty task, i.e. residue task ratio, thus, consider the situation of off-duty task in the resource consumption of being currently running in arbitrary task queue of task and this task queue and determine whether this task queue is in resource tense situation, resource consumption feature is more big with the product of residue task ratio, show that the resource tense situation of corresponding task queue is more serious, more need in time corresponding task queue to be processed, therefore, the application passes through above-mentioned steps, choose the goal task queue that in task queue, resource tense situation is the most serious to process, thus priority treatment is badly in need of the off-duty task in processed task queue, not only better ensure that the fairness of task scheduling, and substantially increase tasks carrying efficiency, decrease the total time of tasks carrying.
In a kind of method for scheduling task that the embodiment of the present invention provides, as in figure 2 it is shown, utilize vacant working groove that the off-duty task in goal task queue is processed, it is possible to including:
S31: the ability characteristics of evaluation work node, wherein, ability characteristics is be calculated obtaining according to the property value that the CPU of working node, internal memory, hard disk and bandwidth are corresponding.
Wherein, property value corresponding to CPU, internal memory, hard disk and bandwidth can be CPU frequency, memory size, hard disk size and amount of bandwidth, by utilizing above-mentioned property value to be calculated the ability characteristics obtained, it is possible to characterizes the ability that task is processed by working node.
S32: calculate the resource occupation feature of the off-duty task comprised in goal task queue, wherein, in goal task queue, the resource occupation of arbitrary off-duty task is characterized as the ratio of all meansigma methodss of the work nest quantity that tasks take in work nest quantity that this off-duty task takies and goal task queue.
Above-mentioned resource occupation feature can characterize the occupation condition of off-duty task in task queue, and in other words, off-duty task needs the situation of the resource taken.
S33: choosing the off-duty task that in goal task queue, the ability characteristics of its resource occupation feature and working node matches is goal task, and utilizes vacant working groove that goal task is processed.
By above-mentioned steps disclosed in the present application, choose the off-duty task matched in goal task queue with the ability of working node, so that working node realizes the process for goal task smoothly, effectively prevent " weak bus performs big task ", namely the node execution that ability is more weak needs the situation of the task that resource is bigger to occur, and then is effectively increased tasks carrying efficiency.
Specifically, choosing the off-duty task that in goal task queue, the ability characteristics of its resource occupation feature and working node matches is goal task, it is possible to including:
The ability characteristics of working node and node capacity threshold value are compared, if the ability characteristics of working node is more than node capacity threshold value, then judge the off-duty task that whether there is its resource occupation feature in object queue more than resource occupation threshold value, if object queue exists its resource occupation feature off-duty task more than resource occupation threshold value, then by its resource occupation feature more than the off-duty task of resource occupation threshold value being chosen an off-duty task as goal task, if object queue is absent from its resource occupation feature off-duty task more than resource occupation threshold value, then judge the off-duty task that whether there is its resource occupation feature in object queue equal to resource occupation threshold value, if object queue exists its resource occupation feature off-duty task equal to resource occupation threshold value, then it is equal to by its resource occupation feature in the off-duty task of resource occupation threshold value and chooses an off-duty task as goal task, if object queue is absent from its resource occupation feature off-duty task equal to resource occupation threshold value, then judge the off-duty task that whether there is its resource occupation feature in object queue less than resource occupation threshold value, if object queue exists its resource occupation feature off-duty task less than resource occupation threshold value, then by its resource occupation feature less than the off-duty task of resource occupation threshold value being chosen an off-duty task as goal task;
If the ability characteristics of working node is equal to node capacity threshold value, then judge the off-duty task that whether there is its resource occupation feature in object queue equal to resource occupation threshold value, if object queue exists its resource occupation feature off-duty task equal to resource occupation threshold value, then it is equal to by its resource occupation feature in the off-duty task of resource occupation threshold value and chooses an off-duty task as goal task, if object queue is absent from its resource occupation feature off-duty task equal to resource occupation threshold value, then judge the off-duty task that whether there is its resource occupation feature in object queue less than resource occupation threshold value, if object queue exists its resource occupation feature off-duty task less than resource occupation threshold value, then by its resource occupation feature less than the off-duty task of resource occupation threshold value being chosen an off-duty task as goal task;
If the ability characteristics of working node is less than node capacity threshold value, then judge the off-duty task that whether there is its resource occupation feature in object queue less than resource occupation threshold value, if object queue exists its resource occupation feature off-duty task less than resource occupation threshold value, then by its resource occupation feature less than the off-duty task of resource occupation threshold value being chosen an off-duty task as goal task.
Wherein, node capacity threshold value and resource occupation threshold value can be determined according to actual needs, by all 1 can be elected as, specifically, if the ability characteristics of working node is more than node capacity threshold value, then working node is strong node, if the ability characteristics of working node is equal to node capacity threshold value, then working node is ordinary node, if the ability characteristics of working node is less than node capacity threshold value, then working node is weak bus, in like manner, if the resource occupation feature of off-duty task is more than resource occupation threshold value, then this task is big task, if the resource occupation feature of off-duty task is equal to resource occupation threshold value, then this task is common task, if the resource occupation feature of off-duty task is less than resource occupation threshold value, then this task is little task, thus, by the off-duty task in goal task queue being carried out classification and the ability of working node being carried out classification, the off-duty task that the ability with working node matches is obtained according to ability matching principle, thus substantially increasing tasks carrying efficiency.
Wherein, by meet the off-duty task of either condition chooses an off-duty task as goal task time; as by its resource occupation feature more than the off-duty task of resource occupation threshold value being chosen an off-duty task as goal task etc.; selection principle can be determined according to actual needs; as chosen according to FIFO principle by meeting in the off-duty task of this condition; or randomly select, all within protection scope of the present invention.
It addition, residue operation ratio can be calculated according to the following formula:
J i = slot i avg slot i
Wherein, JiQuantity slot for the off-duty task i operation slot takeniThe number average of the operation slot that all tasks take with object queueRatio.
In a kind of method for scheduling task that the embodiment of the present invention provides, it is also possible to including:
Whole tasks are divided into the first predetermined number task queue, and distribute the second predetermined number the work nest for the task in this task queue is processed for each task queue.
Wherein, the first predetermined number and the second predetermined number can be determined according to actual needs, and, the quantity for each task queue assignment groove can be the same or different.Namely, according to actual needs whole tasks can be divided into the first predetermined number task queue, and, according to being actually needed of comprising in each task queue of task, the work nest of whole working nodes is distributed in different task queues, wherein it is possible to distributed to the quantity of the work nest of different task queue by configuration file amendment.Illustrate, it is assumed that the quantity of there is currently of task is 3n, then all of task can be divided into 3 task queues, and each task queue has n task, remembers that 3 task queues are Q1、Q2And Q3, and, Q1={ a1,a2,…,an, Q2={ b1,b2,…,bn, Q3={ c1,c2,…,cn}.The quantity of working node is M, is designated as M={1,2 ..., m}, and the quantity of operation slot corresponding to working node is { k1,k2,…,km, namely first job node has k1Individual work nest, second working node has k2Individual work nest, by that analogy.And it is possible to be task queue Q1The work nest of distribution 40%, for task queue Q2With task queue Q3It is respectively allocated the work nest etc. of 30%.
On this basis, after certain work nest is assigned to arbitrary task queue, untreated any task is in idle condition or has processed need to processing of task and be in idle condition, just can obtain the solicited message of working node corresponding to this vacant working groove, perform step S1 to step S3, with by this work nest distribution to suitable task queue, if after this work nest completes process task in suitable task queue, can continue according to a kind of method for scheduling task that the embodiment of the present invention provides is that it distributes suitable task queue, if all task queue does not all need this work nest, then given back back in the task queue being assigned at first.
In a kind of method for scheduling task that the embodiment of the present invention provides, calculate the resource consumption feature of each task queue, it is possible to including: calculate the resource consumption feature of arbitrary task queue according to the following formula:
C Q i = r c p u × T c p u + r m e m × T m e m + r b a n d × T b a n d
Wherein,Represent the resource consumption feature of arbitrary task queue, TcpuRepresent the meansigma methods of the CPU usage of being currently running in this task queue of task, rcpuRepresent TcpuShared weight, TmemRepresent the meansigma methods of the memory usage of being currently running in this task queue of task, rmemRepresent TmemShared weight, TbandRepresent the meansigma methods of the bandwidth usage of being currently running in this task queue of task, rbandRepresent TbandShared weight, rcpu、rmemAnd rbandConcrete value can be determined according to actual needs, and rcpu+rmem+rband=1.
Wherein, obtain the meansigma methods of CPU usage, the meansigma methods of memory usage, bandwidth usage meansigma methods can pass through to check the running log of task, the above-mentioned resource taken when obtaining each task run, and take whole task take respective resources meansigma methods obtain.
It addition, the ability characteristics of evaluation work node, it is possible to including: the ability characteristics of evaluation work node according to the following formula:
N r e s o u r c e = w c u c p u avg c p u + w m u m e m avg m e m + w d u d i s k avg d i s k + w b u b a n d avg b a n d
Wherein, NresourceRepresent the ability characteristics of working node, ucpuRepresent the CPU frequency of working node, wcRepresent ucpuShared weight, umemRepresent the memory size of working node, wmRepresent umemShared weight, udiskRepresent the hard disk size of working node, wdRepresent udiskShared weight, ubandRepresent the amount of bandwidth of working node, wbRepresent ubandShared weight, avgcpuRepresent the meansigma methods of working node institute whole working node CPU frequencys in the cluster, avgmemRepresent the meansigma methods of working node institute whole working node memory sizes in the cluster, avgdiskRepresent the meansigma methods of working node institute whole working node hard disk size in the cluster, avgbandRepresent the meansigma methods of working node institute whole working node amount of bandwidth in the cluster.Wherein, wc、wm、wdAnd wbConcrete value can be determined according to actual needs, and, wc+wm+wd+wb=1.And the above-mentioned parameter for the ability characteristics of evaluation work node is static, all can directly obtain after Hadoop cluster initializes.
In a kind of method for scheduling task that the embodiment of the present invention provides, it is also possible to including:
The load condition of evaluation work node, wherein, load condition is calculated according to the CPU usage of working node, memory usage and bandwidth usage;
Judge that whether load condition is less than node load threshold value, if it is, calculate resource consumption feature and the residue task ratio of each task queue, if it is not, then refusal distributes task for working node.
Wherein, node load threshold value can be determined according to actual needs, by arranging node load threshold value, when the load condition of node is more than or equal to node load threshold value, the quantity of its vacant working groove can be set to 0, and refuse to distribute task for it, to avoid node overload.
Wherein, the load condition of evaluation work node, it is possible to including: the load condition of evaluation work node according to the following formula:
L=fcpu×Lcpu+fmem×Lmem+fband×Lband
Wherein, L represents the load condition of working node, LcpuRepresent the CPU usage of working node, fcpuRepresent LcpuShared weight, LmemRepresent the memory usage of working node, fmemRepresent LmemShared weight, LbandRepresent the bandwidth usage of working node, fbandRepresent LbandShared weight.Wherein, fcpu、fmemAnd fbandConcrete value can be determined according to actual needs, and fcpu+fmem+fband=1.
Corresponding with said method embodiment, the embodiment of the present invention additionally provides a kind of task scheduling apparatus, as it is shown on figure 3, may include that
Monitoring module 1, for obtaining the solicited message of working node, solicited message is that working node exists the information of the request task of transmission during vacant working groove;
Computing module 2, for calculating resource consumption feature and the residue task ratio of each task queue, wherein, the resource consumption of arbitrary task queue is characterized as and is calculated obtaining according to the CPU usage of being currently running in this task queue of task, memory usage and bandwidth usage, and the residue task ratio of arbitrary task queue is the ratio of off-duty task and whole tasks in this task queue;
Choosing module 3, it is goal task queue that the product for choosing its resource consumption feature in task queue and residue task ratio is not less than the task queue of other task queues, and utilizes vacant working groove that the task in goal task queue is processed.
A kind of task scheduling apparatus that the embodiment of the present invention also provides for, chooses module and may include that
Choosing unit, be used for: the ability characteristics of evaluation work node, wherein, ability characteristics is be calculated obtaining according to the property value that the CPU of working node, internal memory, hard disk and bandwidth are corresponding;Calculate the resource occupation feature of the off-duty task comprised in goal task queue, wherein, in goal task queue, the resource occupation of arbitrary off-duty task is characterized as the ratio of all meansigma methodss of the work nest quantity that tasks take in work nest quantity that this off-duty task takies and goal task queue;Choosing the off-duty task that in goal task queue, the ability characteristics of its resource occupation feature and working node matches is goal task, and utilizes vacant working groove that goal task is processed.
A kind of task scheduling apparatus that the embodiment of the present invention also provides for, chooses unit and may include that
Choose subelement, for: the ability characteristics of working node and node capacity threshold value are compared, if the ability characteristics of working node is more than node capacity threshold value, then judge the off-duty task that whether there is its resource occupation feature in object queue more than resource occupation threshold value, if it is, by its resource occupation feature more than the off-duty task of resource occupation threshold value being chosen a task as goal task;If the ability characteristics of working node is equal to node capacity threshold value, then judge the off-duty task that whether there is its resource occupation feature in object queue equal to resource occupation threshold value, a task is chosen as goal task if it is, be equal in the off-duty task of resource occupation threshold value by its resource occupation feature;If the ability characteristics of working node is less than node capacity threshold value, then judge the off-duty task that whether there is its resource occupation feature in object queue less than resource occupation threshold value, if it is, by its resource occupation feature less than the off-duty task of resource occupation threshold value being chosen a task as goal task.
A kind of task scheduling apparatus that the embodiment of the present invention also provides for, it is also possible to including:
Forward allocator module, for whole tasks are divided into the first predetermined number task queue, and distributes the second predetermined number the work nest for the task in this task queue is processed for each task queue.
A kind of task scheduling apparatus that the embodiment of the present invention also provides for, computing module may include that
Computing unit, for calculating the resource consumption feature of arbitrary task queue according to the following formula:
C Q i = r c p u × T c p u + r m e m × T m e m + r b a n d × T b a n d
Wherein,Represent the resource consumption feature of arbitrary task queue, TcpuRepresent the meansigma methods of the CPU usage of being currently running in this task queue of task, rcpuRepresent TcpuShared weight, TmemRepresent the meansigma methods of the memory usage of being currently running in this task queue of task, rmemRepresent TmemShared weight, TbandRepresent the meansigma methods of the bandwidth usage of being currently running in this task queue of task, rbandRepresent TbandShared weight.
A kind of task scheduling apparatus that the embodiment of the present invention also provides for, chooses unit and may include that
Computation subunit, for the ability characteristics of evaluation work node according to the following formula:
N r e s o u r c e = w c u c p u avg c p u + w m u m e m avg m e m + w d u d i s k avg d i s k + w b u b a n d avg b a n d
Wherein, NresourceRepresent the ability characteristics of working node, ucpuRepresent the CPU frequency of working node, wcRepresent ucpuShared weight, umemRepresent the memory size of working node, wmRepresent umemShared weight, udiskRepresent the hard disk size of working node, wdRepresent udiskShared weight, ubandRepresent the amount of bandwidth of working node, wbRepresent ubandShared weight, avgcpuRepresent the meansigma methods of working node institute whole working node CPU frequencys in the cluster, avgmemRepresent the meansigma methods of working node institute whole working node memory sizes in the cluster, avgdiskRepresent the meansigma methods of working node institute whole working node hard disk size in the cluster, avgbandRepresent the meansigma methods of working node institute whole working node amount of bandwidth in the cluster.
A kind of task scheduling apparatus that the embodiment of the present invention also provides for, it is also possible to including:
Overload monitor module, is used for: the load condition of evaluation work node, and wherein, load condition is calculated according to the CPU usage of working node, memory usage and bandwidth usage;Judge that whether load condition is less than node load threshold value, if it is, calculate resource consumption feature and the residue task ratio of each task queue, if it is not, then refusal distributes task for working node.
A kind of task scheduling apparatus that the embodiment of the present invention also provides for, overload monitor module may include that
Monitoring unit, for the load condition of evaluation work node according to the following formula:
L=fcpu×Lcpu+fmem×Lmem+fband×Lband
Wherein, L represents the load condition of working node, LcpuRepresent the CPU usage of working node, fcpuRepresent LcpuShared weight, LmemRepresent the memory usage of working node, fmemRepresent LmemShared weight, LbandRepresent the bandwidth usage of working node, fbandRepresent LbandShared weight.
In a kind of task scheduling apparatus that the embodiment of the present invention provides, the explanation of relevant portion refers to the detailed description of corresponding part in a kind of method for scheduling task that the embodiment of the present invention provides, and does not repeat them here.
Described above to the disclosed embodiments, makes those skilled in the art be capable of or uses the present invention.The multiple amendment of these embodiments be will be apparent from for a person skilled in the art, and generic principles defined herein can without departing from the spirit or scope of the present invention, realize in other embodiments.Therefore, the present invention is not intended to be limited to the embodiments shown herein, and is to fit to the widest scope consistent with principles disclosed herein and features of novelty.

Claims (10)

1. a method for scheduling task, it is characterised in that including:
Obtaining the solicited message of working node, described solicited message is that described working node exists the information of the request task of transmission during vacant working groove;
Calculate resource consumption feature and the residue task ratio of each task queue, wherein, the resource consumption of arbitrary task queue is characterized as and is calculated obtaining according to the CPU usage of being currently running in this task queue of task, memory usage and bandwidth usage, and the residue task ratio of arbitrary task queue is the ratio of off-duty task and whole tasks in this task queue;
Choosing the product of its resource consumption feature in described task queue and residue task ratio, to be not less than the task queue of other task queues be goal task queue, and utilizes described vacant working groove that the off-duty task in described goal task queue is processed.
2. method according to claim 1, it is characterised in that utilize described vacant working groove that the off-duty task in described goal task queue is processed, including:
Calculating the ability characteristics of described working node, wherein, described ability characteristics is be calculated obtaining according to the property value that the CPU of described working node, internal memory, hard disk and bandwidth are corresponding;
Calculate the resource occupation feature of the off-duty task comprised in described goal task queue, wherein, in described goal task queue, the resource occupation of arbitrary off-duty task is characterized as the ratio of all meansigma methodss of the work nest quantity that tasks take in work nest quantity that this off-duty task takies and described goal task queue;
Choosing the off-duty task that in described goal task queue, the ability characteristics of its resource occupation feature and described working node matches is goal task, and utilizes described vacant working groove that described goal task is processed.
3. method according to claim 2, it is characterised in that choosing the off-duty task that in described goal task queue, the ability characteristics of its resource occupation feature and described working node matches is goal task, including:
The ability characteristics of described working node and node capacity threshold value are compared, if the ability characteristics of described working node is more than described node capacity threshold value, then judge the off-duty task that whether there is its resource occupation feature in described object queue more than resource occupation threshold value, if it is, by its resource occupation feature more than the off-duty task of resource occupation threshold value being chosen a task as goal task;
If the ability characteristics of described working node is equal to described node capacity threshold value, then judge the off-duty task that whether there is its resource occupation feature in described object queue equal to resource occupation threshold value, a task is chosen as goal task if it is, be equal in the off-duty task of resource occupation threshold value by its resource occupation feature;
If the ability characteristics of described working node is less than described node capacity threshold value, then judge the off-duty task that whether there is its resource occupation feature in described object queue less than resource occupation threshold value, if it is, by its resource occupation feature less than the off-duty task of resource occupation threshold value being chosen a task as goal task.
4. method according to claim 2, it is characterised in that also include:
Whole tasks are divided into the first predetermined number task queue, and distribute the second predetermined number the work nest for the task in this task queue is processed for each task queue.
5. method according to claim 2, it is characterised in that calculate the resource consumption feature of each task queue, including: calculate the resource consumption feature of arbitrary task queue according to the following formula:
C Q i = r c p u × T c p u + r m e m × T m e m + r b a n d × T b a n d
Wherein,Represent the resource consumption feature of arbitrary task queue, TcpuRepresent the meansigma methods of the CPU usage of being currently running in this task queue of task, rcpuRepresent TcpuShared weight, TmemRepresent the meansigma methods of the memory usage of being currently running in this task queue of task, rmemRepresent TmemShared weight, TbandRepresent the meansigma methods of the bandwidth usage of being currently running in this task queue of task, rbandRepresent TbandShared weight.
6. method according to claim 5, it is characterised in that calculate the ability characteristics of described working node, including: calculate the ability characteristics of described working node according to the following formula:
N r e s o u r c e = w c u c p u avg c p u + w m u m e m avg m e m + w d u d i s k avg d i s k + w b u b a n d avg b a n d
Wherein, NresourceRepresent the ability characteristics of described working node, ucpuRepresent the CPU frequency of described working node, wcRepresent ucpuShared weight, umemRepresent the memory size of described working node, wmRepresent umemShared weight, udiskRepresent the hard disk size of described working node, wdRepresent udiskShared weight, ubandRepresent the amount of bandwidth of described working node, wbRepresent ubandShared weight, avgcpuRepresent described working node the meansigma methods of whole working node CPU frequencys in the cluster, avgmemRepresent described working node the meansigma methods of whole working node memory sizes in the cluster, avgdiskRepresent described working node the meansigma methods of whole working node hard disk size in the cluster, avgbandRepresent described working node the meansigma methods of whole working node amount of bandwidth in the cluster.
7. the method according to any one of claim 1 to 6, it is characterised in that also include:
Calculating the load condition of described working node, wherein, described load condition is calculated according to the CPU usage of described working node, memory usage and bandwidth usage;
Judge that whether described load condition is less than node load threshold value, if it is, calculate resource consumption feature and the residue task ratio of each task queue, if it is not, then refusal is described working node distribution task.
8. method according to claim 7, it is characterised in that calculate the load condition of described working node, including: calculate the load condition of described working node according to the following formula:
L=fcpu×Lcpu+fmem×Lmem+fband×Lband
Wherein, L represents the load condition of described working node, LcpuRepresent the CPU usage of described working node, fcpuRepresent LcpuShared weight, LmemRepresent the memory usage of described working node, fmemRepresent LmemShared weight, LbandRepresent the bandwidth usage of described working node, fbandRepresent LbandShared weight.
9. a task scheduling apparatus, it is characterised in that including:
Monitoring module, for obtaining the solicited message of working node, described solicited message is that described working node exists the information of the request task of transmission during vacant working groove;
Computing module, for calculating resource consumption feature and the residue task ratio of each task queue, wherein, the resource consumption of arbitrary task queue is characterized as and is calculated obtaining according to the CPU usage of being currently running in this task queue of task, memory usage and bandwidth usage, and the residue task ratio of arbitrary task queue is the ratio of off-duty task and whole tasks in this task queue;
Choose module, it is goal task queue that product for choosing its resource consumption feature in described task queue and residue task ratio is not less than the task queue of other task queues, and utilizes described vacant working groove that the off-duty task in described goal task queue is processed.
10. device according to claim 9, it is characterised in that choose module and include:
Choosing unit, be used for: calculate the ability characteristics of described working node, wherein, described ability characteristics is be calculated obtaining according to the property value that the CPU of described working node, internal memory, hard disk and bandwidth are corresponding;Calculate the resource occupation feature of the off-duty task comprised in described goal task queue, wherein, in described goal task queue, the resource occupation of arbitrary off-duty task is characterized as the ratio of all meansigma methodss of the work nest quantity that tasks take in work nest quantity that this off-duty task takies and described goal task queue;Choosing the off-duty task that in described goal task queue, the ability characteristics of its resource occupation feature and described working node matches is goal task, and utilizes described vacant working groove that described goal task is processed.
CN201610029299.2A 2016-01-15 2016-01-15 A kind of method for scheduling task and device Active CN105718317B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610029299.2A CN105718317B (en) 2016-01-15 2016-01-15 A kind of method for scheduling task and device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610029299.2A CN105718317B (en) 2016-01-15 2016-01-15 A kind of method for scheduling task and device

Publications (2)

Publication Number Publication Date
CN105718317A true CN105718317A (en) 2016-06-29
CN105718317B CN105718317B (en) 2019-05-07

Family

ID=56147621

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610029299.2A Active CN105718317B (en) 2016-01-15 2016-01-15 A kind of method for scheduling task and device

Country Status (1)

Country Link
CN (1) CN105718317B (en)

Cited By (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106201721A (en) * 2016-07-12 2016-12-07 浪潮(北京)电子信息产业有限公司 A kind of internal memory dynamic adjusting method based on Intel Virtualization Technology and system
CN106776025A (en) * 2016-12-16 2017-05-31 郑州云海信息技术有限公司 A kind of computer cluster job scheduling method and its device
CN107168806A (en) * 2017-06-29 2017-09-15 上海联影医疗科技有限公司 Resource regulating method, system and the computer equipment of distribution scheduling machine
CN107291544A (en) * 2017-08-03 2017-10-24 山东浪潮云服务信息科技有限公司 Method and device, the distributed task scheduling execution system of task scheduling
CN107438284A (en) * 2017-09-12 2017-12-05 北京邮电大学 Resource allocation methods and device
CN108228349A (en) * 2017-12-26 2018-06-29 北京市商汤科技开发有限公司 For handling the method for task, system and storage medium
CN108270693A (en) * 2017-12-29 2018-07-10 珠海国芯云科技有限公司 The adaptive optimization leading method and device of website visiting
CN108540559A (en) * 2018-04-16 2018-09-14 北京航空航天大学 A kind of SDN controllers for supporting IPSec VPN load balancing
WO2018177036A1 (en) * 2017-03-31 2018-10-04 平安科技(深圳)有限公司 Database data archiving method and system, terminal, and computer readable storage medium
CN109947565A (en) * 2019-03-08 2019-06-28 北京百度网讯科技有限公司 Method and apparatus for distributing calculating task
CN110119315A (en) * 2019-05-14 2019-08-13 思力科(深圳)电子科技有限公司 Rendering method, relevant device and system based on block chain
WO2019205370A1 (en) * 2018-04-24 2019-10-31 平安科技(深圳)有限公司 Electronic device, task distribution method and storage medium
CN110609742A (en) * 2019-09-25 2019-12-24 苏州浪潮智能科技有限公司 Method and device for configuring queues of Kubernetes scheduler
CN110838987A (en) * 2019-10-08 2020-02-25 福建天泉教育科技有限公司 Queue current limiting method and storage medium
CN112269650A (en) * 2020-12-18 2021-01-26 北京达佳互联信息技术有限公司 Task scheduling method and device, electronic equipment and storage medium
CN112272148A (en) * 2020-10-15 2021-01-26 新华三信息安全技术有限公司 Multi-priority queue management method, device and storage medium
CN112783651A (en) * 2021-01-22 2021-05-11 苏州浪潮智能科技有限公司 Load balancing scheduling method, medium and device for vGPU of cloud platform
CN113014663A (en) * 2021-03-12 2021-06-22 中南大学 Task and resource matching method supporting cross-node computing task survivability and succession

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101126992A (en) * 2006-08-15 2008-02-20 国际商业机器公司 Method and system for dispensing multiple tasks at multiple node of network
CN102902587A (en) * 2011-07-28 2013-01-30 中国移动通信集团四川有限公司 Distribution type task scheduling method, distribution type task scheduling system and distribution type task scheduling device
US20130326537A1 (en) * 2012-06-05 2013-12-05 International Business Machines Corporation Dependency management in task scheduling
CN103945548A (en) * 2014-04-29 2014-07-23 西安电子科技大学 Resource distribution system and task/service scheduling method in C-RAN

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101126992A (en) * 2006-08-15 2008-02-20 国际商业机器公司 Method and system for dispensing multiple tasks at multiple node of network
US20080046895A1 (en) * 2006-08-15 2008-02-21 International Business Machines Corporation Affinity dispatching load balancer with precise CPU consumption data
CN102902587A (en) * 2011-07-28 2013-01-30 中国移动通信集团四川有限公司 Distribution type task scheduling method, distribution type task scheduling system and distribution type task scheduling device
US20130326537A1 (en) * 2012-06-05 2013-12-05 International Business Machines Corporation Dependency management in task scheduling
CN103945548A (en) * 2014-04-29 2014-07-23 西安电子科技大学 Resource distribution system and task/service scheduling method in C-RAN

Cited By (28)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106201721A (en) * 2016-07-12 2016-12-07 浪潮(北京)电子信息产业有限公司 A kind of internal memory dynamic adjusting method based on Intel Virtualization Technology and system
CN106201721B (en) * 2016-07-12 2019-05-10 浪潮(北京)电子信息产业有限公司 A kind of memory dynamic adjusting method and system based on virtualization technology
CN106776025A (en) * 2016-12-16 2017-05-31 郑州云海信息技术有限公司 A kind of computer cluster job scheduling method and its device
WO2018177036A1 (en) * 2017-03-31 2018-10-04 平安科技(深圳)有限公司 Database data archiving method and system, terminal, and computer readable storage medium
CN107168806A (en) * 2017-06-29 2017-09-15 上海联影医疗科技有限公司 Resource regulating method, system and the computer equipment of distribution scheduling machine
CN107168806B (en) * 2017-06-29 2021-01-12 上海联影医疗科技股份有限公司 Resource scheduling method and system of distribution scheduler and computer equipment
CN107291544A (en) * 2017-08-03 2017-10-24 山东浪潮云服务信息科技有限公司 Method and device, the distributed task scheduling execution system of task scheduling
CN107291544B (en) * 2017-08-03 2020-07-28 浪潮云信息技术股份公司 Task scheduling method and device and distributed task execution system
CN107438284A (en) * 2017-09-12 2017-12-05 北京邮电大学 Resource allocation methods and device
CN107438284B (en) * 2017-09-12 2020-04-28 北京邮电大学 Resource allocation method and device
CN108228349A (en) * 2017-12-26 2018-06-29 北京市商汤科技开发有限公司 For handling the method for task, system and storage medium
CN108270693A (en) * 2017-12-29 2018-07-10 珠海国芯云科技有限公司 The adaptive optimization leading method and device of website visiting
CN108540559A (en) * 2018-04-16 2018-09-14 北京航空航天大学 A kind of SDN controllers for supporting IPSec VPN load balancing
CN108540559B (en) * 2018-04-16 2020-12-18 北京航空航天大学 SDN controller supporting IPSec VPN load balancing
WO2019205370A1 (en) * 2018-04-24 2019-10-31 平安科技(深圳)有限公司 Electronic device, task distribution method and storage medium
CN109947565A (en) * 2019-03-08 2019-06-28 北京百度网讯科技有限公司 Method and apparatus for distributing calculating task
CN109947565B (en) * 2019-03-08 2021-10-15 北京百度网讯科技有限公司 Method and apparatus for distributing computing tasks
US11210131B2 (en) 2019-03-08 2021-12-28 Beijing Baidu Netcom Science And Technology Co., Ltd. Method and apparatus for assigning computing task
CN110119315A (en) * 2019-05-14 2019-08-13 思力科(深圳)电子科技有限公司 Rendering method, relevant device and system based on block chain
CN110609742A (en) * 2019-09-25 2019-12-24 苏州浪潮智能科技有限公司 Method and device for configuring queues of Kubernetes scheduler
CN110838987A (en) * 2019-10-08 2020-02-25 福建天泉教育科技有限公司 Queue current limiting method and storage medium
CN110838987B (en) * 2019-10-08 2022-07-05 福建天泉教育科技有限公司 Queue current limiting method and storage medium
CN112272148A (en) * 2020-10-15 2021-01-26 新华三信息安全技术有限公司 Multi-priority queue management method, device and storage medium
CN112272148B (en) * 2020-10-15 2022-05-27 新华三信息安全技术有限公司 Multi-priority queue management method, device and storage medium
CN112269650A (en) * 2020-12-18 2021-01-26 北京达佳互联信息技术有限公司 Task scheduling method and device, electronic equipment and storage medium
CN112783651A (en) * 2021-01-22 2021-05-11 苏州浪潮智能科技有限公司 Load balancing scheduling method, medium and device for vGPU of cloud platform
CN113014663A (en) * 2021-03-12 2021-06-22 中南大学 Task and resource matching method supporting cross-node computing task survivability and succession
CN113014663B (en) * 2021-03-12 2022-03-18 中南大学 Task and resource matching method supporting cross-node computing task survivability and succession

Also Published As

Publication number Publication date
CN105718317B (en) 2019-05-07

Similar Documents

Publication Publication Date Title
CN105718317A (en) Task scheduling method and task scheduling device
CN105718479B (en) Execution strategy generation method and device under cross-IDC big data processing architecture
CN108270805B (en) Resource allocation method and device for data processing
CN106445675B (en) B2B platform distributed application scheduling and resource allocation method
CN105159782A (en) Cloud host based method and apparatus for allocating resources to orders
CN104901989B (en) A kind of Site Service offer system and method
CN103761146B (en) A kind of method that MapReduce dynamically sets slots quantity
CN105589750A (en) CPU (Central Processing Unit) resource scheduling method and server
CN108111337B (en) Method and equipment for arbitrating main nodes in distributed system
CN105022668B (en) Job scheduling method and system
CN106790706A (en) A kind of method of data resource optimization, platform and system
CN111246586B (en) Method and system for distributing smart grid resources based on genetic algorithm
CN105404554B (en) Method and apparatus for Storm stream calculation frame
CN106528288A (en) Resource management method, device and system
CN112996056A (en) Method and device for unloading time delay optimized computing task under cloud edge cooperation
CN104301241B (en) A kind of SOA dynamic load distributing methods and system
CN103744735B (en) The dispatching method and device of a kind of multi-core resource
CN109815019A (en) Method for scheduling task, device, electronic equipment and readable storage medium storing program for executing
CN106034047B (en) Data processing method and device
Murata et al. A distributed and cooperative load balancing mechanism for large-scale P2P systems
CN111176840A (en) Distributed task allocation optimization method and device, storage medium and electronic device
Shah et al. Agent based priority heuristic for job scheduling on computational grids
Gong et al. Computation offloading-based task scheduling in the vehicular communication environment for computation-intensive vehicular tasks
CN103685558A (en) Wireless sensor network and task processing method thereof
CN106302241A (en) Online message array dispatching method and device

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
CB03 Change of inventor or designer information

Inventor after: Wang Hongli

Inventor after: Su Zhiyuan

Inventor after: Qi Kaiyuan

Inventor before: Su Zhiyuan

Inventor before: Qi Kaiyuan

CB03 Change of inventor or designer information
GR01 Patent grant
GR01 Patent grant