Method and system for realizing instance scheduling in cloud computing environment

By adopting the sharing method of multi-scheduling nodes and multi-scheduling processes in the cloud computing environment, the computing nodes are divided into sub-lists, which solves the problems of low deployment efficiency and poor stability in large-scale cloud computing environments, and realizes efficient resource utilization and the improvement of deployment efficiency.

CN118939366APending Publication Date: 2024-11-12HARBIN INST OF TECH +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410623841.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-05-20
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

In a large-scale cloud computing environment, the deployment efficiency of existing instance deployment methods is low, poor deployment stability and large resource utilization, making it difficult to meet the needs of rapidly deploying multiple instances.

Method used

By adopting multi-scheduling nodes and multi-scheduling processes in the cloud computing environment, the computing nodes are divided into sub-lists that do not overlap with each other. Each scheduling node is responsible for managing a part of the computing nodes, using multiple processes to share the scheduling pressure, and through the division of the scheduling nodes. The slice scheduling improves efficiency, and regularly detects the health of nodes for segmentation and resource redistribution.

Benefits of technology

It improves the efficiency and stability of instance deployment, reduces resource usage, avoids manual intervention, and ensures the efficient operation of scheduling services.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118939366A_ABST
    Figure CN118939366A_ABST
Patent Text Reader

Abstract

The invention discloses a method and a system for realizing instance scheduling in a cloud computing environment, and belongs to the technical field of cloud computing. The problem that an existing instance deployment method is low in deployment efficiency, poor in deployment stability and large in resource occupation in a large-scale cloud computing environment is solved. According to the method, scheduling is shared in a multi-scheduling-node and multi-scheduling-process mode, a computing node which is responsible for scheduling is segmented at a single scheduling node, and scheduling services are enabled to share different node scheduling in a multi-process mode; and meanwhile, a fragmentation scheduling mode is adopted for a plurality of scheduling nodes, and each node is responsible for scheduling a part of computing nodes. And after the scheduling request is received, scheduling is carried out by each scheduling process of the scheduling node in turn, and if the scheduling node meeting the requirement is not selected after the scheduling of the scheduling node is finished, the scheduling is carried out by other scheduling nodes. The method can be applied to scheduling of instances in a cloud computing environment.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of cloud computing, and in particular relates to a method and system for implementing instance scheduling in a cloud computing environment. Background Art

[0002] In the field of cloud computing, instance deployment is inevitable. The larger the cloud computing system, the more frequent the instance deployment will be, and the more instances will be deployed at a time. As the scale of the cloud computing system increases and the number of instances deployed at a time increases, the deployment time will become slower and slower. This not only affects the customer experience, but also makes the deployment process more unstable. In order to quickly deploy multiple instances in a large-scale cloud computing system, the following two methods are currently used:

[0003] Method 1: No scheduling is performed for each virtual instance, but resources are manually calculated and node deployment is specified. The advantage of this method is that it does not involve the scheduling part, which reduces the processing consumption of the scheduling process, improves efficiency, and directly places the virtual instance on the node. The disadvantage is that the use of resources needs to be considered manually, which occupies a large amount of resources; and improper consideration may cause the virtual instance to fail to generate on the computing node, and the deployment stability is poor.

[0004] Method 2: Improve the hardware performance of the scheduling node. Increasing the number of CPUs and RAM on a single node can also improve scheduling efficiency to a certain extent. However, this method of improving hardware is an improvement over large-scale cloud computing environments, but it is a drop in the bucket and has an upper limit.

[0005] Therefore, it is necessary to propose a new method to improve the deployment efficiency and stability of virtual machine instances in large-scale cloud computing environments and reduce resource usage. Summary of the invention

[0006] The purpose of the present invention is to solve the problems of low deployment efficiency, poor deployment stability and large resource occupation of existing instance deployment methods in a large-scale cloud computing environment, and to propose a method and system for realizing instance scheduling in a cloud computing environment.

[0007] The technical solution adopted by the present invention to solve the above technical problems is:

[0008] According to one aspect of the present invention, a method for implementing instance scheduling in a cloud computing environment comprises the following steps:

[0009] Step 1: The number of scheduling nodes is recorded as N. After each scheduling node obtains all computing nodes in the resource cache list, all computing nodes are divided into N sub-lists that do not overlap with each other;

[0010] Each scheduling node determines the sublist it is responsible for managing based on the position of its own identification among all scheduling nodes;

[0011] Step 2: For any scheduling node, according to the number of scheduling processes in the scheduling node, each scheduling process of the scheduling node divides the sublist managed by the scheduling node, and then determines the computing nodes that each scheduling process is responsible for scheduling according to the position of each scheduling process in the list;

[0012] Similarly, obtain the computing nodes that each scheduling process under each scheduling node is responsible for scheduling;

[0013] Step 3: After receiving an external scheduling request, each scheduling node collaborates to find a computing node that meets the scheduling requirements, and deploys instances on the found computing nodes according to the scheduling request.

[0014] Furthermore, the method of dividing all computing nodes into N mutually non-overlapping sublists is as follows:

[0015] Sort all computing nodes in descending order of remaining CPUs, and record the sorted computing nodes as the first, second, ... Mth computing nodes respectively;

[0016] Divide the i×N+1th computing node into the first sublist, i≤M / N-1;

[0017] Divide the i×N+2th computing node into the second sublist;

[0018] …

[0019] Divide the i×N+Nth computing node into the Nth sublist;

[0020] And after a fixed period of time, all computing nodes are re-split.

[0021] Furthermore, the method of dividing all computing nodes into N mutually non-overlapping sublists is as follows:

[0022] All computing nodes are sorted in descending order according to the remaining memory, and the sorted computing nodes are recorded as the first, second, ... Mth computing nodes respectively;

[0023] Divide the i×N+1th computing node into the first sublist, i≤M / N-1;

[0024] Divide the i×N+2th computing node into the second sublist;

[0025] …

[0026] Divide the i×N+Nth computing node into the Nth sublist;

[0027] And after a fixed period of time, all computing nodes are re-split.

[0028] Furthermore, the method further comprises:

[0029] Each scheduling node periodically detects the health of other scheduling nodes in the entire cloud computing environment. When an abnormal scheduling node is detected in the cloud computing environment, the normal scheduling node re-segments all computing nodes in the resource cache list and updates the sub-list it is responsible for managing.

[0030] Furthermore, each scheduling node periodically detects the health of other scheduling nodes in the entire cloud computing environment, and the detection method adopted is a communication handshake method.

[0031] Furthermore, the specific process of step three is:

[0032] Step 31: Initialize the number of scheduling nodes n=1;

[0033] Step 32: Initialize the number of scheduling processes of the nth scheduling node l=1;

[0034] Step 33: Determine whether there is a computing node that meets the scheduling requirements among the computing nodes scheduled by the lth scheduling process of the nth scheduling node;

[0035] If there is a computing node that meets the scheduling requirements, execute step 36;

[0036] If there is no computing node that meets the scheduling requirements, execute steps 3 and 4;

[0037] Step 34: determine whether the number of scheduling processes l is less than L, where L is the total number of scheduling processes of the nth scheduling node;

[0038] If l is less than L, set l=l+1 and return to step 33;

[0039] If l is equal to L, then execute steps 3 and 5;

[0040] Step 35: Set n=n+1, and return to step 32;

[0041] Step 36: Whether the number of computing nodes that meet the scheduling requirements is greater than 1;

[0042] If it is greater than 1, a computing node is selected from the computing nodes that meet the scheduling requirements, and the instance is deployed on the selected computing node according to the scheduling request;

[0043] If it is equal to 1, the instance is deployed directly on the computing node that meets the scheduling requirements according to the scheduling request.

[0044] Furthermore, the determination of whether there is a computing node that meets the scheduling requirement among the computing nodes that are scheduled by the lth scheduling process of the nth scheduling node is specifically as follows:

[0045] If the remaining CPU and remaining memory of the computing node both meet the scheduling requirements, the computing node is a computing node that meets the scheduling requirements.

[0046] Furthermore, the step of selecting a computing node from the obtained computing nodes that meet the scheduling requirements is specifically:

[0047] If the deployment strategy is centralized deployment, a computing node with the smallest remaining CPU is selected from the computing nodes that meet the scheduling requirements;

[0048] If the deployment strategy is distributed deployment, a computing node with the largest remaining CPU is selected from the computing nodes that meet the scheduling requirements.

[0049] According to another aspect of the present invention, a system for implementing instance scheduling in a cloud computing environment includes N scheduling nodes and a resource cache list;

[0050] The resource cache list is used to store all computing nodes;

[0051] The N scheduling nodes are used to divide the computing nodes in the resource cache list and determine the sub-lists that they are responsible for managing;

[0052] The N scheduling nodes are also used to divide the computing nodes in the sublist according to the number of scheduling processes in the scheduling node, and obtain the computing nodes that each scheduling process is responsible for scheduling;

[0053] Each scheduling process is used to collaboratively find computing nodes that meet the scheduling requirements.

[0054] The beneficial effects of the present invention are:

[0055] The present invention uses multiple scheduling nodes and multiple scheduling processes to share scheduling. In a single scheduling node, the computing nodes that it is responsible for scheduling are divided, so that the scheduling service can perform multi-process scheduling of different nodes; at the same time, a shard scheduling method is adopted for multiple scheduling nodes, and each node is responsible for the scheduling of a part of the computing nodes. This can balance the pressure, make full use of scheduling node resources, improve scheduling efficiency, and thus improve the efficiency of instance deployment.

[0056] After receiving the scheduling request, if a suitable host can be found in the node that it is responsible for scheduling, the scheduling is successful; if no suitable computing node is selected for landing, it will be handed over to other scheduling processes of this scheduling node for scheduling in turn. If the scheduling node still does not select a scheduling node that meets the requirements after the scheduling is completed, it will be handed over to other scheduling nodes for scheduling until the requirements are met. At the same time, the method of the present invention improves the deployment stability, does not require manual consideration of resource usage, and reduces resource occupation. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a schematic diagram of the packet switching logic;

[0058] Figure 2 It is a schematic diagram of the scheduling request processing process. DETAILED DESCRIPTION

[0059] Specific implementation method 1: This implementation method provides a method for implementing instance scheduling in a cloud computing environment, and the method specifically includes the following steps:

[0060] Step 1: The number of scheduling nodes is recorded as N (the number of scheduling nodes can be automatically configured according to the scale of computing nodes during installation and deployment). After each scheduling node obtains all computing nodes in the resource cache list, it divides all computing nodes into N non-overlapping sub-lists;

[0061] Each scheduling node determines the sublist it is responsible for managing based on the position of its own identifier among all scheduling nodes (the identifiers can be sorted according to certain rules or randomly assigned, and each scheduling node knows the total number of scheduling nodes and the identifiers of each scheduling node in the entire cloud computing environment);

[0062] Step 2: For any scheduling node, according to the number of scheduling processes in the scheduling node, each scheduling process of the scheduling node divides the sublist managed by the scheduling node (the division method can be the same as step 1), and then determines the computing node that each scheduling process is responsible for scheduling according to the position of each scheduling process in the list;

[0063] Similarly, obtain the computing nodes that each scheduling process under each scheduling node is responsible for scheduling;

[0064] Step 3: After receiving an external scheduling request, each scheduling node collaborates to find a computing node that meets the scheduling requirements, and deploys instances on the found computing nodes according to the scheduling request.

[0065] The purpose of grouping computing nodes in the present invention is to improve the parallelism of multiple scheduling processes and to share the time consumption of the pure computing part in the scheduling process among multiple scheduling instances, thereby achieving a speed-up effect.

[0066] Specific implementation method 2: This implementation method is different from the specific implementation method 1 in that all computing nodes are divided into N sub-lists that do not overlap with each other, specifically:

[0067] Sort all computing nodes in descending order of remaining CPUs, and record the sorted computing nodes as the first, second, ... Mth computing nodes respectively;

[0068] Divide the i×N+1th computing node into the first sublist, i≤M / N-1;

[0069] Divide the i×N+2th computing node into the second sublist;

[0070] …

[0071] Divide the i×N+Nth computing node into the Nth sublist;

[0072] And after a fixed period of time, all computing nodes are re-split.

[0073] By re-splitting, a new sublist can be obtained, so that the nodes scheduled by each scheduling process of each scheduling node can be obtained again. Because the remaining CPUs on each computing node may change greatly as the scheduling process proceeds, re-splitting is required to ensure the scheduling effect.

[0074] The other steps and parameters are the same as those in the first embodiment.

[0075] The present invention divides each sublist into computing nodes with larger remaining CPU and computing nodes with smaller remaining CPU, so as to ensure that there are computing nodes that meet the scheduling requirements in each sublist as much as possible, thereby improving the efficiency of scheduling. Moreover, by dividing the sublist, the resource conflict when scheduling multiple instances and multiple processes to run concurrently can be reduced as much as possible, thereby speeding up the scheduling action.

[0076] Specific implementation method three: This implementation method is different from specific implementation methods one or two in that all computing nodes are divided into N mutually non-overlapping sub-lists, specifically:

[0077] All computing nodes are sorted in descending order according to the remaining memory, and the sorted computing nodes are recorded as the first, second, ... Mth computing nodes respectively;

[0078] Divide the i×N+1th computing node into the first sublist, i≤M / N-1;

[0079] Divide the i×N+2th computing node into the second sublist;

[0080] …

[0081] Divide the i×N+Nth computing node into the Nth sublist;

[0082] And after a fixed period of time, all computing nodes are re-split.

[0083] By re-splitting, a new sublist can be obtained, so that the nodes scheduled by each scheduling process of each scheduling node can be obtained again. Because the remaining memory on each computing node may change greatly as the scheduling process proceeds, re-splitting is required to ensure the scheduling effect.

[0084] The other steps and parameters are the same as those in the first or second embodiment.

[0085] The present invention divides each sublist into computing nodes with larger remaining memory and computing nodes with smaller remaining memory, so that each sublist can be guaranteed to have computing nodes that meet the scheduling requirements as much as possible, thereby improving the efficiency of scheduling. Moreover, by dividing the sublist, the resource conflict when scheduling multiple instances and multiple processes to run concurrently can be reduced as much as possible, thereby speeding up the scheduling action.

[0086] Moreover, in addition to the quantitative division method according to the second and third specific implementation methods, the division can also be performed according to the computing node type or business. It can also be divided according to the total resources that the computing nodes can provide, so that each sublist includes both computing nodes that provide large total resources and computing nodes that provide small total resources, and when dividing according to the total resources that the computing nodes can provide, there is no need to re-divide at fixed intervals.

[0087] Specific implementation method 4: This implementation method is different from any one of the specific implementation methods 1 to 3 in that the method further includes:

[0088] Each scheduling node periodically checks the health of other scheduling nodes in the entire cloud computing environment. When an abnormal scheduling node is detected in the cloud computing environment, the normal scheduling node re-divides all computing nodes in the resource cache list and updates the sub-list it is responsible for managing (at this time, the abnormal scheduling node will no longer allocate computing nodes).

[0089] The other steps and parameters are the same as those in Specific Embodiments 1 to 3.

[0090] Specific implementation method five: This implementation method is different from any one of specific implementation methods one to four in that each scheduling node periodically detects the health of other scheduling nodes in the entire cloud computing environment, and the detection method adopted is a communication handshake method.

[0091] The other steps and parameters are the same as those in Specific Embodiments 1 to 4.

[0092] Specific implementation method 6: This implementation method is different from the specific implementation methods 1 to 5 in that the specific process of step 3 is as follows:

[0093] Step 31: Initialize the number of scheduling nodes n=1;

[0094] Step 32: Initialize the number of scheduling processes of the nth scheduling node l=1;

[0095] Step 33: Determine whether there is a computing node that meets the scheduling requirements among the computing nodes scheduled by the lth scheduling process of the nth scheduling node;

[0096] If there is a computing node that meets the scheduling requirements, execute step 36;

[0097] If there is no computing node that meets the scheduling requirements, execute steps 3 and 4;

[0098] Step 34: determine whether the number of scheduling processes l is less than L, where L is the total number of scheduling processes of the nth scheduling node;

[0099] If l is less than L, set l=l+1 and return to step 33;

[0100] If l is equal to L, then execute steps 3 and 5;

[0101] Step 35: Set n=n+1, and return to step 32;

[0102] Step 36: Whether the number of computing nodes that meet the scheduling requirements is greater than 1;

[0103] If it is greater than 1, a computing node is selected from the computing nodes that meet the scheduling requirements, and the instance is deployed on the selected computing node according to the scheduling request;

[0104] If it is equal to 1, the instance is deployed directly on the computing node that meets the scheduling requirements according to the scheduling request.

[0105] The other steps and parameters are the same as those in Specific Implementation Methods 1 to 5.

[0106] like Figure 1 and Figure 2As shown, after receiving an external scheduling request, the second scheduling node first responds to the request (which scheduling node processes the scheduling request first can be randomly determined, and generally the first scheduling node can be selected to start first). Inside the second scheduling node, starting from any scheduling process, it is determined whether there is a computing node that meets the scheduling requirements in the scheduling process, until a computing node that meets the scheduling requirements is found, and the traversal of the scheduling process is stopped. If a computing node that meets the scheduling requirements is still not found after traversing all scheduling processes in the second scheduling node, the scheduling request will continue to be processed by the next scheduling node, and a computing node that meets the scheduling request will eventually be found. In some cases, a scheduling process may include several computing nodes that meet the scheduling requirements, so the number of computing nodes that meet the scheduling requirements found at one time may not be unique, so a computing node can be selected from the computing nodes that meet the scheduling requirements according to the requirements.

[0107] Specific implementation method 7: This implementation method is different from any one of specific implementation methods 1 to 6 in that the determination of whether there is a computing node that meets the scheduling requirements among the computing nodes that are scheduled by the lth scheduling process of the nth scheduling node is specifically as follows:

[0108] If the remaining CPU and remaining memory of the computing node both meet the scheduling requirements, the computing node is a computing node that meets the scheduling requirements.

[0109] The other steps and parameters are the same as those in Specific Embodiments 1 to 6.

[0110] Specific implementation eight: This implementation differs from any one of specific implementations one to seven in that the step of selecting a computing node from the obtained computing nodes that meet the scheduling requirements is as follows:

[0111] If the deployment strategy is centralized deployment, a computing node with the smallest remaining CPU is selected from the computing nodes that meet the scheduling requirements;

[0112] If the deployment strategy is distributed deployment, a computing node with the largest remaining CPU is selected from the computing nodes that meet the scheduling requirements.

[0113] The other steps and parameters are the same as those in Specific Embodiments 1 to 7.

[0114] Specific implementation method 9: A system for implementing instance scheduling in a cloud computing environment described in this implementation method includes N scheduling nodes and a resource cache list;

[0115] The resource cache list is used to store all computing nodes;

[0116] The N scheduling nodes are used to divide the computing nodes in the resource cache list and determine the sub-lists that they are responsible for managing;

[0117] The N scheduling nodes are also used to divide the computing nodes in the sublist according to the number of scheduling processes in the scheduling node, and obtain the computing nodes that each scheduling process is responsible for scheduling;

[0118] Each scheduling process is used to collaboratively find computing nodes that meet the scheduling requirements.

[0119] Specific implementation method ten: This implementation method is different from specific implementation method nine in that the N scheduling nodes are used to divide the computing nodes in the resource cache list, and the switching method adopts method (1) or method (2):

[0120] Method (1): Sort all computing nodes in descending order of remaining CPUs, and record the sorted computing nodes as the first, second, ..., Mth computing nodes respectively;

[0121] Divide the i×N+1th computing node into the first sublist;

[0122] Divide the i×N+2th computing node into the second sublist;

[0123] …

[0124] Divide the i×N+Nth computing node into the Nth sublist;

[0125] And after a fixed period of time, all computing nodes are re-split.

[0126] Method (2): Sort all computing nodes in descending order of remaining memory, and record the sorted computing nodes as the first, second, ... Mth computing nodes respectively;

[0127] Divide the i×N+1th computing node into the first sublist;

[0128] Divide the i×N+2th computing node into the second sublist;

[0129] …

[0130] Divide the i×N+Nth computing node into the Nth sublist;

[0131] And after a fixed period of time, all computing nodes are re-split.

[0132] The other steps and parameters are the same as those in the ninth embodiment.

[0133] The above calculation examples of the present invention are only used to explain the calculation model and calculation process of the present invention in detail, and are not intended to limit the implementation methods of the present invention. For ordinary technicians in the relevant field, other different forms of changes or modifications can be made based on the above description. It is impossible to list all the implementation methods here. All obvious changes or modifications derived from the technical solution of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for implementing instance scheduling in a cloud computing environment, characterized in that: The method specifically comprises the following steps: Step 1: The number of scheduling nodes is recorded as N. After each scheduling node obtains all computing nodes in the resource cache list, all computing nodes are divided into N sub-lists that do not overlap with each other; Each scheduling node determines the sublist it is responsible for managing based on the position of its own identification among all scheduling nodes; Step 2: For any scheduling node, according to the number of scheduling processes in the scheduling node, each scheduling process of the scheduling node divides the sublist managed by the scheduling node, and then determines the computing nodes that each scheduling process is responsible for scheduling according to the position of each scheduling process in the list; Similarly, obtain the computing nodes that each scheduling process under each scheduling node is responsible for scheduling; Step 3: After receiving an external scheduling request, each scheduling node collaborates to find a computing node that meets the scheduling requirements, and deploys instances on the found computing nodes according to the scheduling request.

2. A method for implementing instance scheduling in a cloud computing environment according to claim 1, characterized in that: The method of dividing all computing nodes into N mutually non-overlapping sublists is as follows: Sort all computing nodes in descending order of remaining CPUs, and record the sorted computing nodes as the first, second, ... Mth computing nodes respectively; Divide the i×N+1th computing node into the first sublist, i≤M / N-1; Divide the i×N+2th computing node into the second sublist; …… Divide the i×N+Nth computing node into the Nth sublist; And after a fixed period of time, all computing nodes are re-split.

3. The method for implementing instance scheduling in a cloud computing environment according to claim 1, characterized in that: The method of dividing all computing nodes into N mutually non-overlapping sublists is as follows: All computing nodes are sorted in descending order according to the remaining memory, and the sorted computing nodes are recorded as the first, second, ... Mth computing nodes respectively; Divide the i×N+1th computing node into the first sublist, i≤M / N-1; Divide the i×N+2th computing node into the second sublist; …… Divide the i×N+Nth computing node into the Nth sublist; And after a fixed period of time, all computing nodes are re-split.

4. The method for implementing instance scheduling in a cloud computing environment according to claim 1, characterized in that: The method further comprises: Each scheduling node periodically detects the health of other scheduling nodes in the entire cloud computing environment. When an abnormal scheduling node is detected in the cloud computing environment, the normal scheduling node re-segments all computing nodes in the resource cache list and updates the sub-list it is responsible for managing.

5. A method for implementing instance scheduling in a cloud computing environment according to claim 4, characterized in that: Each scheduling node periodically detects the health of other scheduling nodes in the entire cloud computing environment, and the detection method adopted is a communication handshake method.

6. The method for implementing instance scheduling in a cloud computing environment according to claim 1, characterized in that: The specific process of step three is: Step 31: Initialize the number of scheduling nodes n=1; Step 32: Initialize the number of scheduling processes of the nth scheduling node l=1; Step 33: Determine whether there is a computing node that meets the scheduling requirements among the computing nodes scheduled by the lth scheduling process of the nth scheduling node; If there is a computing node that meets the scheduling requirements, execute step 36; If there is no computing node that meets the scheduling requirements, execute steps 3 and 4; Step 34: determine whether the number of scheduling processes l is less than L, where L is the total number of scheduling processes of the nth scheduling node; If l is less than L, set l=l+1 and return to step 33; If l is equal to L, then execute steps 3 and 5; Step 35: Set n=n+1, and return to step 32; Step 36: Whether the number of computing nodes that meet the scheduling requirements is greater than 1; If it is greater than 1, a computing node is selected from the computing nodes that meet the scheduling requirements, and the instance is deployed on the selected computing node according to the scheduling request; If it is equal to 1, the instance is deployed directly on the computing node that meets the scheduling requirements according to the scheduling request.

7. A method for implementing instance scheduling in a cloud computing environment according to claim 6, characterized in that: The step of judging whether there is a computing node that meets the scheduling requirement among the computing nodes that are scheduled by the lth scheduling process of the nth scheduling node is specifically: If the remaining CPU and remaining memory of the computing node both meet the scheduling requirements, the computing node is a computing node that meets the scheduling requirements.

8. A method for implementing instance scheduling in a cloud computing environment according to claim 6 or 7, characterized in that: The step of selecting a computing node from the obtained computing nodes that meet the scheduling requirements is specifically: If the deployment strategy is centralized deployment, a computing node with the smallest remaining CPU is selected from the computing nodes that meet the scheduling requirements; If the deployment strategy is distributed deployment, a computing node with the largest remaining CPU is selected from the computing nodes that meet the scheduling requirements.

9. A system for implementing instance scheduling in a cloud computing environment, characterized in that: The system includes N scheduling nodes and a resource cache list; The resource cache list is used to store all computing nodes; The N scheduling nodes are used to divide the computing nodes in the resource cache list and determine the sub-lists that they are responsible for managing; The N scheduling nodes are also used to divide the computing nodes in the sublist according to the number of scheduling processes in the scheduling node, and obtain the computing nodes that each scheduling process is responsible for scheduling; Each scheduling process is used to collaboratively find computing nodes that meet the scheduling requirements.

10. A system for implementing instance scheduling in a cloud computing environment according to claim 9, characterized in that: The N scheduling nodes are used to divide the computing nodes in the resource cache list, and the switching method adopts method (1) or method (2): Method (1): Sort all computing nodes in descending order of remaining CPUs, and record the sorted computing nodes as the first, second, ..., Mth computing nodes respectively; Divide the i×N+1th computing node into the first sublist; Divide the i×N+2th computing node into the second sublist; …… Divide the i×N+Nth computing node into the Nth sublist; And after a fixed period of time, all computing nodes are re-split; Method (2): Sort all computing nodes in descending order of remaining memory, and record the sorted computing nodes as the first, second, ... Mth computing nodes respectively; Divide the i×N+1th computing node into the first sublist; Divide the i×N+2th computing node into the second sublist; …… Divide the i×N+Nth computing node into the Nth sublist; And after a fixed period of time, all computing nodes are re-split.