A chain server non-perception application inter-cloud deployment method and system

Abstract

The application discloses a kind of chain server non-perception application's intercloud deployment method and system, the method of the present application includes obtaining the application deployment request submitted by user, including the code of each function in chain server non-perception application And deployment condition including chain relationship;All functions of chain server non-perception application are deployed in the form of single function on all intercloud nodes, and the FaaS service and BaaS service used by function are limited on single intercloud node;Small data is used to trigger the function running of chain server non-perception application in batches;Query function running log and collect the running data of chain server non-perception application;According to the running data collected and preset cost model and running delay model, select the best deployment scheme.The present application aims to shield the heterogeneity of different intercloud nodes, and carry out cross-cloud deployment for multiple functions in chain server non-perception application to reduce cost and running delay.

Description

Technical Field

[0001] The present invention relates to the technical field of cloud application deployment, and in particular to a cloud deployment method and system for chained server-unaware applications. Background Art

[0002] Cloud computing is a network-based computing model. Its core concept is to provide computing resources (including computing power, storage, and network resources) as services to users through virtualization technology. Users can access and release these resources as needed. Cloud computing is a new generation of cloud computing that aims to achieve the "1+1>2" capability by combining multiple clouds. Based on open collaboration between multiple cloud service entities, cloud computing deeply integrates multi-cloud resources and customizes cloud services through a "software-defined" approach. This allows developers to leverage resources from multiple cloud providers, creating value beyond the cloud itself. The recently emerging new cloud computing paradigm, serverless computing, with its lightweight and stateless nature, is injecting new vitality into cloud computing.

[0003] Serverless services can be thought of as a combination of FaaS (Function as a Service) and BaaS (Backend as a Service). BaaS encompasses a broad range of third-party services, such as classic object storage services, database storage services, and caching services, providing users with various foundational components. FaaS represents user-defined functions, typically running in short-lived function instances (sandboxes, such as microVMs and containers) with constrained resources like CPU and memory. FaaS acts as the "glue" between the underlying infrastructure components (also known as BaaS services), providing computing power and connecting BaaS services such as message queues, object storage events, and database binlogs to efficiently complete user requests. In a serverless architecture, infrastructure management is effectively separated from application development, allowing application developers to focus on writing functions that perform specific tasks and setting corresponding triggers without having to worry about the underlying infrastructure. When a trigger is activated, the serverless computing platform automatically allocates a sandbox (container) for the function to run. However, users may face a number of challenges when actually using serverless services. First, the performance delivery capabilities of serverless platforms are often unknown. Because resource allocation and scheduling are automatically maintained by the platform, it is difficult for users to accurately understand the underlying resource allocation. Although some micro-benchmarks and application tests exist, the performance of serverless platforms remains difficult to clearly quantify. Second, the function services provided by different serverless platforms vary significantly in performance specifications and cost models, and different cloud service providers also have different technology selections and optimization solutions. Therefore, users need to select the most appropriate function-as-a-service platform for deployment based on their workload characteristics to achieve the best balance between performance and cost-effectiveness. Finally, vendor lock-in is also a key consideration. Due to geographical distribution or policy restrictions, business data and computing resources may be distributed across different cloud service providers. However, data and computing capacity migration between different cloud service providers may be subject to certain limitations, including differences in interfaces and fees for data migration. Although some cloud service providers are working to develop open standards, migration between serverless providers can still involve challenges. Therefore, there is an urgent need to design a cross-cloud collaboration mechanism that can shield the heterogeneity of different serverless applications and deploy multiple functions in chained serverless applications (chained server-unaware applications) across clouds to reduce costs and latency. Summary of the Invention

[0004] Technical problem to be solved by the present invention: In response to the above-mentioned problems of the prior art, a cloud deployment method and system for chain server unaware applications are provided. The present invention aims to shield the heterogeneity of different cloud nodes and perform cross-cloud deployment of multiple functions in chain server unaware applications to reduce costs and operating delays.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0006] A cloud deployment method for chained server-unaware applications includes the following steps:

[0007] S1, obtaining an application deployment request submitted by a user, wherein the application deployment request includes the code of each function in the chain server-unaware application and the deployment conditions including the chain relationship;

[0008] S2 deploys all functions of the chained server-unaware application as a single function on all cloud nodes. The FaaS and BaaS services used by the function are limited to a single cloud node.

[0009] S3 uses small data to batch trigger the function execution of chained servers without the perception of applications;

[0010] S4, query the function's running log to collect the chain server's non-aware application's running data;

[0011] S5: Select the best deployment solution based on the collected operation data and the preset cost model and operation delay model.

[0012] Optionally, in step S2, when deploying all functions of the chained server-unaware application on all cloud nodes in the form of a single function, it includes: obtaining the general interface call information of different FaaS services and BaaS services for all cloud nodes in the function, converting the general interface call information into API and SDK calls of the dedicated FaaS services and BaaS services of the corresponding cloud nodes, thereby deploying all functions of the chained server-unaware application on all cloud nodes in the form of a single function.

[0013] Optionally, in step S3, using small data to batch trigger the function execution of the chain server unaware application means using test data that is smaller than the benchmark test of the chain server unaware application to batch trigger the function execution of the chain server unaware application.

[0014] Optionally, the function expression of the cost model preset in step S5 is:

[0015]

[0016] In the above formula, C is the total operating cost, N is the number of functions in the chain server-unaware application, Deploy the i-th function on cloud node j i Storage cost C_S on i , Deploy the i-th function on cloud node j i Operating cost C_E i , The output result of the i-th function is from the cloud node j i Transmit to cloud node j i+1 The transmission cost C_Tran i .

[0017] Optionally, the storage cost C_S i , operating costs C_E i and the transmission cost C_Tran i The function expression is:

[0018] C_S i =D_St i *R_St i +1*R_OP i ,

[0019]

[0020] C_Tran i =D_St i+1 *D_StOut i ,

[0021] In the above formula, D_St i is the size of the storage data of the function at the i-th cloud node, R_St i is the storage unit price, R_OP i is the unit price of data operation, T i is the running time, M i is the memory usage of the i-th function, R Mem_i is the unit price of memory, CPU i is the CPU usage of the i-th function, R cpu_i is the CPU unit price, is the price of a single request, D_St i+1 D_StOut is the size of the storage data of the function at the i+1th cloud node. i is the data outflow price of the i-th cloud node.

[0022] Optionally, the function expression of the running delay model preset in step S5 is:

[0023]

[0024] In the above formula, D is the total running delay, N is the number of functions in the chain server unaware application, Deploy the i-th function on cloud node j i Trigger delay on Deploy the i-th function on cloud node j i The running delay on The output result of the i-th function is from the cloud node j i Transmit to cloud node j i+1 The transmission delay on , and:

[0025]

[0026] In the above formula, D_St i+1 is the size of the storage data of the function at the i+1th cloud node, The output result of the i-th function is from the cloud node j i Transmit to cloud node j i+1 The transmission speed on.

[0027] Optionally, when selecting the best deployment solution based on the collected operation data and the preset cost model and operation delay model in step S5, the following indicators are constructed based on the preset cost model and operation delay model:

[0028] In the above formula, α is the weight parameter, is the normalized delay, is the normalized cost, and:

[0029]

[0030] In the above formula, Deploy the i-th function on cloud node j i Trigger delay on Deploy the i-th function on cloud node j i The running delay on The output result of the i-th function is from the cloud node j i-1 Transmit to cloud node j i The transmission delay on Deploy the i-th function on cloud node j i Storage cost C_S on i , Deploy the i-th function on cloud node j i Operating cost C_E i , The output result of the i-th function is from the cloud node ji-1 Transmit to cloud node j i The transmission cost C_Tran i ; Combined indicators The beam search algorithm is used to find the best deployment solution.

[0031] In addition, the present invention also provides a cloud deployment system for chain server unaware applications, including a microprocessor and a memory connected to each other, and the microprocessor is programmed or configured to execute the cloud deployment method for chain server unaware applications.

[0032] In addition, the present invention also provides a computer-readable storage medium, which stores a computer program or instruction, and the computer program or instruction is programmed or configured to execute the cloud deployment method of the chain server-unaware application through a processor.

[0033] In addition, the present invention also provides a computer program product, including a computer program or instructions, which are programmed or configured to execute the cloud deployment method of the chain server-unaware application through a processor.

[0034] Compared with the prior art, the present invention mainly has the following advantages: due to the existence of certain differences in performance and pricing among different cloud platforms, cross-cloud deployment becomes necessary. The method of the present invention provides a cloud deployment mechanism for chain server unaware applications, including obtaining an application deployment request submitted by a user, including the code of each function in the chain server unaware application and the deployment conditions including the chain relationship; deploying all functions of the chain server unaware application on all cloud nodes in the form of a single function, and the FaaS service and BaaS service used by the function are limited to a single cloud node; batch triggering the function operation of the chain server unaware application in the form of small data; querying the function operation log to collect the operation data of the chain server unaware application; selecting the best deployment plan based on the collected operation data and the preset cost model and operation delay model, which can shield the heterogeneity of different cloud nodes, and perform cross-cloud deployment for multiple functions in the chain server unaware application to reduce costs and operation delays, thereby effectively helping users choose a suitable cross-cloud deployment plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 Schematic diagram of the basic process of the method of the embodiment of the present invention.

[0036] Figure 2 The figure is a flow chart of finding the best deployment solution by using the beam search algorithm in an embodiment of the present invention. DETAILED DESCRIPTION

[0037] like Figure 1 As shown, the cloud deployment method of the chained server-unaware application in this embodiment includes the following steps:

[0038] S1, obtaining an application deployment request submitted by a user, wherein the application deployment request includes the code of each function in the chain server-unaware application and the deployment conditions including the chain relationship;

[0039] S2 deploys all functions of the chained server-unaware application as a single function on all cloud nodes. The FaaS and BaaS services used by the function are limited to a single cloud node.

[0040] S3 uses small data to batch trigger the function execution of chained servers without the perception of applications;

[0041] S4, query the function's running log to collect the chain server's non-aware application's running data;

[0042] S5: Select the best deployment solution based on the collected operation data and the preset cost model and operation delay model.

[0043] As an optional implementation, steps S1 to S5 of this embodiment are specifically executed using a deployment optimizer as the execution subject. When the deployment optimizer obtains the application deployment request submitted by the user, the format of the application deployment request can be selected according to the actual needs to select a feasible implementation. For example, as an optional implementation, the code of each function in the chain server unaware application in step S1 of this embodiment is in the form of a container or compressed package, and the deployment conditions of the chain relationship are in the format of a json file. In addition, other forms can be used to submit the code of each function in the chain server unaware application and the deployment conditions of the chain relationship as needed.

[0044] The step S2 of the embodiment includes: obtaining the general interface call information of the functions of the chain server application for different FaaS services and BaaS services of all the inter-cloud nodes, converting the general interface call information to the API and SDK calls of the corresponding special FaaS services and BaaS services of the inter-cloud nodes, and thus deploying all the functions of the chain server application in the form of a single function on all the inter-cloud nodes. The deployment optimizer deploys all the functions in the form of a single function on all the inter-cloud nodes, and the FaaS and BaaS services used by the functions are limited on a single inter-cloud node. Different cloud vendors achieve platform locking by setting obstacles at the API and SDK levels. When service deployment is needed on different server application platforms, the developer needs to adapt the API and SDK of different cloud platforms, which greatly reduces the development efficiency. Therefore, the embodiment designs related components to help users flexibly use various inter-cloud nodes. Specifically, on the one hand, the embodiment designs a set of general interfaces for different FaaS and BaaS platforms, which serves as an abstraction layer of cloud-specific API and SDK. When the developer develops using the general interface, automatic conversion can be performed to adapt to different inter-cloud nodes. On the other hand, considering that the developer is usually familiar with the interface of a certain cloud, in order to avoid increasing the burden of the developer, the embodiment further provides an interface conversion function from one cloud vendor to another, so that the code of a certain inter-cloud node can be easily ported to other server application platforms.

[0045] In step S3 of the embodiment, the function of the chain server application is triggered in batches in the small data mode, which means that test data smaller than the benchmark test of the chain server application is used to trigger the function of the chain server application in batches. In order to help users select appropriate inter-cloud nodes, the embodiment uses the small data batch triggering mode to test the performance and cost of different inter-cloud nodes under different workloads. It should be noted that the embodiment does not use benchmark tests to measure the performance of different inter-cloud nodes. This is because the benchmark test of the inter-cloud node cannot directly reflect the real performance of the user function on different server application platforms. In addition, the pay-as-you-go mode of the server application platform makes the small batch data triggering mode not generate too much cost. The embodiment can obtain the corresponding triggering delay, running time and other performance characteristics by querying the log records of different inter-cloud nodes.

[0046] The function expression of the preset cost model in step S5 of the embodiment is:

[0047]

[0048] In the above formula, C is the total operating cost, N is the number of functions in the chain server-unaware application, Deploy the i-th function on cloud node j i Storage cost C_S on i , Deploy the i-th function on cloud node j i Operating cost C_E i , The output result of the i-th function is from the cloud node j i Transmit to cloud node j i+1 The transmission cost C_Tran i , that is, the jth i The public network outflow fee of each cloud node.

[0049] About operating costs C_E i The billing items of different cloud platforms are usually composed of two parts: resource usage fee and request fee. The resource usage fee depends on the function running time and resource configuration. i ,The storage services of different cloud platforms are charged based on the size of the stored data and the operations (read, write and delete) on the stored data. i When data flows from the cloud to the public network, major cloud vendors will charge based on the amount of data flowing into the public network. Therefore, in this embodiment, the storage cost C_S i , operating costs C_E i and the transmission cost C_Tran i The function expression is:

[0050] C_S i =D_St i *R_St i +1*R_OP i ,

[0051]

[0052] C_Tran i =D_St i+1 *D_StOut i ,

[0053] In the above formula, D_St i is the size of the storage data of the function at the i-th cloud node, R_St i is the storage unit price, R_OP i is the unit price of data operation, T i is the running time, M i is the memory usage of the i-th function, R Mem_i is the unit price of memory, CPUi is the CPU usage of the i-th function, R cpu_i is the CPU unit price, is the price of a single request, D_St i+1 D_StOut is the size of the storage data of the function at the i+1th cloud node. i is the data outflow price of the i-th cloud node.

[0054] When a user function is deployed across clouds, its total running delay mainly consists of three parts: trigger delay Operation delay and data transmission delays Regarding trigger delay, when a function is triggered to run through a storage trigger, it is usually called in the form of an asynchronous trigger. At this time, the function request needs to wait in the queue for a period of time, so there will be a certain trigger delay; regarding running delay, the function is charged according to the actual running time, and its running delay is obtained by reading the function execution log; regarding transmission delay, after the function is executed, its output result cannot be directly passed to other functions, but needs to be handed over to the next function with the assistance of an intermediate service, such as passing it to the storage service of this cloud or other clouds. At this time, there is a certain transmission delay. In actual calculations, for data transmission in the same data center, the transmission delay between its function service and the storage service can usually be ignored, while different cloud vendors or data centers need to be transmitted through the public network. At this time, the transmission delay is determined by the amount of data transmitted and the network transmission speed. Therefore, the function expression of the preset running delay model in step S5 of this embodiment is:

[0055]

[0056] In the above formula, D is the total running delay, N is the number of functions in the chain server unaware application, Deploy the i-th function on cloud node j i Trigger delay on Deploy the i-th function on cloud node j i The running delay on The output result of the i-th function is from the cloud node j i Transmit to cloud node j i+1 The transmission delay on , and:

[0057]

[0058] In the above formula, D_St i+1 is the size of the storage data of the function at the i+1th cloud node, The output result of the i-th function is from the cloud node j i Transmit to cloud node ji+1 The transmission speed on.

[0059] In step S5 of this embodiment, when selecting the best deployment solution based on the collected operation data and the preset cost model and operation delay model, it includes constructing the following indicator based on the preset cost model and operation delay model:

[0060] In the above formula, α is the weight parameter, is the normalized delay, is the normalized cost, and:

[0061]

[0062] In the above formula, Deploy the i-th function on cloud node j i Trigger delay on Deploy the i-th function on cloud node j i The running delay on The output result of the i-th function is from the cloud node j i-1 Transmit to cloud node j i The transmission delay on Deploy the i-th function on cloud node j i Storage cost C_S on i , Deploy the i-th function on cloud node j i Operating cost C_E i , The output result of the i-th function is from the cloud node j i-1 Transmit to cloud node j i The transmission cost C_Tran i ; Combined indicators The beam search algorithm is used to find the best deployment solution. In order to comprehensively consider the operation delay and operation cost, an indicator is constructed in step S5 of this embodiment. Among them, α is specified by the user, and the α indicator represents the user's trade-off between cost and operation delay. For example, when α = 0, it means that the user wants to minimize the cost and does not consider the impact of delay, while when α = 1, it means that the user minimizes delay without considering the impact of cost. Secondly, in order to avoid the influence of the value size of different stages, this embodiment uses normalized calculation and The upper part of the fraction represents the time when the i-th function is deployed in the j-th iThe delay and cost on each cloud platform, and the lower half of the fraction represents the maximum delay and cost when all functions are deployed on different cloud platforms. Finally, this embodiment uses the existing beam search algorithm to find the best deployment solution, such as Figure 2 As shown, the steps of the beam search algorithm to find the best deployment solution include: 1) Initialize an empty priority queue as the bundle B of the beam search algorithm; 2) Each function in the chain server-unaware application has several deployment options, and all the options of all N functions are constructed into a tree, the root node r of the tree is initialized, and it is inserted into the bundle B with the initial score, and the first layer is initialized as the current layer; 3) Determine whether the number of functions that have been searched (i.e., the path length) is less than N. If so, jump to the next step, otherwise jump to step 7). 4) Initialize the temporary bundle B' as an empty priority queue; 5) For each node n in the bundle B, generate each child node c of n (up to P child nodes), combined with the indicator The function expression calculates the index of the child node c Insert (c, score) into the temporary bundle B'. 6) Select the first k nodes with the lowest scores from the temporary bundle B' as the next layer of bundle B, and skip to step 3. 7) Finally, select the path associated with the lowest score as the best path and return it. The resulting best path best_path is the optimal deployment solution.

[0063] To summarize, in response to the problems such as certain differences in performance and pricing among different cloud platforms, the method of this embodiment provides a cloud deployment mechanism for chained server-unaware applications, including obtaining an application deployment request submitted by a user, including the code of each function in the chained server-unaware application and the deployment conditions including the chain relationship; deploying all functions of the chained server-unaware application as a single function on all cloud nodes, and the FaaS service and BaaS service used by the function are limited to a single cloud node; using small data to batch trigger the function execution of the chained server-unaware application; querying the function's execution log to collect the operation data of the chained server-unaware application; based on the collected operation data and the preset cost model and operation delay model, using the beam search algorithm combined with the indicator model constructed based on the preset cost model and operation delay model to implement deployment search to select the best deployment solution, which can shield the heterogeneity of different cloud nodes, and perform cross-cloud deployment for multiple functions in the chained server-unaware application to reduce costs and operation delays, thereby effectively helping users choose a suitable cross-cloud deployment solution.

[0064] In addition, this embodiment also provides a cloud deployment system for chain server unaware applications, including a microprocessor and a memory connected to each other, and the microprocessor is programmed or configured to execute the cloud deployment method for chain server unaware applications.

[0065] In addition, this embodiment also provides a computer-readable storage medium, which stores a computer program or instruction, and the computer program or instruction is programmed or configured to execute the cloud deployment method of the chain server-unaware application through a processor.

[0066] In addition, this embodiment also provides a computer program product, including a computer program or instructions, which are programmed or configured to execute the cloud deployment method of the chain server-unaware application through a processor.

[0067] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, may be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including the instruction device, which implements the function specified in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the process. Figure 1a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0068] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the concept of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A cloud deployment method for chained server-unaware applications, characterized in that: The steps include: S1, obtaining an application deployment request submitted by a user, wherein the application deployment request includes the code of each function in the chain server-unaware application and the deployment conditions including the chain relationship; S2 deploys all functions of the chained server-unaware application as a single function on all cloud nodes. The FaaS and BaaS services used by the function are limited to a single cloud node. S3 uses small data to batch trigger the function execution of chained servers without the perception of applications; S4, query the function's running log to collect the chain server's non-aware application's running data; S5, select the best deployment plan based on the collected operation data and the preset cost model and operation delay model, including constructing the indicators shown in the following formula based on the preset cost model and operation delay model To choose the best deployment option: ， In the above formula, is the weight parameter, is the normalized delay, is the normalized cost, and: ， ， In the above formula, For the Functions are deployed on cloud nodes Trigger delay on For the Functions are deployed on cloud nodes The running delay on For the The output of the function is from the cloud node Transmit to cloud nodes The transmission delay on For the Functions are deployed on cloud nodes Storage costs on , For the Functions are deployed on cloud nodes Operating costs , For the The output of the function is from the cloud node Transmit to cloud nodes Transmission costs ; Combined indicators A beam search algorithm is used to find the best deployment solution, including: 1) initializing an empty priority queue as the beam B of the beam search algorithm; 2) For the various deployment options of each function in the chained server-unaware application, construct all the options of all N functions into a tree, initialize the root node r of the tree, insert the root node r with the initial score into the bundle B, and initialize the first layer as the current layer; 3) Determine whether the number of searched functions is less than N. If so, jump to the next step, otherwise jump to step 7); 4) Initialize the temporary bundle B' as an empty priority queue; 5) For each node n in the bundle B, generate n child nodes c, combined with the indicator The function expression calculates the index of the child node c ,Will Insert it into the temporary bundle B'; 6) Select the first k nodes with the lowest score from the temporary bundle B' as the bundle B of the next layer, and jump to step 3); 7) Finally, select the path associated with the lowest score as the optimal path and return it. The obtained optimal path is the best deployment solution.

2. The cloud deployment method for chained server-unaware applications according to claim 1, characterized in that: In step S2, when all functions of the chain server unaware application are deployed on all cloud nodes in the form of a single function, it includes: obtaining the general interface call information of different FaaS services and BaaS services for all cloud nodes in the function, converting the general interface call information into the API and SDK calls of the dedicated FaaS services and BaaS services of the corresponding cloud nodes, thereby deploying all functions of the chain server unaware application on all cloud nodes in the form of a single function.

3. The cloud deployment method of chained server-unaware applications according to claim 1, characterized in that: In step S3, using small data to batch trigger the function running of the chain server unaware application means using test data that is smaller than the benchmark test of the chain server unaware application to batch trigger the function running of the chain server unaware application.

4. The cloud deployment method of chained server-unaware applications according to claim 1, characterized in that: The function expression of the cost model preset in step S5 is: ， In the above formula, is the total operating cost, is the number of functions in the chained server-unaware application, For the Functions are deployed on cloud nodes Storage costs on , For the Functions are deployed on cloud nodes Operating costs , For the The output of the function is from the cloud node Transmit to cloud nodes Transmission costs .

5. The cloud deployment method of chained server-unaware applications according to claim 4 is characterized in that: The storage cost , operating costs and transmission costs The function expression is: ， ， ， In the above formula, For the function The size of the storage data of each cloud node, is the storage unit price, is the unit price of data operation, For the running time, For the The memory usage of the function, is the unit price of memory, For the The CPU usage of each function, is the CPU unit price, is the price of a single request, For the function The size of the storage data of each cloud node, For the The data outflow price of each cloud node.

6. The cloud deployment method of chained server-unaware applications according to claim 5, characterized in that: The function expression of the operation delay model preset in step S5 is: ， In the above formula, is the total running delay, is the number of functions in the chained server-unaware application, For the Functions are deployed on cloud nodes Trigger delay on For the Functions are deployed on cloud nodes The running delay on For the The output of the function is from the cloud node Transmit to cloud nodes The transmission delay on , and: ， In the above formula, For the function The size of the storage data of each cloud node, For the The output of the function is from the cloud node Transmit to cloud nodes The transmission speed on.

7. A cloud deployment system for chained server-unaware applications, comprising interconnected microprocessors and memories, characterized in that: The microprocessor is programmed or configured to execute the cloud deployment method of chain server-unaware applications as described in any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program or instruction stored therein, characterized in that: The computer program or instruction is programmed or configured to execute the cloud deployment method of the chain server unaware application described in any one of claims 1 to 6 through a processor.

9. A computer program product comprising a computer program or instructions, characterized in that The computer program or instruction is programmed or configured to execute the cloud deployment method of the chain server unaware application described in any one of claims 1 to 6 through a processor.

Images