CN113961320A - Adaptive gain cluster scheduling control method - Google Patents
Adaptive gain cluster scheduling control method Download PDFInfo
- Publication number
- CN113961320A CN113961320A CN202111084842.6A CN202111084842A CN113961320A CN 113961320 A CN113961320 A CN 113961320A CN 202111084842 A CN202111084842 A CN 202111084842A CN 113961320 A CN113961320 A CN 113961320A
- Authority
- CN
- China
- Prior art keywords
- task
- tasks
- time
- cluster
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 21
- 230000003044 adaptive effect Effects 0.000 title claims description 15
- 239000003016 pheromone Substances 0.000 claims abstract description 30
- 238000012163 sequencing technique Methods 0.000 claims abstract description 5
- 241000257303 Hymenoptera Species 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 9
- 238000004891 communication Methods 0.000 claims description 4
- 230000001276 controlling effect Effects 0.000 claims description 4
- 230000001105 regulatory effect Effects 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 abstract description 3
- 230000009545 invasion Effects 0.000 abstract 1
- 238000005516 engineering process Methods 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 239000003795 chemical substances by application Substances 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/48—Program initiating; Program switching, e.g. by interrupt
- G06F9/4806—Task transfer initiation or dispatching
- G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
- G06F9/4881—Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
- G06F9/5038—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering the execution order of a plurality of tasks, e.g. taking priority or time dependency constraints into consideration
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
- G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
- G06F9/505—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals considering the load
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a self-adaptive gain cluster scheduling control method, which relates to the technical field of cluster scheduling control and comprises the following steps: the method comprises the steps of calibrating the predicted execution time of the operation in advance, establishing an operation execution list corresponding to the operation execution time, obtaining the earliest completion time of all tasks, sequencing all the tasks according to the earliest completion time, sequentially adding the tasks into a task cache queue from the early to the late, updating pheromones of resource nodes in a cluster, updating cluster resource nodes based on the calibrated task, and updating local pheromones of a path. The invention has small invasion to the system in the cluster system environment, has the advantages of parallelism, effective global search balance capability, simple calculation and good robustness, and in addition, on the premise of ensuring the safety and the reliability of the cluster system, the task is executed by selecting the execution node through the scheduling node, the task is prevented from being executed on the fixed cluster node all the time, the load of the cluster system is balanced, and the task scheduling difficulty is reduced.
Description
Technical Field
The invention relates to the technical field of cluster scheduling control, in particular to a self-adaptive gain cluster scheduling control method.
Background
With the gradual expansion of cluster scale and the increasing improvement of performance, the storage technology, the load balancing and the scheduling strategy of the cluster become the hot and difficult points of cluster research, and how to improve the reliability of a large storage system and keep data continuously and effectively accessed becomes the system bottleneck to be solved at present. The cluster RAID system is the fusion of the traditional RAID system technology and the cluster storage system, and has high reliability. One or more disks can be prevented from failing by the RAID data redundancy relationship established among the network storage nodes, and even the phenomenon that data cannot be accessed due to the failure of a single node or a disk cabinet can be avoided.
The cluster system is realized on the basis of resource sharing, and the resources in the whole system network are efficiently shared and operated through the job scheduling technology and the management of the resources, so that the aims of improving the system throughput and the system resource utilization rate are fulfilled, and high performance is obtained. Therefore, the effective utilization of resources is a key problem of cluster system software research, and the job scheduling technology is an effective way for improving the computing efficiency of the whole system and realizing resource sharing. However, currently, there are problems that when a high-performance computing system is used by multiple users, the system utilization rate and the user service quality are balanced, and it is difficult to simultaneously guarantee a high utilization rate and a short job waiting time, for example, pursuing a short job waiting time results in a low utilization rate, and pursuing a high utilization rate results in a long job waiting time of the user, which is mostly caused by a scheduling policy.
The aim of the load balancing scheduling strategy in task scheduling is to balance the load of each node while distributing user operation among each node through a certain scheduling strategy, so that the utilization rate of each node is maximized, and further, the parallel computation of a system is realized to reduce the response time of user tasks. Therefore, considering load balancing in the scheduling strategy is a very critical factor for improving the system throughput and simultaneously reducing the job waiting time.
Therefore, a method for controlling adaptive gain cluster scheduling is needed.
An effective solution to the problems in the related art has not been proposed yet.
Disclosure of Invention
The present invention provides a method for controlling adaptive gain cluster scheduling to overcome the above technical problems in the related art.
The technical scheme of the invention is realized as follows:
a self-adaptive gain cluster scheduling control method comprises the following steps:
step S1, calibrating the expected execution time of the operation in advance, and establishing the corresponding operation execution list, wherein the list comprises a calibration task queue set and a task buffer queue;
step S2, acquiring the earliest completion time of all tasks, sequencing all the tasks according to the earliest completion time, and sequentially adding the tasks into a task cache queue from morning to evening;
step S3, the pheromone tau is carried out on the resource nodes in the cluster at the t +1 momentij(t +1) update, expressed as:
τij(t+1)=(1-ρ)*τij(t)+ρΔτij(t);
where ρ is the sum of pheromone volatilization factors, τij(t) indicates the pheromone on path ij at time t, Δ τij(t) is the local increment of the pheromone;
step S4, performing cluster resource nodes based on the calibration task, which is expressed as:
wherein, tauij(t) is the pheromone on path ij,representing the Q value of the first node moving to the jth node, wherein alpha and beta are weight factors;
step S5, the local pheromone of the path is updated, and is expressed as:
wherein eta is the volatilization rate of the pheromone,mu is a regulating factor for the processing capacity of the cluster resource node;
step S6, determining whether the task buffer queue is empty, and calibrating T to T +1 and determining that the current iteration time T is equal to the maximum iteration time T,
and step S7, acquiring node scheduling schemes corresponding to all tasks.
Wherein, still include the following step:
step S101, calibrating task execution time and aiming at given task tiE.g. T, assigned to a certain node PEjHas a running time of PEtiExpressed as:
wherein m isiAs task tiSpe (j) is the node PEjThe processing speed of (2);
step S102, calibrating the earliest start time starti(ti,PE(ti) For any task t)iE T, expressed as:
starti(ti,PE(ti))=max{startj+tij+ti,startPE+tPE};
step S103, calibrating the earliest completion time finishi(ti,PE(ti) For any task t)iE T, expressed as:
finishi(ti,PE(ti))=starti+ti。
wherein, the Δ τij(t) is the local increment of the pheromone, expressed as:
wherein, comijIs the communication time of path ij.
Wherein said sorting all tasks according to earliest completion time comprises the steps of:
step S201, calibrating K ants to select K tasks from a task cache queue;
and if the number of the tasks in the task cache queue is less than K, sending the ant number which is the same as the number of the tasks in the task cache queue.
The method for judging whether the task buffer queue is empty comprises the following steps:
step S601, if the current task cache queue is empty, calibrating T to T +1 and determining that the current iteration time T is equal to the maximum iteration time T;
step S602, if the current task cache queue is not empty, the scheduling scheme of the current K tasks is stored, and the K tasks are selected from the task cache queue by calibrating K ants.
Wherein, the judging that the current iteration time T is equal to the maximum iteration time T comprises the following steps:
step S603, if the current iteration time T is equal to the maximum iteration time T, outputting a node scheduling scheme corresponding to all tasks;
step S604, if the current iteration time T is not equal to the maximum iteration time T, K ants are calibrated to select K tasks from the task cache queue.
The invention has the beneficial effects that:
the invention relates to a self-adaptive gain cluster scheduling control method, which comprises the steps of calibrating the predicted execution time of a job in advance, establishing a job execution list corresponding to the predicted execution time, obtaining the earliest completion time of all tasks, sequencing all tasks according to the earliest completion time, sequentially adding task cache queues from the morning to the evening, updating pheromones of resource nodes in a cluster, updating local pheromones of a path based on the clustered resource nodes of the calibrated task, judging whether the task cache queues are empty or not, judging the current iteration number to be equal to the maximum iteration number, obtaining a node scheduling scheme corresponding to all tasks, realizing adaptive gain cluster scheduling control, having the advantages of small intrusion to a system under a cluster system environment, parallelism, effective global search balance capability, simple calculation and good robustness, in addition, on the premise of ensuring the safety and reliability of the cluster system, the scheduling node selects the execution node to execute the task, so that the task is prevented from being executed on the fixed cluster node all the time, the load of the cluster system is balanced, and the task scheduling difficulty is reduced.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
Fig. 1 is a flowchart illustrating an adaptive gain cluster scheduling control method according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments that can be derived by one of ordinary skill in the art from the embodiments given herein are intended to be within the scope of the present invention.
According to an embodiment of the present invention, there is provided a method for controlling adaptive gain cluster scheduling.
As shown in fig. 1, the adaptive gain cluster scheduling control method according to the embodiment of the present invention includes the following steps:
step S1, calibrating the expected execution time of the operation in advance, and establishing the corresponding operation execution list, wherein the list comprises a calibration task queue set and a task buffer queue;
step S2, acquiring the earliest completion time of all tasks, sequencing all the tasks according to the earliest completion time, and sequentially adding the tasks into a task cache queue from morning to evening;
step S3, the pheromone tau is carried out on the resource nodes in the cluster at the t +1 momentij(t +1) update, expressed as:
τij(t+1)=(1-ρ)*τij(t)+ρΔτij(t);
where ρ is the sum of pheromone volatilization factors, τij(t) indicates the pheromone on path ij at time t, Δ τij(t) is the local increment of the pheromone;
step S4, performing cluster resource nodes based on the calibration task, which is expressed as:
wherein, tauij(t) is the pheromone on path ij,representing the Q value of the first node moving to the jth node, wherein alpha and beta are weight factors;
step S5, the local pheromone of the path is updated, and is expressed as:
wherein eta is the volatilization rate of the pheromone,mu is a regulating factor for the processing capacity of the cluster resource node;
step S6, determining whether the task buffer queue is empty, and calibrating T to T +1 and determining that the current iteration time T is equal to the maximum iteration time T,
and step S7, acquiring node scheduling schemes corresponding to all tasks.
Wherein, still include the following step:
step S101, calibrating task execution time and aiming at given task tiE.g. T, assigned to a certain node PEjHas a running time of PEtiExpressed as:
wherein m isiAs task tiSpe (j) is the node PEjThe processing speed of (2);
step S102, calibrating the earliest start time starti(ti,PE(ti) For any task t)iE T, expressed as:
starti(ti,PE(ti))=max{startj+tij+ti,startPE+tPE};
step S103, calibrating the earliest completion time finishi(ti,PE(ti) For any task t)iE T, expressed as:
finishi(ti,PE(ti))=starti+ti。
wherein, the Δ τij(t) is the local increment of the pheromone, expressed as:
wherein, comijIs the communication time of path ij.
Wherein said sorting all tasks according to earliest completion time comprises the steps of:
step S201, calibrating K ants to select K tasks from a task cache queue;
and if the number of the tasks in the task cache queue is less than K, sending the ant number which is the same as the number of the tasks in the task cache queue.
The method for judging whether the task buffer queue is empty comprises the following steps:
step S601, if the current task cache queue is empty, calibrating T to T +1 and determining that the current iteration time T is equal to the maximum iteration time T;
step S602, if the current task cache queue is not empty, the scheduling scheme of the current K tasks is stored, and the K tasks are selected from the task cache queue by calibrating K ants.
Wherein, the judging that the current iteration time T is equal to the maximum iteration time T comprises the following steps:
step S603, if the current iteration time T is equal to the maximum iteration time T, outputting a node scheduling scheme corresponding to all tasks;
step S604, if the current iteration time T is not equal to the maximum iteration time T, K ants are calibrated to select K tasks from the task cache queue.
In addition, in the technical scheme, the method further comprises the following steps:
calibrating task execution time for a given task tiE.g. T, assigned to a certain node PEjHas a running time of PEtiExpressed as:
wherein m isiAs task tiCalculated amount of (spe)(j) Is a node PEjThe processing speed of (2);
calibrating data transmission time by tijRepresenting a task tiAnd tjWhen two tasks are allocated on the same processing node, the transmission time is 0, otherwise, the ratio of the data transmission amount to the communication bandwidth of the two processing nodes is expressed as:
calibrating task entry path for any task tiE.g. T, the task entry path is entry task TinReach the current task tiThe longest path length of (1) is the In-pathiExpressed as:
calibrating task exit path for any task tiE.g. T, the task exit path is the current task TiTo the egress task toutThe longest path length of (1) is Out-pathiExpressed as:
demarcating the earliest start time, for any task tiE T, the earliest start time of the new task is the earliest completion time of all the predecessor tasks or the earliest idle time of the distributed nodes, and is marked as starti(ti,PE(ti) Expressed as:
starti(ti,PE(ti))=max{startj+tij+ti,startPE+tPE};
demarcating earliest completion time, for any task tiE.g. T, the earliest completion time of which means to put the task TiTo the processor PE (t)i) Expressed as:
finishi(ti,PE(ti))=starti+ti;
in addition, specifically, when the Agent is applied, alpha and beta are weight factors and satisfy the sum of 1, and setting alpha and beta to different values can effectively balance the Agent between the local optimizing capacity and the global optimizing capacity.
In addition, for the above step S6, it is determined whether the task buffer queue is empty, and T is marked as T +1 and the current iteration number T is determined to be equal to the maximum iteration number T, which is applied,
judging whether the task cache queue is empty, if so, turning to the step S7, otherwise, storing the scheduling schemes of the current K tasks, and turning to the step S201;
in addition, specifically, the step of calibrating T to T +1 and judging that the current iteration time T is equal to the maximum iteration time T includes the following steps: if the node scheduling schemes are equal to the preset scheduling schemes, the algorithm is ended, and node scheduling schemes corresponding to all tasks are output; otherwise, the operation is continued in step S201.
In summary, with the above technical solution of the present invention, by calibrating the predicted execution time of the job in advance and establishing the job execution list corresponding to the predicted execution time, obtaining the earliest completion time of all the tasks, sorting all the tasks according to the earliest completion time, sequentially adding the tasks to the task cache queues from the beginning to the end, performing pheromone update on resource nodes in the cluster, performing cluster resource node based on the calibrated task, updating local pheromones of the path, determining whether the task cache queue is empty, determining the current iteration number equal to the maximum iteration number, obtaining the node scheduling scheme corresponding to all the tasks, implementing adaptive gain cluster scheduling control, not only having little intrusion to the system in the cluster system environment, but also having advantages of parallelism, effective global search balance capability, simple calculation, and good robustness, in addition, on the premise of ensuring the safety and reliability of the cluster system, the scheduling node selects the execution node to execute the task, so that the task is prevented from being executed on the fixed cluster node all the time, the load of the cluster system is balanced, and the task scheduling difficulty is reduced.
While the foregoing is directed to the preferred embodiment of the present invention, other and further embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This application is intended to cover any variations, uses, or adaptations of the disclosure following, in general, the principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.
It will be understood that the present disclosure is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.
Claims (6)
1. A method for controlling adaptive gain cluster scheduling, comprising the steps of:
pre-calibrating the expected execution time of the operation, and establishing an operation execution list corresponding to the operation execution list, wherein the operation execution list comprises a calibration task queue set and a task cache queue;
acquiring the earliest completion time of all tasks, sequencing all the tasks according to the earliest completion time, and sequentially adding the tasks into a task cache queue from morning to evening;
performing pheromone tau on resource nodes in the cluster at the t +1 momentij(t +1) update, expressed as:
τij(t+1)=(1-ρ)*τij(t)+ρΔτij(t);
where ρ is the sum of pheromone volatilization factors, τij(t) indicates the pheromone on path ij at time t, Δ τij(t) is the local increment of the pheromone;
performing cluster resource nodes based on the calibration task, which is expressed as:
wherein, tauij(t) is the pheromone on path ij,representing the Q value of the first node moving to the jth node, wherein alpha and beta are weight factors;
the local pheromone of the path is updated and is represented as:
wherein eta is the volatilization rate of the pheromone,mu is a regulating factor for the processing capacity of the cluster resource node;
judging whether the task cache queue is empty, calibrating T to be T +1, and judging that the current iteration time T is equal to the maximum iteration time T;
and acquiring node scheduling schemes corresponding to all tasks.
2. The adaptive gain cluster scheduling control method of claim 1, further comprising the steps of:
calibrating task execution time for a given task tiE.g. T, assigned to a certain node PEjHas a running time of PEtiExpressed as:
wherein m isiAs task tiSpe (j) is the node PEjThe processing speed of (2);
calibrating an earliest start time starti(ti,PE(ti) For any task t)iE T, expressed as:
starti(ti,PE(ti))=max{startj+tij+ti,startPE+tPE};
calibrating earliest completion time finishi(ti,PE(ti) For any task t)iE T, expressed as:
finishi(ti,PE(ti))=starti+ti。
4. The adaptive gain cluster scheduling control method of claim 3, wherein said ordering all tasks according to earliest completion time comprises the steps of:
calibrating K ants to select K tasks from the task cache queue;
and if the number of the tasks in the task cache queue is less than K, sending the ant number which is the same as the number of the tasks in the task cache queue.
5. The adaptive gain cluster scheduling control method of claim 4, wherein the determining whether the task buffer queue is empty comprises:
if the current task cache queue is empty, calibrating T as T +1 and judging that the current iteration time T is equal to the maximum iteration time T;
and if the current task cache queue is not empty, storing the scheduling scheme of the current K tasks, and switching to calibrate K ants to select K tasks from the task cache queue.
6. The adaptive gain cluster scheduling control method of claim 5, wherein said determining that the current iteration number T is equal to the maximum iteration number T comprises:
if the current iteration times T are equal to the maximum iteration times T, outputting node scheduling schemes corresponding to all tasks;
and if the current iteration time T is not equal to the maximum iteration time T, calibrating K ants to select K tasks from the task cache queue.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111084842.6A CN113961320A (en) | 2021-09-16 | 2021-09-16 | Adaptive gain cluster scheduling control method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111084842.6A CN113961320A (en) | 2021-09-16 | 2021-09-16 | Adaptive gain cluster scheduling control method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113961320A true CN113961320A (en) | 2022-01-21 |
Family
ID=79461749
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111084842.6A Pending CN113961320A (en) | 2021-09-16 | 2021-09-16 | Adaptive gain cluster scheduling control method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113961320A (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110764912A (en) * | 2019-10-25 | 2020-02-07 | 东北大学 | Self-adaptive task scheduler and method |
WO2020206705A1 (en) * | 2019-04-10 | 2020-10-15 | 山东科技大学 | Cluster node load state prediction-based job scheduling method |
CN112346839A (en) * | 2020-11-06 | 2021-02-09 | 东方红卫星移动通信有限公司 | Associated task scheduling method based on evolutionary algorithm |
-
2021
- 2021-09-16 CN CN202111084842.6A patent/CN113961320A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020206705A1 (en) * | 2019-04-10 | 2020-10-15 | 山东科技大学 | Cluster node load state prediction-based job scheduling method |
CN110764912A (en) * | 2019-10-25 | 2020-02-07 | 东北大学 | Self-adaptive task scheduler and method |
CN112346839A (en) * | 2020-11-06 | 2021-02-09 | 东方红卫星移动通信有限公司 | Associated task scheduling method based on evolutionary algorithm |
Non-Patent Citations (2)
Title |
---|
张新华: "基于强化学习和蚁群算法的协同依赖多任务网格集群调度", 长沙大学学报, vol. 30, no. 02, 15 March 2016 (2016-03-15), pages 50 - 53 * |
袁秀利等: "基于强化蚁群算法的任务DAG在线网格集群资源调度", 计算机测量与控制, vol. 23, no. 01, 25 January 2015 (2015-01-25), pages 287 - 290 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11509596B2 (en) | Throttling queue for a request scheduling and processing system | |
CN110297699B (en) | Scheduling method, scheduler, storage medium and system | |
US5504894A (en) | Workload manager for achieving transaction class response time goals in a multiprocessing system | |
JP3658420B2 (en) | Distributed processing system | |
CN109960585A (en) | A kind of resource regulating method based on kubernetes | |
US7388839B2 (en) | Methods, apparatus and computer programs for managing performance and resource utilization within cluster-based systems | |
US7712103B2 (en) | Method and system for managing load balancing in data processing system | |
US20110161294A1 (en) | Method for determining whether to dynamically replicate data | |
CN112799823B (en) | Online dispatching and scheduling method and system for edge computing tasks | |
CN111782627B (en) | Task and data cooperative scheduling method for wide-area high-performance computing environment | |
CN103294548A (en) | Distributed file system based IO (input output) request dispatching method and system | |
CN111143036A (en) | Virtual machine resource scheduling method based on reinforcement learning | |
US20050207439A1 (en) | Method and apparatus for dynamically scheduling requests | |
Maiti et al. | Internet of Things applications placement to minimize latency in multi-tier fog computing framework | |
CN113672391A (en) | Parallel computing task scheduling method and system based on Kubernetes | |
CN115878260A (en) | Low-carbon self-adaptive cloud host task scheduling system | |
CN112954012B (en) | Cloud task scheduling method based on improved simulated annealing algorithm of load | |
US12093730B2 (en) | Optimal dispatching of Function-as-a-Service in heterogeneous accelerator environments | |
Lu et al. | An efficient load balancing algorithm for heterogeneous grid systems considering desirability of grid sites | |
CN113672372A (en) | Multi-edge cooperative load balancing task scheduling method based on reinforcement learning | |
CN113961320A (en) | Adaptive gain cluster scheduling control method | |
CN114518951A (en) | Distributed stream data processing system virtualized resource flexible scheduling management research | |
KR102681134B1 (en) | Method and apparatus for resource allocation in cluster computing system | |
CN114968512A (en) | Energy perception scheduling method based on block chain intelligent contract | |
Valarmathi et al. | Differed service broker scheduling for data centres in cloud environment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |