A cost optimization method under a multi-cloud environment

Abstract

The application discloses a cost optimization method in a multi-cloud environment, and particularly relates to the technical field of cost optimization; the method comprises the following steps: collecting multi-cloud charging rule data, and performing unified parameter format conversion to construct a standardized multi-cloud charging structure model; generating a multi-dimensional cost influence factor matrix based on the multi-cloud charging structure model; extracting discount mutual exclusion and pricing mismatch characteristic data based on the multi-dimensional cost influence factor matrix; constructing business load adaptability analysis data based on the multi-dimensional cost influence factor matrix; combining the discount mutual exclusion and pricing mismatch characteristic data and the business load adaptability analysis data to construct a global cost optimization objective function and a parameter constraint set, and performing optimal distribution on business loads in multi-cloud resources to output a business deployment scheme with minimized cost; and the method reduces the multi-cloud deployment cost of users and improves resource utilization.

Description

Technical Field

[0001] This invention relates to the field of cost optimization technology, and more specifically, to a cost optimization method in a multi-cloud environment. Background Technology

[0002] As cloud computing technology matures and becomes more widespread, enterprise users typically deploy their businesses using resources from multiple cloud service providers. Currently, each cloud service provider sets its own billing rules, and these rules differ. Therefore, in actual resource deployment and usage, enterprise users usually need to analyze the billing rules of each cloud service provider separately, making it difficult to accurately assess and predict the overall resource usage costs in a multi-cloud environment.

[0003] Existing multi-cloud resource cost analysis methods are unable to effectively address the complex interactions between the differentiated billing rules of multiple cloud service providers, resulting in enterprise users being unable to obtain a global cost optimization strategy for multi-cloud environments. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a cost optimization method in a multi-cloud environment to address the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] A cost optimization method for multi-cloud environments includes the following steps:

[0007] S1. Collect multi-cloud billing rule data from cloud service providers in a multi-cloud environment, perform unified parameter format conversion, and generate multi-cloud billing structure model data.

[0008] S2. Based on the multi-cloud billing structure model data, analyze the interaction characteristics of discount thresholds and segmented outbound pricing among cloud service providers, and construct a multi-dimensional cost influencing factor matrix.

[0009] S3. Based on the multidimensional cost influencing factor matrix, identify the overlap between the quota threshold trigger interval and the outbound traffic step change interval, and extract the discount mutual exclusion and pricing mismatch feature data.

[0010] S4. Based on the multidimensional cost influencing factor matrix, evaluate the adaptability of business traffic to different cloud resource modes and outbound billing strategies, and generate business load adaptability analysis data.

[0011] S5. Based on the characteristics of discount mutual exclusion and pricing mismatch, and the business load adaptability analysis data, establish a global cost optimization objective function and parameter constraint set;

[0012] S6. Based on the global cost optimization objective function and parameter constraint set, solve the optimal allocation strategy of business load in multi-cloud resources and output the business deployment scheme with minimal cost.

[0013] In a preferred embodiment, S1 specifically refers to:

[0014] By obtaining resource discount strategy parameters, outbound traffic tiered billing parameters, and real-time business traffic monitoring data from the public application programming interfaces of cloud service providers, the initial multi-cloud billing rule data file is generated.

[0015] The initial multi-cloud billing rule data file is mapped to fields, billing units are unified, timestamps are aligned, and missing values ​​are filled to obtain the multi-cloud billing rule dataset.

[0016] A cost element correspondence table is established based on the multi-cloud billing rule dataset, and a multi-cloud billing structure model data is constructed through hierarchical grouping and key-value indexing.

[0017] In a preferred embodiment, S2 specifically refers to:

[0018] Based on the multi-cloud billing structure model data, the discount threshold value sequence of resource discount strategy parameters and the segmented rate sequence of outbound traffic tiered billing parameters are extracted according to the cloud service provider identifier and resource type identifier.

[0019] Perform interval intersection operation on the discount threshold value sequence and the segmented fee rate sequence, and map the business traffic monitoring data to each overlapping interval to obtain the corresponding business traffic interval load value;

[0020] A multidimensional cost influencing factor matrix is ​​constructed using cloud service provider identifier, resource type identifier, pricing granularity identifier, interval overlap identifier, and business traffic interval load value as dimensions.

[0021] In a preferred embodiment, S3 specifically refers to:

[0022] The endpoints of the discount threshold range and the segmented rate range are extracted based on the multidimensional cost influencing factor matrix;

[0023] The intersection of the discount threshold range and the segmented fee rate range is calculated to determine the overlapping range, and an index of the overlapping range is established.

[0024] Based on the overlapping interval index, match the service traffic interval load value to generate discount mutual exclusion identifier data and pricing mismatch identifier data;

[0025] The discount mutual exclusion identifier data and the pricing mismatch identifier data are aggregated using cloud service provider identifier, resource type identifier, pricing granularity identifier and overlapping interval index to obtain discount mutual exclusion and pricing mismatch feature data.

[0026] In a preferred embodiment, S4 specifically refers to:

[0027] Based on the multidimensional cost influencing factor matrix, extract the resource discount strategy parameters and outbound traffic tiered billing parameters for each cloud service provider's resource type.

[0028] Based on the resource discount strategy parameters and outbound traffic tiered billing parameters, calculate the discount applicability index and billing tier adaptation index for each type of business traffic in different cloud service providers' resource types.

[0029] Based on the load value of the business traffic range, the discount applicability index and the billing tier adaptation index are weighted and calculated to obtain the discount adaptation weight and the billing adaptation weight.

[0030] Based on the cloud service provider identifier, resource type identifier, and pricing granularity identifier, the discount adaptation weight and billing adaptation weight are combined to generate business load adaptability analysis data.

[0031] In a preferred embodiment, S5 specifically refers to:

[0032] Based on the characteristics of mutually exclusive discounts and pricing mismatch, the discount threshold of the resource discount strategy and the rate jump point in the outbound traffic tiered billing parameters are determined as piecewise constraints of the objective function.

[0033] Based on the business load adaptability analysis data, the product relationship between the business traffic range load value and the discount adaptability weight and the billing adaptability weight is determined as a linear combination term of the objective function;

[0034] Establish a mapping relationship between multi-cloud resource configuration variables and business traffic range load values, and construct decision variables in the optimization objective function;

[0035] An optimization objective function and parameter constraint set are established using piecewise constraints, linear combination terms, and decision variables, forming a global cost optimization objective function and parameter constraint set.

[0036] In a preferred embodiment, S6 specifically refers to:

[0037] Based on the global cost optimization objective function and the set of parameter constraints, a mixed integer optimization method is selected as the solution algorithm.

[0038] An integer linear programming model is constructed by linearizing the piecewise constraints by introducing interval indicator binary variables for each rate jump point;

[0039] Based on the integer linear programming model, the solution algorithm is invoked to perform variable relaxation, branch and bound, and iterative update to obtain the optimal solution vector of multi-cloud resource configuration variables;

[0040] The optimal solution vector of multi-cloud resource configuration variables is mapped to a mapping table between business load and multi-cloud resources, generating a business deployment plan with minimal cost.

[0041] The technical effects and advantages of the cost optimization method in a multi-cloud environment of the present invention are as follows:

[0042] By standardizing the parameter format conversion of multi-cloud billing rule data from cloud service providers in a multi-cloud environment, multi-cloud billing structure model data is obtained, avoiding cost calculation deviations caused by inconsistent standards. A multi-dimensional cost influencing factor matrix is ​​constructed to systematically characterize the interaction between discount thresholds and tiered billing, improving the visualization and quantifiability of cost influencing factors. By identifying the overlap between the threshold trigger interval and the outbound traffic tiered jump interval, the phenomenon of discount exclusivity and pricing mismatch is effectively revealed, providing targeted input for optimization strategies. Furthermore, by evaluating the impact of business traffic on different cloud resource commitment models... The system assesses the compatibility between the billing model and outbound billing strategies, quantifies the fit between business load and billing conditions, and enhances decision-making transparency. Based on discount mutual exclusion and pricing mismatch characteristic data and business load adaptability analysis data, a global cost optimization objective function and parameter constraint set are established, which can accurately characterize cost constraints and optimization objectives. By solving the optimal allocation strategy of business load in multi-cloud resources, a business deployment scheme with minimized costs is obtained, achieving global minimization of business deployment costs. This improves the efficiency and accuracy of cost optimization, reduces cloud resource procurement and usage costs, and enhances the elasticity of multi-cloud deployment. Attached Figure Description

[0043] Figure 1 This is a schematic diagram of a cost optimization method for a multi-cloud environment according to the present invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0045] Example

[0046] Figure 1 This invention presents a cost optimization method for a multi-cloud environment, comprising the following steps:

[0047] S1. Collect multi-cloud billing rule data from cloud service providers in a multi-cloud environment, perform unified parameter format conversion, and generate multi-cloud billing structure model data.

[0048] S2. Based on the multi-cloud billing structure model data, analyze the interaction characteristics of discount thresholds and segmented outbound pricing among cloud service providers, and construct a multi-dimensional cost influencing factor matrix.

[0049] S3. Based on the multidimensional cost influencing factor matrix, identify the overlap between the quota threshold trigger interval and the outbound traffic step change interval, and extract the discount mutual exclusion and pricing mismatch feature data.

[0050] S4. Based on the multidimensional cost influencing factor matrix, evaluate the adaptability of business traffic to different cloud resource modes and outbound billing strategies, and generate business load adaptability analysis data.

[0051] S5. Based on the characteristics of discount mutual exclusion and pricing mismatch, and the business load adaptability analysis data, establish a global cost optimization objective function and parameter constraint set;

[0052] S6. Based on the global cost optimization objective function and parameter constraint set, solve the optimal allocation strategy of business load in multi-cloud resources and output the business deployment scheme with minimal cost.

[0053] S1. Collect multi-cloud billing rule data from cloud service providers in a multi-cloud environment, perform unified parameter format conversion, and generate multi-cloud billing structure model data, including:

[0054] By obtaining resource discount strategy parameters, outbound traffic tiered billing parameters, and real-time business traffic monitoring data from the public application programming interfaces of cloud service providers, the initial multi-cloud billing rule data file is generated.

[0055] Cloud service provider public application programming interfaces (APIs) refer to standardized data interfaces provided by various cloud service providers to users via the internet. Accessing these APIs allows users to obtain billing-related data from cloud service providers, including resource discount policy parameters, outbound traffic tiered billing parameters, and real-time business traffic monitoring data. Resource discount policy parameters refer to the policy data of cloud service providers regarding fee reductions and price discounts offered to users for long-term purchases, fixed resource reservations, or advance payments. These parameters include resource reservation limits, discount percentages, usage periods, discount thresholds, and applicable discount scope. For example, a user paying 12 months' worth of virtual computing instance fees in advance through cloud service provider A may receive a certain percentage discount; the discount percentage, duration, and applicable instance type constitute the resource discount policy parameters. Outbound traffic tiered billing parameters refer to the differentiated billing standards set by cloud service providers for users under different outbound traffic usage scales. That is, as the total outbound traffic usage of users increases, cloud service providers will gradually reduce the cost per unit of traffic. The data of the cost standards constitute the outbound traffic tiered billing parameters. For example, cloud service provider B charges users 0.12 yuan per GB for monthly outbound traffic between 0 and 10TB. When the monthly outbound traffic of users is between 10TB and 50TB, the cost per GB is reduced to 0.10 yuan. After exceeding 50TB, the rate continues to decrease. Similar billing and corresponding billing ranges form the outbound traffic tiered billing parameters. Real-time business traffic monitoring data refers to the traffic data generated during the operation of actual business systems deployed by users on various cloud service providers. This includes the actual usage of business traffic, traffic volume, traffic source and destination, traffic type, and information on traffic distribution within the cloud service provider. For example, if a user deploys an e-commerce business system on cloud service provider A, and sends 2TB of data files daily from provider A to provider B, and transmits 1TB of data files daily from provider B to provider A, the real-time monitoring data constitutes the real-time business traffic monitoring data. Resource discount strategy parameters, outbound traffic tiered billing parameters, and real-time business traffic monitoring data are merged based on the cloud service provider's identifier and the same time window to form the initial multi-cloud billing rule data file.

[0056] The initial multi-cloud billing rule data file is mapped to fields, billing units are unified, timestamps are aligned, and missing values ​​are filled to obtain the multi-cloud billing rule dataset.

[0057] Field mapping refers to standardizing field names that have the same meaning but inconsistent naming conventions in data provided by different cloud service providers. This establishes unified data field names. For example, provider A might use "resource reservation quota" to represent pre-booked resources, while provider B might use "committed quota" for the same meaning. Field mapping will standardize these to "resource discount committed quota." Billing unit unification refers to standardizing the differences in billing data units used by different cloud service providers, unifying them to the same unit of measurement. For example, different cloud service providers might use different currencies or express traffic in MB or GB. In this case, all currencies need to be unified, and all traffic units need to be converted to GB to ensure a unified billing unit. Timestamp alignment refers to unifying the differences in data collection frequency across different cloud service providers. For example, one provider might collect traffic data every 5 minutes, while another collects it every hour. In this case, the timestamps need to be aligned to the hourly level, allowing for unified comparison and analysis of data within the same time interval. Missing value imputation refers to supplementing missing data in the initial multi-cloud billing rule data file due to incomplete data collection caused by various reasons. For example, missing values ​​can be added based on the mean of historical data or interpolation algorithms. After performing the above-mentioned field mapping, billing unit unification, timestamp alignment, and missing value imputation on the initial multi-cloud billing rule data file, a standardized multi-cloud billing rule dataset is obtained.

[0058] A cost element correspondence table is established based on the multi-cloud billing rule dataset, and a multi-cloud billing structure model data is constructed through hierarchical grouping and key-value indexing.

[0059] The cost element mapping table establishes a one-to-one correspondence between resource discount strategy parameters, outbound traffic tiered billing parameters, and real-time business traffic monitoring data, using cloud service provider, resource type, and pricing granularity as the main dimensions. For example, in the compute instance resource type of cloud service provider A, with hourly pricing granularity, the discount strategy parameters of the resource type, the corresponding outbound traffic tiered billing parameters, and the real-time monitored business traffic data for that period are correlated to form corresponding entries. Corresponding entries for other providers or other resource types are then established sequentially to form the cost element mapping table. Hierarchical grouping refers to grouping data hierarchically according to the order of cloud service provider, resource type, and pricing granularity. For example, all data is first divided into multiple provider groups according to cloud service provider, then further subdivided by resource type within each provider group, and then further subdivided by pricing granularity under each resource type, forming a clear hierarchical structure. Key-value indexes assign a unique index identifier to each data item in the cost element mapping table for quick access and querying of data. For example, an identifier is created for the hourly pricing granularity data item of the virtual instance resource type of provider A, allowing for quick data retrieval through the identifier. Multi-cloud billing structure model data can be constructed by using hierarchical grouping and key-value indexing.

[0060] S2. Based on multi-cloud billing structure model data, analyze the interaction characteristics of discount thresholds and segmented outbound pricing among cloud service providers, and construct a multi-dimensional cost influencing factor matrix, including:

[0061] Based on the multi-cloud billing structure model data, the discount threshold value sequence of resource discount strategy parameters and the segmented rate sequence of outbound traffic tiered billing parameters are extracted according to the cloud service provider identifier and resource type identifier.

[0062] The discount threshold sequence in resource discount strategy parameters refers to the conditions and thresholds under which cloud service providers offer users fee reductions or discounts. It is usually presented in a segmented form, meaning that when a user reserves or commits to using a certain number of resources, they can receive a corresponding percentage discount. The threshold values ​​are arranged in ascending order to form a sequence. For example, cloud service provider A sets three discount threshold values ​​for long-term reservations or advance payments of computing resources: the first threshold is 50 virtual computing instances, receiving a 10% discount upon reaching this threshold; the second threshold is 100 virtual computing instances, receiving a 15% discount; and the third threshold is 200 virtual computing instances, receiving a 20% discount. These three threshold values ​​constitute the discount threshold sequence. Similarly, the segmented rate sequence in outbound traffic tiered billing parameters refers to the differentiated price rate standards set by cloud service providers based on different outbound traffic usage amounts. These rates are segmented based on different traffic usage intervals. For example, cloud service provider B stipulates that the outbound traffic is charged at RMB 0.12 per GB in the range of 0 to 10TB; RMB 0.10 per GB in the range of 10TB to 50TB; and RMB 0.08 per GB in the range of 50TB to 200TB, which constitutes a segmented fee rate sequence.

[0063] Perform interval intersection operation on the discount threshold value sequence and the segmented fee rate sequence, and map the business traffic monitoring data to each overlapping interval to obtain the corresponding business traffic interval load value;

[0064] Interval intersection operation refers to comparing each interval in the discount threshold value sequence and the segmented fee rate sequence to find the overlapping intervals, thus obtaining the overlapping intervals that simultaneously satisfy the discount threshold condition and the segmented fee rate condition. For example, cloud service provider A's discount threshold value sequence includes three intervals: 0-50 instances, 50-100 instances, and 100-200 instances; cloud service provider A's outbound traffic tiered billing parameters for the same resource type include three intervals: 0-80TB, 80-150TB, and 150-300TB. Through interval intersection operation, the intervals of the above sequences are intersected to obtain new overlapping intervals; for example, the calculated overlapping intervals are 0-50 instances corresponding to 0-80TB, 50-100 instances corresponding to 80-150TB, and 100-200 instances corresponding to 150-300TB. Each overlapping interval is a composite interval that simultaneously meets the discount threshold and fee rate jump conditions. Mapping business traffic monitoring data to each overlapping interval means matching the real-time monitored business traffic data with the cloud service provider, resource type, and actual business load. For example, if a user's actual business traffic usage on provider A is 120TB and the user has 120 instances, then the business traffic of 120TB and the number of instances both fall within the overlapping intervals of 100-200 instances and 150-300TB. The overlapping interval can then be used to determine the corresponding business traffic load value as 120TB. By accurately matching business traffic data to each overlapping interval in this way, the business traffic load value is obtained and used to establish a multi-dimensional cost influence factor matrix.

[0065] A multidimensional cost influencing factor matrix is ​​constructed using cloud service provider identifier, resource type identifier, pricing granularity identifier, interval overlap identifier, and business traffic interval load value as dimensions;

[0066] A multidimensional cost impact factor matrix is ​​a matrix that combines data from various dimensions. In this matrix, the cloud service provider identifier represents the name of the provider (e.g., provider A); the resource type identifier represents the type of resource (e.g., compute instance resources); the pricing granularity identifier represents the time scale for resource billing (e.g., hourly or monthly billing); the interval overlap identifier represents the interval information that simultaneously meets the discount threshold and segmented rate conditions (e.g., 100-200 instances and 150-300TB); and the business traffic interval load value represents the overlap. The actual business traffic value within the overlapping interval, such as the mapped 120TB business traffic value; using cloud service provider identifier as the first-level dimension, resource type identifier as the second-level dimension, pricing granularity identifier as the third-level dimension, interval overlap identifier as the fourth-level dimension, and business traffic interval load value as the data value in the matrix, a multi-dimensional structured data matrix is ​​constructed; for example, in the multi-dimensional structured data matrix, a data record can be represented as a combination of dimensions such as provider A, computing instance resources, monthly billing granularity, 100-200 instances / 150-300TB interval, and 120TB business load value. By establishing a multi-dimensional cost influencing factor matrix, users can accurately grasp the discount strategies and segmented billing conditions corresponding to business traffic under different cloud service providers, resource types, and pricing conditions, thereby achieving cost analysis, optimization, and decision-making.

[0067] S3. Based on a multi-dimensional cost influencing factor matrix, identify the overlap between the quota threshold trigger interval and the outbound traffic step change interval, and extract discount mutual exclusion and pricing mismatch feature data, including:

[0068] The endpoints of the discount threshold range and the segmented rate range are extracted based on the multidimensional cost influencing factor matrix;

[0069] The endpoints of a discount threshold range refer to the starting and ending values ​​of the range of resource quantities for which a discount is applied, as defined in the resource discount strategy parameters. For example, if cloud service provider A specifies a discount strategy for virtual computing instance resource types, defining ranges of 0-50 instances, 50-100 instances, and 100-200 instances, then the starting values ​​0, 50, and 100, and the ending values ​​50, 100, and 200 within these ranges constitute the endpoints of the discount threshold range. The endpoints of a segmented rate range refer to the starting and ending values ​​of traffic usage within different rate applicable ranges, as defined in the outbound traffic tiered billing parameters. For example, if cloud service provider A defines outbound traffic billing rate ranges of 0-80TB, 80-150TB, and 150-300TB for the same resource type, then the starting values ​​0, 80, and 150, and the ending values ​​80, 150, and 300 within these ranges constitute the endpoints of the segmented rate range.

[0070] The intersection of the discount threshold range and the segmented fee rate range is calculated to determine the overlapping range, and an index of the overlapping range is established.

[0071] The extracted discount threshold ranges and segmented fee rate ranges are compared separately, and the common range that simultaneously satisfies both the discount and fee rate conditions is calculated; this is called the overlapping range. For example, taking cloud service provider A as an example, the discount threshold ranges include 0-50 instances, 50-100 instances, and 100-200 instances, while the segmented fee rate ranges include 0-80TB, 80-150TB, and 150-300TB. Through intersection calculation, the 0-50 instances of the discount threshold and the 0-80TB of the fee rate range form an overlapping range of 0-50 instances / 0-80TB; the 50-100 instances of the discount threshold and the 80-150TB of the fee rate range form an overlapping range of 50-100 instances / 80-150TB; and the 100-200 instances of the discount threshold and the 150-300TB of the fee rate range form an overlapping range of 100-200 instances / 150-300TB. The calculation results of overlapping intervals indicate that the business load meets both the specific discount threshold and the corresponding outbound traffic rate conditions. Each overlapping interval reflects the cost characteristics under different combinations. Establishing an overlapping interval index involves assigning a unique index identifier to each calculated overlapping interval for quick location and use. For example, the 0-50 instances / 0-80TB interval is labeled as overlapping interval index "001", the 50-100 instances / 80-150TB interval as overlapping interval index "002", and the 100-200 instances / 150-300TB interval as overlapping interval index "003". By creating a unique index for each overlapping interval, the overlapping intervals can be referenced and matched, enabling efficient mapping and corresponding analysis of business load values.

[0072] Based on the overlapping interval index, match the service traffic interval load value to generate discount mutual exclusion identifier data and pricing mismatch identifier data;

[0073] Overlapping interval index matching of business traffic interval load values ​​refers to determining the overlapping interval into which the actual business traffic interval load value falls by combining real-time monitored business traffic data according to a combination of cloud service provider identifiers, resource type identifiers, and pricing granularity identifiers. For example, if a user actually deploys 120 virtual computing instances on provider A and generates 120TB of outbound traffic, the overlapping interval index can match a third overlapping interval, 100-200 instances / 150-300TB, i.e., index "003," with a corresponding business traffic interval load value of 120TB. Through this matching process, discount mutual exclusion identifier data and pricing mismatch identifier data are generated. The discount exclusion flag indicates that when the load value of a business traffic range is unevenly distributed across different cloud service providers, the discount thresholds defined by each provider cannot be fully met, leading to discount exclusion. For example, a user deploys 45 instances on provider A and 55 instances on provider B, for a total of 100 instances, reaching provider B's threshold of 100 instances. However, due to the distributed deployment, the threshold is only reached on provider B, not on provider A, which is a discount exclusion. The pricing mismatch flag indicates that when the load value of a business traffic range is distributed across multiple cloud service providers, the increased cost of rate jumps due to the lack of coordination among segmented rate ranges. For example, a user generates 75TB of outbound traffic on provider A and 75TB on provider B respectively, and is billed separately at a higher unit rate range. If the traffic is concentrated on provider A, the user's total 150TB of outbound traffic can enter a lower rate tier. Therefore, the current situation is a pricing mismatch.

[0074] The discount mutual exclusion identifier data and the pricing mismatch identifier data are aggregated by cloud service provider identifier, resource type identifier, pricing granularity identifier and overlapping interval index to obtain discount mutual exclusion and pricing mismatch feature data;

[0075] Aggregation operations refer to combining and summarizing discount exclusion and pricing mismatch identification data according to four dimensions: cloud service provider, resource type, billing granularity, and overlapping interval index, to form discount exclusion and pricing mismatch feature data that is easy to analyze and query. For example, for provider A, compute instance resources, monthly billing granularity, and overlapping interval index "003", the number of times the discount exclusion and pricing mismatch identification occur under the business traffic interval load value is counted and recorded. The feature data records can show the occurrence of discount exclusion and pricing mismatch under different provider, resource type, billing granularity, and interval conditions.

[0076] S4. Based on a multidimensional cost influencing factor matrix, evaluate the adaptability of business traffic to different cloud resource models and outbound billing strategies, and generate business load adaptability analysis data, including:

[0077] Based on the multidimensional cost influencing factor matrix, extract the resource discount strategy parameters and outbound traffic tiered billing parameters for each cloud service provider's resource type.

[0078] Resource discount strategy parameters describe the fee discount conditions and discounts offered by cloud service providers to users when their committed long-term resource usage reaches a certain amount. For example, cloud service provider A might set resource discount strategy parameters for computing resource types, stipulating a 15% discount when resource usage reaches 100 instances and a 20% discount when it reaches 200 instances. The discount rate and usage amount correspond to specific providers and resource types. Outbound traffic tiered billing parameters define the segmented tiered rates set by cloud service providers based on the different monthly or annual outbound traffic usage scales of users. For example, cloud service provider B might charge users 0.12 yuan per GB for monthly outbound traffic in the 0-10TB range, 0.10 yuan per GB for the 10-50TB range, and a rate that decreases to 0.08 yuan per GB for traffic exceeding 50TB. The rates correspond to the outbound traffic usage scale range and the cloud service provider's identifier. Therefore, through a multi-dimensional cost influencing factor matrix, resource discount strategy parameters and outbound traffic tiered billing parameters for each cloud service provider under different resource types are extracted.

[0079] Based on the resource discount strategy parameters and outbound traffic tiered billing parameters, calculate the discount applicability index and billing tier adaptation index for each type of business traffic in different cloud service providers' resource types.

[0080] The discount applicability index is used to measure the actual applicability of the discount policy defined by the cloud service provider to the user under the current resource usage level. For example, if a user deploys 120 virtual instances of computing resources on cloud service provider A, and according to the resource discount policy parameters of provider A, a 20% discount can be obtained in the range of 100-200 instances, then the discount applicability index corresponding to the user in the range is 20%. If the number of resources deployed by the user is at the edge of the discount threshold, such as 99 instances, which does not reach the threshold of 100 instances, then the discount applicability index is reduced, such as 0 or a lower discount percentage, to reflect the actual applicability of the discount policy. The billing tier adaptation index describes the actual degree of matching between the business traffic load value and the outbound traffic tiered billing parameters set by the cloud service provider. For example, if a user's monthly outbound traffic is 120TB, according to the tiered billing parameters defined by cloud service provider B, the rate per GB for outbound traffic is 0.10 yuan in the 10-50TB range and 0.08 yuan per GB in the 50-200TB range. The user's actual load of 120TB falls within the more favorable 50-200TB range. Therefore, the billing tier adaptation index indicates that the user is in the optimal tier range and is given a higher adaptation index accordingly.

[0081] Based on the load value of the business traffic range, the discount applicability index and the billing tier adaptation index are weighted and calculated to obtain the discount adaptation weight and the billing adaptation weight.

[0082] The service traffic range load value refers to the actual service traffic usage of a user under specific vendor, resource type, and billing granularity conditions, representing the actual amount of cloud resources consumed by the user. For example, if a user deploys 120 virtual instances and generates 120TB of outbound traffic, this is the service traffic range load value. The discount adaptation weight and billing adaptation weight are calculated by multiplying the service traffic range load value by the discount applicability index and the billing tier adaptation index, respectively. For example, if a user deploys 120 virtual compute instances with vendor A, and the discount applicability index is 20%, then multiplying the discount applicability index by 120 instances yields a discount adaptation weight of 24. The billing tier adaptation index for the user's 120TB of outbound traffic is at the optimal tier (0.08 yuan per GB), and multiplying 120TB by the billing tier adaptation index yields a corresponding billing adaptation weight of 9.6TB yuan. Through weighted calculation, the actual adaptation status of discounts and billing to different service traffic loads can be quantified.

[0083] Based on the cloud service provider identifier, resource type identifier, and pricing granularity identifier, the discount adaptation weight and the billing adaptation weight are combined to generate business load adaptability analysis data.

[0084] The workload adaptability analysis data records the combined combination of discount adaptation weights and billing adaptation weights, organized and displayed according to cloud service provider identifiers, resource type identifiers, and billing granularity identifiers. For example, for cloud service provider A's virtual computing instance resources and monthly billing granularity, the data records the combination of discount adaptation weight 24 and billing adaptation weight 9.6TB yuan; similarly, for provider B's storage resource type and hourly billing granularity, the corresponding combinations of discount adaptation weights and billing adaptation weights are recorded. By combining discount adaptation weights and billing adaptation weights into workload adaptability analysis data, users can understand the adaptability between actual resource discounts and outbound traffic tiered billing strategies under each workload scenario.

[0085] S5. Based on discount mutual exclusion and pricing mismatch characteristic data and business load adaptability analysis data, establish a global cost optimization objective function and parameter constraint set, including:

[0086] Based on the characteristics of mutually exclusive discounts and pricing mismatch, the discount threshold of the resource discount strategy and the rate jump point in the outbound traffic tiered billing parameters are determined as piecewise constraints of the objective function.

[0087] The discount threshold in a resource discount strategy refers to the numerical point at which a user commits to purchasing or long-term reservations of cloud resources. Different thresholds correspond to different discount rates. For example, a user purchasing 50, 100, and 200 virtual instances from cloud service provider A might receive a 10%, 15%, and 20% discount respectively. These virtual instance numbers correspond to different thresholds, forming a series of discount thresholds. The rate jump point in outbound traffic tiered billing parameters refers to the numerical point at which a user's monthly or annual traffic usage reaches a preset threshold, resulting in a significant reduction in the rate. For example, cloud service provider B might offer tiered rates of 0.12 yuan per GB, 0.10 yuan per GB, and 0.08 yuan per GB for monthly outbound traffic in the ranges of 0-10TB, 10-50TB, and 50-200TB respectively. The rate jump points are at 10TB and 50TB, demonstrating the significant reduction in unit billing cost after traffic usage increases to the preset threshold. The segmented constraints consist of the discount threshold and rate jump points mentioned above. These segmented constraints on the objective function ensure that the optimization process adheres to the discount policies under different threshold conditions and the billing standards under different rate tiers, guaranteeing that the optimization results accurately reflect the actual cost policies provided by cloud service providers. For example, when a user actually deploys 120 virtual instances and has 120TB of outbound traffic per month, the discount threshold corresponds to a 20% discount constraint for the 100-200 instance range, and the rate jump point corresponds to a constraint of 0.08 yuan per GB for the 50-200TB range. These segmented constraints, consisting of the discount threshold and rate jump points, ensure that the objective function is calculated based on real-world conditions during optimization.

[0088] Based on the business load adaptability analysis data, the product relationship between the business traffic range load value and the discount adaptability weight and the billing adaptability weight is determined as a linear combination term of the objective function;

[0089] The business traffic range load value represents the actual amount of resources a user uses within a specific cloud service provider, resource type, and billing cycle, such as 120 virtual instances and 120TB of outbound traffic. The discount adaptation weight quantifies the combination of user resource usage and discount applicability indicators. For example, a user's actual usage of 120 instances combined with a 20% discount applicability indicator yields a discount adaptation weight of 24, reflecting the match between resource usage scale and discount benefits. The billing adaptation weight quantifies the combination of user business traffic usage and tiered pricing rate applicability indicators. For example, a user's monthly outbound traffic of 120TB falls within the 50-200TB range with a rate tier of 0.08 yuan per GB. Multiplying the business traffic range load value of 120TB by the applicability indicator of 0.08 yuan yields a billing adaptation weight of 9.6TB yuan, reflecting the actual fit between traffic usage scale and tiered pricing rates. In the linear combination term of the objective function, the load value of the business traffic interval, the discount adaptation weight, and the billing adaptation weight are represented as a product. The linear combination term is formed by multiplying the weight coefficient by the load value of the business traffic interval, which is expressed as the number of business instances 120 multiplied by the discount weight 24, and the business traffic 120TB multiplied by the billing weight 9.6TB yuan. The above linear combination term reflects the real cost relationship between the user's resource usage scale and the cloud service provider's cost policy, and is used to construct the objective function to optimize the cost structure.

[0090] Establish a mapping relationship between multi-cloud resource configuration variables and business traffic range load values, and construct decision variables in the optimization objective function;

[0091] Multi-cloud resource configuration variables represent the actual number of resources deployed by a user across various cloud service providers and the allocation of outbound traffic. For example, a user might deploy 60 virtual computing instances (VCIs) with provider A and 60 VCIs with provider B, and the user's actual monthly outbound traffic is 60TB with provider A and 60TB with provider B. The business traffic range load value, for example, 120 VCIs and 120TB of traffic, represents the actual number of resources used in the user's business scenario. By establishing a mapping relationship, multi-cloud resource configuration variables are correlated with business traffic range load values. For instance, the 60 instances from provider A are mapped to the corresponding proportion within the total load of 120 instances, meaning that provider A's actual number of instances accounts for 50% of the total business load instance count, and the same applies to provider B. Similarly, the 60TB of outbound traffic each from providers A and B is also mapped to a 50% share within the total 120TB business traffic range load value. By establishing the above mapping relationship, the decision variables in the objective function are constructed, namely the proportion of the actual resource usage of each cloud service provider to the overall business traffic range load value. This ensures that the resource deployment ratio of each provider can be adjusted when optimizing the objective function to achieve the cost optimization goal.

[0092] By establishing the set of optimization objective functions and parameter constraints using piecewise constraints, linear combination terms, and decision variables, a global cost optimization objective function and parameter constraint set is formed.

[0093] The objective function, also known as the cost function, is primarily expressed as a linear combination of business traffic range load values, discount adaptation weights, and billing adaptation weights. It uses the proportional mapping of multi-cloud resource configuration variables as decision variables and is constrained by segmented constraints related to discount thresholds and rate jump points. For example, the objective function is a linear combination of 120 instances, a discount adaptation weight of 24, 120TB outbound traffic, and a billing adaptation weight of 9.6TB. It uses the resource usage ratio of vendors A and B as decision variables and is subject to segmented constraints such as instance number thresholds of 100 and 200, and traffic thresholds of 50TB and 150TB. Specifically:

[0094] Total cost = Number of instances × Discount adaptation weight + Outbound traffic × Billing adaptation weight;

[0095] The set of parameter constraints is that the resource usage ratio decision variable must satisfy the condition that the sum of the total resource usage of all vendors equals the business traffic range load value of 120 instances and 120TB of traffic, and also satisfy the segmented constraints of discount thresholds and rate jump points of all vendors; a cost optimization mathematical model is established through the above optimization objective function and parameter constraint set.

[0096] S6. Based on the global cost optimization objective function and parameter constraint set, solve for the optimal allocation strategy of business load in multi-cloud resources, and output the business deployment scheme that minimizes cost, including:

[0097] Based on the global cost optimization objective function and the set of parameter constraints, a mixed integer optimization method is selected as the solution algorithm.

[0098] Since the actual number of deployed resources, the resource discount thresholds provided by cloud service providers, and the traffic tiered billing structure are all discrete integers, a mixed-integer optimization method is used. This method uses integer variables to represent discrete decisions and continuous variables to represent continuously adjustable resource allocation ratios, thus finding the optimal solution for multi-cloud resource configuration variables. For example, if a user needs to allocate a total of 120 virtual computing instances between cloud service provider A and provider B, and the discount thresholds are specified as discrete integer points such as 50, 100, and 200 instances, the user cannot choose non-integer options like 55.5 instances and must determine the resource configuration scheme in integer form. In this case, the mixed-integer optimization method combines the discrete and continuous properties of the decision variables, selecting the number of instances through integer variables and determining the traffic allocation ratio through continuous variables, ultimately determining the most cost-effective solution. Choosing the mixed-integer optimization method as the solution algorithm allows the optimization process to adapt to the actual requirements of the business scenario, accurately reflect the integer characteristics of the discount thresholds and traffic billing structures, and ultimately achieve the lowest-cost resource deployment scheme.

[0099] An integer linear programming model is constructed by linearizing the piecewise constraints by introducing interval indicator binary variables for each rate jump point;

[0100] The binary variable indicating the rate jump point refers to a 0-1 binary variable introduced for each jump point in the outbound traffic tiered billing structure. This variable represents whether the actual traffic load exceeds the rate jump point. For example, if cloud service provider B provides outbound traffic rate jump points at 10TB and 50TB, and the user's actual monthly traffic is 120TB, the first binary variable indicates whether the 10TB rate jump point is exceeded (since the actual traffic exceeds 10TB, the variable is set to 1). The second variable indicates whether the 50TB rate jump point is exceeded (since the actual traffic is 120TB, exceeding 50TB, the variable is also set to 1). By introducing binary variables, the non-linear relationship of rate jumps can be linearized. For example, if the user's actual traffic is in the range of 0-10TB, the first binary variable takes the value of 0; when the user's actual traffic is in the range of 10-50TB, the first variable takes the value of 1 and the second variable takes the value of 0; when the user's actual traffic is greater than 50TB, both variables take the value of 1. In this case, the traffic and the corresponding segmented rate exhibit a linear relationship. By introducing the above binary variables, the originally non-linear rate jump condition is transformed into a linear constraint, thereby constructing an integer linear programming model.

[0101] Based on the integer linear programming model, the solution algorithm is invoked to perform variable relaxation, branch and bound, and iterative update to obtain the optimal solution vector of multi-cloud resource configuration variables;

[0102] After establishing the integer linear programming model, a mixed-integer optimization method is used to solve it. The solution steps include variable relaxation, branch and bound, and iterative update. Variable relaxation refers to temporarily ignoring the constraint that integer variables must take integer values ​​during the optimization process, allowing variables to temporarily take non-integer solutions. The purpose is to obtain an initial lower bound for the objective function. For example, relaxing the constraint that the number of virtual computing instances must be an integer allows the number of instances in the solution to be temporarily a decimal, thus quickly obtaining an initial optimized solution. The branch and bound method refers to gradually refining the variable constraints based on the initial solution obtained from variable relaxation, determining the range of integer values. For example, if the number of instances is 65.7 in the first solution, two subproblems with integer solutions of 65 and 66 instances are determined and solved separately, thereby continuously narrowing the search range. The objective function value of each branch serves as the boundary condition for the next branch. When an objective function value worse than an existing solution appears, that branch is abandoned, which is called bounding. Iterative update refers to repeatedly performing variable relaxation and branch and bound, continuously updating the objective function value until the globally optimal solution under integer conditions is found, thus obtaining the lowest-cost multi-cloud resource configuration scheme. For example, after iterative updates, the optimal solution vector was determined to be deploying 70 virtual instances on cloud service provider A and 50 instances on provider B, with 70TB of outbound traffic from provider A and 50TB of outbound traffic from provider B. Through variable relaxation, branch and bound, and iterative updates, the final optimal solution vector of the multi-cloud resource configuration variables reflects the cost structure in the optimization objective function, ensuring that the optimization result is the solution required for actual deployment.

[0103] The optimal solution vector of multi-cloud resource configuration variables is mapped to a mapping table between business load and multi-cloud resources, generating a business deployment plan with minimal cost.

[0104] The optimal solution vector for multi-cloud resource configuration variables represents the allocation of instance counts and outbound traffic among different cloud service providers. For example, 70 virtual computing instances are configured with Provider A and 50 with Provider B, with outbound traffic of 70TB for Provider A and 50TB for Provider B. Based on the optimal solution vector for multi-cloud resource configuration variables, a business load-multi-cloud resource mapping table is constructed using cloud service provider identifiers, resource type identifiers, and billing granularity identifiers as dimensions. For example, Provider A records the deployment of 70 instances and 70TB of traffic usage under the monthly billing granularity of virtual computing instance resource type, while Provider B records 50 instances and 50TB of traffic usage. The business load-multi-cloud resource mapping table records the actual resource configuration and traffic usage of the business system deployment, such as the resource configuration and actual traffic allocation of each cloud service provider, forming an executable business deployment plan. For example, the mapping table lists: Cloud service provider A, virtual computing instance resources, monthly billing granularity, 70 instances, 70TB of traffic; Cloud service provider B, virtual computing instance resources, monthly billing granularity, 50 instances, 50TB of traffic. The mapping table reflects the deployment scheme of business load in multi-cloud resources after solving the objective function of global cost optimization. It fully meets the constraints of discount threshold and tiered rate structure, ensuring that the deployment scheme is feasible and the cost is minimized.

[0105] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium can be a solid-state drive.

[0106] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0107] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0108] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0109] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0110] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0111] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0112] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0113] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A cost optimization method in a multi-cloud environment, characterized in that, Includes the following steps: S1. Collect multi-cloud billing rule data from cloud service providers in a multi-cloud environment, perform unified parameter format conversion, and generate multi-cloud billing structure model data. S2. Based on the multi-cloud billing structure model data, analyze the interaction characteristics of discount thresholds and segmented outbound pricing among cloud service providers, and construct a multi-dimensional cost influencing factor matrix. S3. Based on the multidimensional cost influencing factor matrix, identify the overlap between the quota threshold trigger interval and the outbound traffic step change interval, and extract the discount mutual exclusion and pricing mismatch feature data. The mutual exclusion of discounts means that when the load value of the business traffic range is unevenly distributed among different cloud service providers, it is impossible to fully meet the discount thresholds defined by each provider, resulting in mutual exclusion of discounts. The pricing mismatch refers to the increased cost of rate jumps caused by the failure to coordinate and align segmented rate ranges after the load value of the business traffic range is distributed among multiple cloud service providers. S4. Based on the multidimensional cost influencing factor matrix, evaluate the adaptability of business traffic to different cloud resource modes and outbound billing strategies, and generate business load adaptability analysis data. S5. Based on the characteristics of discount mutual exclusion and pricing mismatch, and the business load adaptability analysis data, establish a global cost optimization objective function and parameter constraint set; S6. Based on the global cost optimization objective function and parameter constraint set, solve the optimal allocation strategy of business load in multi-cloud resources and output the business deployment scheme with minimal cost.

2. The cost optimization method in a multi-cloud environment according to claim 1, characterized in that, S1, specifically: By obtaining resource discount strategy parameters, outbound traffic tiered billing parameters, and real-time business traffic monitoring data from the public application programming interfaces of cloud service providers, the initial multi-cloud billing rule data file is generated. The initial multi-cloud billing rule data file is mapped to fields, billing units are unified, timestamps are aligned, and missing values ​​are filled to obtain the multi-cloud billing rule dataset. A cost element correspondence table is established based on the multi-cloud billing rule dataset, and a multi-cloud billing structure model data is constructed through hierarchical grouping and key-value indexing.

3. The cost optimization method in a multi-cloud environment according to claim 2, characterized in that, S2, specifically: Based on the multi-cloud billing structure model data, the discount threshold value sequence of resource discount strategy parameters and the segmented rate sequence of outbound traffic tiered billing parameters are extracted according to the cloud service provider identifier and resource type identifier. Perform interval intersection operation on the discount threshold value sequence and the segmented fee rate sequence, and map the business traffic monitoring data to each overlapping interval to obtain the corresponding business traffic interval load value; A multidimensional cost influencing factor matrix is ​​constructed using cloud service provider identifier, resource type identifier, pricing granularity identifier, interval overlap identifier, and business traffic interval load value as dimensions.

4. The cost optimization method in a multi-cloud environment according to claim 3, characterized in that, S3, specifically: The endpoints of the discount threshold range and the segmented rate range are extracted based on the multidimensional cost influencing factor matrix; The intersection of the discount threshold range and the segmented fee rate range is calculated to determine the overlapping range, and an index of the overlapping range is established. Based on the overlapping interval index, match the service traffic interval load value to generate discount mutual exclusion identifier data and pricing mismatch identifier data; The discount mutual exclusion identifier data and the pricing mismatch identifier data are aggregated using cloud service provider identifier, resource type identifier, pricing granularity identifier and overlapping interval index to obtain discount mutual exclusion and pricing mismatch feature data.

5. The cost optimization method in a multi-cloud environment according to claim 4, characterized in that, S4, specifically: Based on the multidimensional cost influencing factor matrix, extract the resource discount strategy parameters and outbound traffic tiered billing parameters for each cloud service provider's resource type. Based on the resource discount strategy parameters and outbound traffic tiered billing parameters, calculate the discount applicability index and billing tier adaptation index for each type of business traffic in different cloud service providers' resource types. Based on the load value of the business traffic range, the discount applicability index and the billing tier adaptation index are weighted and calculated to obtain the discount adaptation weight and the billing adaptation weight. Based on the cloud service provider identifier, resource type identifier, and pricing granularity identifier, the discount adaptation weight and billing adaptation weight are combined to generate business load adaptability analysis data.

6. The cost optimization method in a multi-cloud environment according to claim 5, characterized in that, S5, specifically: Based on the characteristics of mutually exclusive discounts and pricing mismatch, the discount threshold of the resource discount strategy and the rate jump point in the outbound traffic tiered billing parameters are determined as piecewise constraints of the objective function. Based on the business load adaptability analysis data, the product relationship between the business traffic range load value and the discount adaptability weight and the billing adaptability weight is determined as a linear combination term of the objective function; Establish a mapping relationship between multi-cloud resource configuration variables and business traffic range load values, and construct decision variables in the optimization objective function; An optimization objective function and parameter constraint set are established using piecewise constraints, linear combination terms, and decision variables, forming a global cost optimization objective function and parameter constraint set.

7. The cost optimization method in a multi-cloud environment according to claim 6, characterized in that, S6, specifically: Based on the global cost optimization objective function and the set of parameter constraints, a mixed integer optimization method is selected as the solution algorithm. An integer linear programming model is constructed by linearizing the piecewise constraints by introducing interval indicator binary variables for each rate jump point; Based on the integer linear programming model, the solution algorithm is invoked to perform variable relaxation, branch and bound, and iterative update to obtain the optimal solution vector of multi-cloud resource configuration variables; The optimal solution vector of multi-cloud resource configuration variables is mapped to a mapping table between business load and multi-cloud resources, generating a business deployment plan with minimal cost.

Images