Search

CN-121996382-A - Campus cloud platform elastic resource scheduling method and system

CN121996382ACN 121996382 ACN121996382 ACN 121996382ACN-121996382-A

Abstract

The invention discloses a campus cloud platform elastic resource scheduling method and a system, which relate to the technical field of resource management and comprise the steps of collecting execution state information and storage resource states, and generating execution state evidence through solidification after stabilization treatment; the method comprises the steps of carrying out cooperative evolution judgment on an execution state evidence, locating a stage switching boundary in the program execution process and dividing an execution stage interval, carrying out staged screening on the execution state evidence according to the execution stage interval to generate an execution state constraint condition, carrying out attenuation update processing on a continuous out-of-range condition by adopting a continuous deviation judgment algorithm based on the execution state constraint condition to generate an execution scheduling trigger state, carrying out progressive resource scheduling control according to the execution stage interval according to the execution scheduling trigger state, and updating the execution state evidence through a cooling constraint mark. The method and the device can improve the operation stability and the resource scheduling precision of the industrial cloud platform in a complex service scene on the premise of not depending on complex hardware transformation.

Inventors

  • TAN XINXIN

Assignees

  • 湖南大学

Dates

Publication Date
20260508
Application Date
20260127

Claims (10)

  1. 1. A campus cloud platform elastic resource scheduling method is characterized by comprising the steps of, Collecting execution state information and storage resource states, and solidifying after stabilization treatment to generate execution state evidence; performing execution state co-evolution judgment on the execution state evidence, positioning a phase switching boundary in the program execution process and dividing an execution phase interval; According to the execution stage interval, carrying out stage screening on the execution state evidence to generate an execution state constraint condition; based on the execution state constraint condition, adopting a continuous deviation judging algorithm to carry out attenuation update processing on the continuous out-of-range condition, and generating an execution scheduling triggering state; And executing progressive resource scheduling control according to the execution stage interval according to the execution scheduling triggering state, and updating the execution state evidence through the cooling constraint mark.
  2. 2. The campus cloud platform elastic resource scheduling method of claim 1, wherein the execution state information comprises thread scheduling state, blocking duration, I/O access occupation ratio and request response delay information; the storage resource states include an available storage capacity state, a storage access concurrency state, a storage access latency state, and a storage access queue state.
  3. 3. The method for scheduling elastic resources of campus cloud platform according to claim 2, wherein the step of generating the execution state evidence comprises the following steps, Performing sliding window processing on the execution state information according to time sequence, and eliminating transient mutation data only appearing at a single sampling point; And locking and storing the data with the consistent change trend of a plurality of continuous sampling periods in the execution state information processed by the sliding window, and generating the execution state evidence.
  4. 4. The method for scheduling elastic resources of campus cloud platform according to claim 3, wherein the step of performing the collaborative evolution discrimination of the execution state on the execution state evidence, locating the phase switching boundary in the process of program execution comprises the following steps, Performing time stamp binding and trend marking on the execution state evidence to generate an execution state marking track; Generating an execution state linkage direction sequence by judging the change direction of the execution state at adjacent moments based on the execution state marking track; And in the execution state linkage direction sequence, the arrangement sequence of the linkage directions in the adjacent time period is subjected to front-back comparison by a linkage direction sequence comparison and discrimination method, and a phase switching boundary is obtained.
  5. 5. The method of claim 4, wherein dividing the execution phase interval refers to segmenting a time range of the execution state evidence according to a phase switching boundary, and merging the execution state evidence to form the execution phase interval.
  6. 6. The method for scheduling elastic resources of campus cloud platform according to claim 5, wherein the step of performing staged screening on the execution state evidence according to the execution stage interval comprises the following steps, Slicing and grouping the execution state evidence according to each execution stage interval to obtain an execution state evidence fragment set corresponding to each stage; based on the execution state evidence segment set, merging repeated execution state marking paths in the segment into the shortest representative path by adopting an intra-stage mode compression algorithm, and obtaining a representative evolution path set; And selecting a path subset with the largest number of covering pieces in the representative evolution path set by adopting a path coverage priority algorithm, and acquiring a normal execution path coverage set.
  7. 7. The method of claim 6, wherein generating the execution state constraint is extracting a time-sequential state transition relation in a normal execution path coverage set for encapsulation and solidification, and generating the execution state constraint.
  8. 8. The method for scheduling elastic resources of campus cloud platform according to claim 7, wherein the continuous deviation determination algorithm is adopted to perform attenuation update processing on continuous out-of-range conditions based on execution state constraint conditions, and an execution scheduling trigger state is generated by the following steps, Mapping the execution state evidence into an execution state constraint condition, identifying whether crossing a stage switching boundary occurs or not, and recording a corresponding crossing event; Adopting a time attenuation memory method to perform time attenuation memory updating processing on the crossing events in adjacent time periods to generate a deviation evolution sequence; based on the deviation evolution sequence, a continuous deviation judgment algorithm is adopted to judge the time distribution form of the crossing event, and an execution scheduling trigger state is generated.
  9. 9. The method for scheduling elastic resources of campus cloud platform according to claim 8, wherein the step of executing progressive resource scheduling control according to execution phase intervals and updating execution state evidence through cooling constraint marks according to execution scheduling triggering state comprises the following steps, Reading an execution stage interval corresponding to an execution scheduling triggering state, executing progressive resource scheduling control on a storage resource state by adopting a stage constraint progressive scheduling method, and generating updated execution state evidence; Based on the updated execution state evidence, calculating the state deviation convergence degree in the stage to judge whether the current scheduling behavior has fed back the execution state or not, and acquiring the deviation regression discriminant in the stage; And based on the intra-stage deviation regression judgment quantity, marking the corresponding execution stage interval by cooling constraint, and carrying out continuous combination with the updated execution state evidence to generate the scheduled continuous execution state evidence.
  10. 10. A campus cloud platform elastic resource scheduling system based on the campus cloud platform elastic resource scheduling method of any one of claims 1-9 is characterized by comprising, The state evidence module is used for collecting the execution state information and the storage resource state and generating an execution state evidence through solidification after stabilization treatment; The stage identification module is used for carrying out cooperative evolution judgment on the execution state evidence, locating a stage switching boundary in the program execution process and dividing an execution stage interval; The state constraint module performs staged screening on the execution state evidence according to the execution stage interval to generate an execution state constraint condition; The trigger judgment module is used for carrying out attenuation update processing on the continuous out-of-range condition by adopting a continuous deviation judgment algorithm based on the execution state constraint condition to generate an execution scheduling trigger state; and the scheduling updating module is used for executing progressive resource scheduling control according to the execution stage interval according to the execution scheduling triggering state and updating the execution state evidence through the cooling constraint mark.

Description

Campus cloud platform elastic resource scheduling method and system Technical Field The invention relates to the technical field of resource management, in particular to a campus cloud platform elastic resource scheduling method and system. Background With the continuous development of cloud computing and virtualization technologies, a campus informatization infrastructure gradually evolves from a traditional distributed server architecture to a centralized and platform campus cloud platform. Most of the current universities rely on private cloud or hybrid cloud forms to bear a plurality of service loads of a teaching experiment system, scientific research calculation tasks, virtual simulation environments and teaching management application on a unified cloud platform. The existing industrial cloud platform resource scheduling method mostly uses the elastic scheduling thought in the general cloud computing environment, and mainly triggers resource expansion and contraction operation based on static or instantaneous indexes such as CPU utilization rate, memory utilization rate or storage capacity threshold value. At the storage resource scheduling level, the existing campus cloud platform resource scheduling scheme is generally focused on single-index monitoring of storage capacity utilization rate or throughput capacity. On the one hand, the storage resource scheduling decision lacks joint analysis on the program execution state, and running characteristics such as thread blocking, I/O access concurrency degree, storage access time delay and the like cannot be brought into a unified scheduling decision frame, so that real storage access pressure change of the program is difficult to reflect. On the other hand, the storage resource scheduling lacks of staged and continuous constraint, and is easy to excessively adjust the storage resource in the program execution process, so that the overall stability of teaching and scientific research task operation is reduced. Disclosure of Invention The present invention has been made in view of the above-described problems occurring in the prior art. Therefore, the invention provides a campus cloud platform elastic resource scheduling method which solves the problems that the state collaborative analysis is difficult to realize in storage resource scheduling and the constraint of lack of staged continuous scheduling is difficult to realize. In order to solve the technical problems, the invention provides the following technical scheme: The invention provides a campus cloud platform elastic resource scheduling method, which comprises the steps of collecting execution state information and storage resource states, solidifying after stabilizing to generate execution state evidences, judging the cooperative evolution of the execution state for the execution state evidences, locating a stage switching boundary in the program execution process, dividing the execution stage interval, carrying out stage screening on the execution state evidences according to the execution stage interval to generate execution state constraint conditions, carrying out attenuation update processing on continuous out-of-range conditions by adopting a continuous deviation judging algorithm based on the execution state constraint conditions to generate an execution scheduling trigger state, and executing progressive resource scheduling control according to the execution stage interval and updating the execution state evidences through cooling constraint marks according to the execution scheduling trigger state. As a preferable scheme of the campus cloud platform elastic resource scheduling method, the execution state information comprises thread scheduling state, blocking duration time, I/O access occupation proportion and request response delay information; the storage resource states include an available storage capacity state, a storage access concurrency state, a storage access latency state, and a storage access queue state. As a preferable scheme of the campus cloud platform elastic resource scheduling method, the method generates the execution state evidence, and comprises the following specific steps, Performing sliding window processing on the execution state information according to time sequence, and eliminating transient mutation data only appearing at a single sampling point; And locking and storing the data with the consistent change trend of a plurality of continuous sampling periods in the execution state information processed by the sliding window, and generating the execution state evidence. As a preferable scheme of the campus cloud platform elastic resource scheduling method, the invention comprises the steps of performing execution state co-evolution discrimination on the execution state evidence, positioning a phase switching boundary in the program execution process, specifically comprising the following steps, Performing time stamp binding and trend marking on the execution s