Multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method and system

Abstract

The invention discloses an intelligent scheduling method and system for concurrent requests of a multi-tenant cloud computer operation management platform, belongs to the technical field of AI intelligent application, and aims at solving the technical problem of how to overcome the defects of resource preemption and low service quality of the multi-tenant cloud computer platform in a high-concurrency scene. According to the technical scheme, the resource utilization efficiency and the service quality guarantee capability of a multi-tenant cloud computer platform are improved, and the method comprises the steps of tenant behavior intelligent analysis: collecting related original data of each tenant in real time, constructing a tenant portrait model, and achieving multi-dimensional priority dynamic evaluation; resource demand time sequence prediction: predicting resource demands of a CPU, a GPU and a memory in a set period through a Prophet prediction model in combination with historical data and a real-time load, establishing a resource demand early warning mechanism, and performing resource pre-allocation in advance; an elastic resource allocation strategy is established, resource sufficiency and shortage scenes are distinguished, dynamic resource quota adjustment is achieved, and response time is shortened; and abnormal behaviors are monitored in real time and rapidly isolated.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of AI intelligent application, in particular to a multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method and system. BACKGROUND

[0002] A multi-tenant cloud computer system realizes resource isolation through virtualization technology (such as virtual machines, containers) or containerization technology (such as Kubernetes namespaces), and its core architecture is divided into three layers: infrastructure layer, platform management layer and application service layer.

[0003] Among them, the infrastructure layer functions as follows:

[0004] ①Computing resources: CPU, memory, GPU and other hardware resources are divided into independent virtual resource pools through virtualization technology, and each tenant is allocated a dedicated quota.

[0005] ②Storage resources: A distributed storage system (such as Ceph, HDFS) is used to realize tenant data isolation through data encryption and access control.

[0006] ③Network resources: Virtual local area networks (VLANs), software-defined networks (SDNs) or zero-trust network architectures are used to ensure that network traffic between tenants does not interfere with each other.

[0007] The platform management layer functions as follows:

[0008] ①Resource scheduling: Resources are dynamically allocated based on tenant priority, resource demand and load balancing algorithms (such as round robin, minimum number of connections).

[0009] ②Permission management: Through RBAC (role-based access control) or ABAC (attribute-based access control) models, tenant user permissions (such as administrators, ordinary users, read-only users) are refined.

[0010] ③Monitoring and auditing: Real-time collection of CPU usage, memory occupancy, network bandwidth and other indicators, and recording of user operation logs to support compliance review.

[0011] The application service layer functions as follows:

[0012] ①Shared application instances: The same application service (such as office software, development environment) serves multiple tenants through a multi-tenant architecture, reducing the cost of repeated deployment.

[0013] ②Personalized configuration: Tenants can customize interface themes, functional modules and data storage paths to meet differentiated needs.

[0014] The main technical problems faced by current multi-tenant cloud computer systems include:

[0015] ① Static resource allocation strategy cannot adapt to the dynamic changes of tenant behavior;

[0016] ② Lack of effective tenant priority evaluation mechanism, important business requests may be delayed;

[0017] ③ Abnormal resource occupation behavior leads to decline of system stability;

[0018] ④ Resource scheduling response is not timely under burst traffic scenario.

[0019] Therefore, how to overcome the defects of resource preemption and low service quality of the multi-tenant cloud computer platform under high concurrency scenario, and improve the resource utilization efficiency and service quality guarantee capability of the multi-tenant cloud computer platform are the technical problems to be solved at present. SUMMARY

[0020] The technical task of the present application is to provide a multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method and system to solve the problem of how to overcome the defects of resource preemption and low service quality of the multi-tenant cloud computer platform under high concurrency scenario, and improve the resource utilization efficiency and service quality guarantee capability of the multi-tenant cloud computer platform.

[0021] The technical task of the present application is realized in the following manner, a multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method, the method is as follows:

[0022] Intelligent analysis of tenant behavior: real-time collection of relevant raw data of each tenant and construction of tenant portrait model, realization of multi-dimensional priority dynamic evaluation;

[0023] Resource demand time series prediction: combining historical data and real-time load, predicting the resource demand of CPU, GPU and memory in a set period through a Prophet prediction model, and establishing a resource demand early warning mechanism to pre-allocate resources;

[0024] Establishing an elastic resource allocation strategy, distinguishing between resource sufficient and resource tight scenarios, realizing dynamic resource quota adjustment, and speeding up response time;

[0025] Real-time monitoring and rapid isolation of abnormal behavior: identifying DDOS attack behavior through ntopng open source tool, and automatically triggering resource isolation strategy to speed up isolation response time, while supporting manual intervention and rule customization to realize resource quota adjustment.

[0026] As a preferred, the intelligent analysis of tenant behavior is as follows:

[0027] Data collection: real-time collection of tenant request frequency, resource occupation time, and tenant-related raw data of task type;

[0028] Feature extraction: extract tenant feature data such as request frequency, resource occupation and task type from tenant-related raw data;

[0029] Model training: use hierarchical clustering algorithm to build tenant portrait model, update tenant priority score regularly, and train tenant portrait model with extracted feature data to obtain trained tenant portrait model;

[0030] Dynamic update: dynamically update tenant behavior feature library according to tenant feature data, and support real-time priority adjustment; wherein, the information of tenant behavior feature library includes request frequency, resource occupation and task type.

[0031] As preferred, the tenant type includes VIP tenant, ordinary tenant and potential abnormal tenant;

[0032] Among them, VIP tenant refers to a tenant with high request frequency, stable resource usage and high value task;

[0033] Ordinary tenant refers to a tenant with medium or fluctuating resource usage and regular tasks;

[0034] Potential abnormal tenant refers to a tenant with abnormally high or low request frequency and abnormal resource occupation mode.

[0035] As preferred, the training based on Prophet prediction model is as follows:

[0036] By inputting the usage information of CPU, GPU and memory within 7 days into the Prophet prediction model, the Prophet prediction model outputs the prediction of CPU, GPU and memory usage within 5-30 minutes in the future.

[0037] As preferred, the resource demand warning mechanism is as follows:

[0038] When the prediction result of the Prophet prediction model is greater than 60% in any of CPU, GPU and memory usage, send warning information;

[0039] When the prediction result of the Prophet prediction model is greater than 80% in any of CPU, GPU and memory usage, pre-allocate more corresponding resources for the tenant.

[0040] As preferred, the elastic resource allocation strategy is as follows:

[0041] When the resources are sufficient, use weighted fair queue algorithm to optimize task queuing;

[0042] When the resources are tight, start a three-level priority protection mechanism, which is as follows:

[0043] Level 1: guarantee core business tenant resource demand;

[0044] Level 2: reserve basic resource quota for ordinary tenants;

[0045] Level 3: limit resource requests of suspicious tenants;

[0046] Wherein, the judgment standard of whether the resource is sufficient is: when any one of CPU, GPU and memory condition is greater than 80% and lasts for more than 5 minutes, it is determined that the resource is tight; otherwise, it is determined that the resource is sufficient.

[0047] More preferably, the resource isolation strategy is as follows:

[0048] Level 1: flow limiting, that is, limiting 50% traffic;

[0049] Level 2: temporary ban, that is, banning for 5 minutes;

[0050] Level 3: complete isolation.

[0051] A multi-tenant cloud computer operation management platform concurrent request intelligent scheduling system, the system is used to realize the multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method as described above; the system comprises:

[0052] A tenant behavior intelligent analysis module is used to collect relevant raw data of each tenant in real time and build a tenant portrait model, and realize multi-dimensional priority dynamic evaluation;

[0053] A resource demand time sequence prediction module is used to combine historical data and real-time load, predict the resource demand of CPU, GPU and memory in a set period through a Prophet prediction model, and establish a resource demand early warning mechanism to pre-allocate resources;

[0054] A strategy establishment module is used to establish an elastic resource allocation strategy, distinguish between resource sufficient and tight scenarios, realize dynamic resource quota adjustment, and speed up response time;

[0055] An abnormal behavior real-time monitoring and rapid isolation module is used to identify DDOS attack behavior through ntopng open source tool, and automatically trigger resource isolation strategy to speed up isolation response time, while supporting manual intervention and rule customization to realize resource quota adjustment.

[0056] An electronic device comprising a memory and at least one processor;

[0057] Wherein, the memory has stored thereon a computer program;

[0058] The at least one processor executes the computer program stored in the memory, so that the at least one processor executes the multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method as described above.

[0059] A computer readable storage medium, the computer readable storage medium stores a computer program, the computer program can be executed by a processor to implement the multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method as described above.

[0060] The multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method and system of the application has the following advantages:

[0061] (I) The application realizes multi-dimensional tenant priority dynamic evaluation by constructing a tenant portrait model and a Prophet prediction model, and uses a time series prediction algorithm to predict resource demand fluctuations, establishes an elastic resource allocation strategy, optimizes the task queuing mechanism when resources are sufficient, intelligently enables priority protection schemes when resources are scarce, and identifies malicious resource occupation in real time according to abnormal behavior detection results, ensuring the stable operation of the system in a high concurrency scenario, and significantly improving the resource utilization efficiency and service quality guarantee capability of the multi-tenant cloud computer platform;

[0062] (II) The application realizes tenant behavior analysis, resource demand prediction and elastic scheduling through AI technology, solves the problems of resource preemption and service quality guarantee in a high concurrency scenario, and improves the resource utilization efficiency and service quality guarantee capability of the multi-tenant cloud computer platform;

[0063] (III) The application improves resource utilization, reduces resource waste through intelligent prediction, shortens high-priority tenant task response time, and improves abnormal behavior identification accuracy, while improving system availability in sudden high concurrency scenarios and guaranteeing basic resources for ordinary tenants;

[0064] (IV) The elastic resource allocation strategy of the application includes a three-level priority protection mechanism to improve response time;

[0065] (V) The abnormal behavior monitoring of the application speeds up the isolation response time and supports rule customization. BRIEF DESCRIPTION OF DRAWINGS

[0066] The application will be further described below with reference to the accompanying drawings.

[0067] The accompanying drawings illustrate the application. Figure 1 is a flowchart of the multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method. DETAILED DESCRIPTION

[0068] The multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method and system of the application will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0069] Embodiment 1:

[0070] As shown in the accompanying drawings Figure 1 The embodiment provides a multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method, which specifically comprises the following steps:

[0071] S1, intelligent analysis of tenant behavior: real-time collection of relevant raw data of each tenant and construction of a tenant portrait model to realize multi-dimensional priority dynamic evaluation;

[0072] S2, resource demand time series prediction: combining historical data and real-time load, predicting the resource demand of CPU, GPU and memory within a set period through a Prophet prediction model, and establishing a resource demand early warning mechanism to pre-allocate resources;

[0073] S3, establish an elastic resource allocation strategy to distinguish between resource sufficient and resource tight scenarios, realize dynamic resource quota adjustment, and speed up response time;

[0074] S4, real-time monitoring and rapid isolation of abnormal behavior: identify DDOS attack behavior through ntopng open source tools and automatically trigger resource isolation strategies to speed up isolation response time while supporting manual intervention and rule customization to achieve resource quota adjustment.

[0075] The tenant behavior intelligent analysis in step S1 of the embodiment specifically comprises the following steps:

[0076] S101, data collection: real-time collection of tenant-related raw data such as request frequency, resource occupation duration, and task type;

[0077] S102, feature extraction: extracting tenant feature data such as request frequency, resource occupation, and task type from tenant-related raw data;

[0078] S103, model training: using hierarchical clustering algorithm to construct tenant portrait model, regularly updating tenant priority score once, and training tenant portrait model through extracted feature data to obtain trained tenant portrait model;

[0079] S104, dynamic update: dynamically updating tenant behavior feature library according to tenant feature data and supporting real-time priority adjustment; wherein the information of tenant behavior feature library includes request frequency, resource occupation, and task type.

[0080] The tenant types in the embodiment include VIP tenants, ordinary tenants, and potential abnormal tenants;

[0081] Among them, VIP tenants refer to tenants with high request frequency, stable resource usage, and high-value tasks;

[0082] Normal tenants refer to tenants with medium or fluctuating resource usage, and regular tasks;

[0083] Potential abnormal tenants refer to tenants with abnormally high or low request frequency and abnormal resource occupation mode.

[0084] The training based on the Prophet prediction model in step S2 of the embodiment is as follows:

[0085] By inputting the usage information of CPU, GPU and memory within 7 days into the Prophet prediction model, the Prophet prediction model outputs the prediction of CPU, GPU and memory usage within 5-30 minutes in the future.

[0086] The resource demand early warning mechanism in step S2 of the embodiment is as follows:

[0087] When any one of the prediction results of CPU, GPU and memory usage based on the Prophet prediction model is greater than 60%, send an early warning message;

[0088] When any one of the prediction results of CPU, GPU and memory usage based on the Prophet prediction model is greater than 80%, pre-allocate more corresponding resources for the tenant.

[0089] The elastic resource allocation strategy in step S3 of the embodiment is as follows:

[0090] When the resources are sufficient, use the weighted fair queue algorithm to optimize task queuing;

[0091] When the resources are tight, start a three-level priority protection mechanism, which is as follows:

[0092] First level: protect the resource demand of core business tenants;

[0093] Second level: reserve basic resource quota for normal tenants;

[0094] Third level: limit the resource request of suspicious tenants;

[0095] Among them, the judgment standard of whether the resources are sufficient is: when any one of CPU, GPU and memory is greater than 80% and lasts for more than 5 minutes, it is determined that the resources are tight; otherwise, it is determined that the resources are sufficient.

[0096] The resource isolation strategy in step S4 of the embodiment is as follows:

[0097] First level: flow limiting, i.e. limiting 50% flow;

[0098] Second level: temporary ban, i.e. banning for 5 minutes;

[0099] Level 3: complete isolation.

[0100] Embodiment 2:

[0101] The embodiment provides a multi-tenant cloud computer operation management platform concurrent request intelligent scheduling system, which is used for implementing the multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method in the embodiment 1; the system comprises:

[0102] A tenant behavior intelligent analysis module is configured to collect relevant raw data of each tenant in real time and build a tenant portrait model, and realize multi-dimensional priority dynamic evaluation.

[0103] A resource demand time sequence prediction module is configured to combine historical data and real-time load, predict resource demands of CPU, GPU and memory in a set period based on a Prophet prediction model, and establish a resource demand early warning mechanism to perform resource pre-allocation in advance.

[0104] A strategy establishment module is configured to establish an elastic resource allocation strategy, distinguish resource sufficient and resource shortage scenarios, realize dynamic resource quota adjustment, and speed up response time.

[0105] An abnormal behavior real-time monitoring and rapid isolation module is configured to identify DDOS attack behavior through an ntopng open source tool, automatically trigger a resource isolation strategy, speed up isolation response time, and support manual intervention and rule customization to realize resource quota adjustment.

[0106] Embodiment 3:

[0107] The embodiment of the application further provides an electronic device, comprising a memory and a processor.

[0108] The memory stores computer execution instructions.

[0109] The processor executes the computer execution instructions stored in the memory, so that the processor executes the multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method in any embodiment of the application.

[0110] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), ready-to-program gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The processor can be a microprocessor, or the processor can be any conventional processor.

[0111] The memory can be used to store computer programs and / or modules, and the processor can realize various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory, and calling data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required by a function, and the like; and the data storage area can store data created according to the use of the terminal, and the like. In addition, the memory can also include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a secure digital (SD) card, a flash memory card, at least one disk storage period, a flash memory device, or other volatile solid-state memory device.

[0112] Embodiment 4:

[0113] The embodiment also provides a computer readable storage medium, in which a plurality of instructions are stored, the instructions are loaded by a processor to make the processor execute the multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method in any embodiment of the application. Specifically, a system or device provided with a storage medium can be provided, and software program code for realizing the functions of any one of the above embodiments is stored on the storage medium, and the computer (or CPU or MPU) of the system or device reads and executes the program code stored in the storage medium.

[0114] In this case, the program code read from the storage medium can realize the functions of any one of the above embodiments, and therefore the program code and the storage medium storing the program code constitute a part of the application.

[0115] The storage medium for providing the program code includes a floppy disk, a hard disk, a magneto-optical disk, an optical disk (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RYM, DVD-RW, DVD+RW), a magnetic tape, a non-volatile memory card, and a ROM. Alternatively, the program code can be downloaded from a server computer through a communication network.

[0116] In addition, it should be clear that not only the program code read by the computer can be executed, but also part or all of the actual operations can be completed by the operating system and the like operating on the computer based on the instructions of the program code, so as to realize the functions of any one of the above embodiments.

[0117] Further, it is understood that the program code read by the storage medium can be written into the memory provided in the expansion board inserted into the computer or the memory provided in the expansion unit connected to the computer, and then the CPU or the like mounted on the expansion board or the expansion unit is caused to perform part or all of the actual operation based on the instruction of the program code, thereby realizing the function of any of the above-described embodiments.

[0118] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not limited thereto; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can still be modified, or part or all of the technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for intelligent scheduling of concurrent requests of a multi-tenant cloud computer operation management platform, characterized in that, The method is specifically as follows: Tenant behavior intelligent analysis: real-time collection of relevant raw data of each tenant and construction of tenant portrait model to realize multi-dimensional priority dynamic evaluation; Resource demand time series prediction: combining historical data and real-time load, predicting CPU, GPU and memory resource demand in a set period based on the Prophet prediction model, and establishing a resource demand warning mechanism to pre-allocate resources; Establishing an elastic resource allocation strategy to distinguish between resource sufficient and tight scenarios, realizing dynamic resource quota adjustment and speeding up response time; Real-time monitoring and rapid isolation of abnormal behavior: identify DDOS attack behavior through ntopng open source tool and automatically trigger resource isolation strategy to speed up isolation response time while supporting manual intervention and rule customization to realize resource quota adjustment. 2.The multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method according to claim 1, characterized in that, Tenant behavior intelligent analysis is specifically as follows: Data collection: real-time collection of tenant request frequency, resource occupation time, and task type-related raw data; Feature extraction: extracting tenant feature data such as request frequency, resource occupation, and task type from tenant-related raw data; Model training: using hierarchical clustering algorithm to build tenant portrait model, regularly updating tenant priority score, and training tenant portrait model through extracted feature data to obtain trained tenant portrait model; Dynamic update: dynamically updating tenant behavior feature library according to tenant feature data and supporting real-time priority adjustment; the information in the tenant behavior feature library includes request frequency, resource occupation, and task type. 3.The multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method of claim 1, wherein, Tenant types include VIP tenants, ordinary tenants, and potential abnormal tenants; Among them, VIP tenants refer to tenants with high request frequency, stable resource usage, and high-value tasks; Ordinary tenants refer to tenants with medium or fluctuating resource usage and routine tasks; Potential abnormal tenants refer to tenants with abnormally high or low request frequency and abnormal resource occupation patterns.

4. The method of claim 1, wherein the method further comprises: The training of the Prophet prediction model is specifically as follows: By inputting 7-day CPU, GPU, and memory usage information into the Prophet prediction model, the Prophet prediction model outputs the prediction of CPU, GPU, and memory usage in the next 5-30 minutes.

5. The method of claim 1, wherein the method further comprises: The resource demand warning mechanism is specifically as follows: When any one of the Prophet prediction model's prediction results CPU, GPU, and memory usage is greater than 60%, send a warning message; When any one of the Prophet prediction model's prediction results CPU, GPU, and memory usage is greater than 80%, pre-allocate corresponding resources for the tenant.

6. The method of claim 1, wherein the method further comprises: The elastic resource allocation strategy is specifically as follows: When resources are sufficient, use the weighted fair queue algorithm to optimize task queuing; When resources are tight, start a three-level priority protection mechanism, which is specifically as follows: First level: protect the resource demand of core business tenants; Second level: reserve basic resource quota for ordinary tenants; Third level: limit the resource requests of suspicious tenants; The judgment standard of whether the resource is sufficient is that: when any one of CPU, GPU and memory condition is greater than 80% and lasts for more than 5 minutes, it is determined that the resource is in shortage; otherwise, it is determined that the resource is sufficient.

7. The multi-tenant cloud computer operations management platform concurrent request intelligent scheduling method according to any one of claims 1 to 6, characterized in that, The resource isolation strategy is as follows: Level one: flow limiting, that is, limiting 50% flow; Level two: temporary ban, that is, banning for 5 minutes; Level three: complete isolation.

8. A multi-tenant cloud computer operation management platform concurrent request intelligent scheduling system, characterized in that, The system is used to realize the multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method according to any one of claims 1 to 7; the system comprises: A tenant behavior intelligent analysis module is used to collect the relevant original data of each tenant in real time and construct a tenant portrait model, realize multi-dimensional priority dynamic evaluation; A resource demand time series prediction module is used to combine historical data and real-time load, predict the resource demand of CPU, GPU and memory in a set period through a Prophet prediction model, and establish a resource demand early warning mechanism to pre-allocate resources; A strategy establishment module is used to establish an elastic resource allocation strategy, distinguish between resource sufficient and resource shortage scenarios, realize dynamic resource quota adjustment, and speed up response time; An abnormal behavior real-time monitoring and rapid isolation module is used to identify DDOS attack behavior through ntopng open source tool, and automatically trigger resource isolation strategy to speed up isolation response time, while supporting manual intervention and rule customization to realize resource quota adjustment.

9. An electronic device, comprising: It comprises: A memory and at least one processor; Wherein, the memory has stored a computer program; The at least one processor executes the computer program stored in the memory, so that the at least one processor executes the multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium has stored a computer program, and the computer program can be executed by the processor to realize the multi-tenant cloud computer operation management platform concurrent request intelligent scheduling method according to any one of claims 1 to 7.

Images