Diversified computing power configuration system and method

Abstract

The invention provides a diversified computing power configuration system and method. The system comprises a resource layer, a sensing layer, a quantization layer and an arrangement layer, the sensing layer is used for acquiring at least one target computing node from the plurality of computing nodes according to the resource information of each computing node in the resource layer and the mirror image architecture label in the resource encapsulation creation request, packaging a mapping matrix between the architecture information and the function information of each target computing node, and analyzing the action information and the architecture feature information corresponding to the resource packaging creation request; the quantization layer is used for carrying out computing power evaluation on each target computing node according to the mapping matrix, the action information and the architecture feature information, and carrying out computing power balanced scheduling and computing power resource configuration according to a plurality of computing power evaluation results; and the arrangement layer is used for arranging and configuring the application program corresponding to the computing power task to be executed according to the computing power balance scheduling result and the computing power resource configuration result. According to the invention, the rationality of diversified computing power configuration and the computing power service performance are improved.

Description

Technical Field

[0001] The present invention relates to the field of cloud-edge computing technology, and in particular to a diversified computing power configuration system and method. Background Art

[0002] In the digital economy era, computing power continues to grow, but it also faces problems such as uneven distribution of computing power, diversified computing power types, and imbalance between supply and demand, which makes it difficult to improve computing power utilization. Therefore, how to reasonably configure diversified computing power is an important issue that the industry needs to solve urgently.

[0003] In the related technology, diversified computing power configuration is unified through the integration of diversified computing power. Specifically, the performance of each computing power device is uniformly modeled through a data-driven method to obtain a unified performance model; the performance data reflected by the hardware device layer is mapped to the performance events recorded in the kernel state layer of the operating system; the performance event layering is implemented in the kernel state layer of the operating system to obtain a platform-related event layer and a platform-independent event layer; the performance events of the platform-independent event layer are aggregated to the user state layer of the operating system; when the computing power device undergoes iterative changes, dynamic changes are achieved by adjusting the mapping relationship between the performance data reflected by the hardware device layer and the performance events of the platform-related event layer.

[0004] However, this computing power configuration method only uses a data-driven approach, converting problems such as the performance behavior description and classification of heterogeneous computing power devices into performance data collection and fusion problems. It is difficult to accurately evaluate and compare the performance of heterogeneous computing power devices, which makes it difficult to ensure the rationality of computing power configuration. Summary of the invention

[0005] The present invention provides a diversified computing power configuration system and method, which are used to solve the defect that the rationality of computing power configuration is difficult to ensure in the prior art, and to improve the rationality of diversified computing power configuration and computing power service performance.

[0006] The present invention provides a diversified computing power configuration system, including: a resource layer, a perception layer, a quantization layer and an orchestration layer; The resource layer includes a plurality of computing nodes of different architectures; The perception layer is used to perceive and obtain resource information of each of the computing nodes, and obtain at least one target computing node from the multiple computing nodes according to the resource information and the image architecture tag in the resource encapsulation creation request, and encapsulate and establish a mapping matrix between the architecture information and function information of each of the target computing nodes, and parse and obtain action information and architecture feature information corresponding to the resource encapsulation creation request; The quantization layer is used to perform computing power evaluation on each of the target computing nodes according to the mapping matrix, the action information and the architecture feature information, obtain multiple computing power evaluation results corresponding to each of the target computing nodes, and perform computing power balancing scheduling and computing power resource allocation on each of the target computing nodes according to the multiple computing power evaluation results; The orchestration layer is used to orchestrate and configure the application corresponding to the computing task to be executed according to the computing power balancing scheduling results and computing power resource configuration results of each target computing node.

[0007] According to a diversified computing power configuration system provided by the present invention, the perception layer includes a collector, an interceptor and a scheduler; The collector is used to sense and obtain resource information of each computing node; The interceptor is used to identify the image architecture tag in the resource encapsulation creation request, match the image architecture tag with the architecture information in the resource information of each computing node, and filter out at least one target computing node matching the image architecture tag from the plurality of computing nodes according to the matching result, and encapsulate and establish a mapping matrix between the architecture information and the function information of each target computing node; The scheduler is used to parse and obtain the action information and architecture feature information corresponding to the resource package creation request, and display the action information and the architecture feature information.

[0008] According to a diversified computing power configuration system provided by the present invention, the quantization layer is further used for: According to the mapping matrix, the action information and the architecture feature information, a multi-level computing power evaluation is performed on each of the target computing nodes to obtain a plurality of computing power evaluation results corresponding to each of the target computing nodes; the plurality of computing power evaluation results include a specification computing power evaluation result, an effective computing power evaluation result and a business computing power evaluation result, the specification computing power evaluation result is an evaluation result of a whole machine specification performance benchmark test on each of the target computing nodes, the effective computing power evaluation result is an evaluation result of a middleware resource encapsulation performance test on each of the target computing nodes, and the business computing power evaluation result is an evaluation result of a business load performance test on each of the target computing nodes; According to the task type and task requirement corresponding to the computing task to be executed, obtaining a first computing power evaluation result from the multiple computing power evaluation results; According to the first computing power evaluation result, perform equivalent computing power evaluation on each of the target computing nodes to obtain a second computing power evaluation result corresponding to each of the target computing nodes; Determining a scheduling priority level of each of the target computing nodes according to the second computing power evaluation result; According to the scheduling priority level, computing power balancing scheduling is performed on each of the target computing nodes.

[0009] According to a diversified computing power configuration system provided by the present invention, the quantization layer is further used for: In a case where the first computing power evaluation result is the standard computing power evaluation result, obtaining the standard computing power coefficient of each resource in each of the target computing nodes according to the standard computing power evaluation result; Combined with the specification computing power coefficient, as well as the maximum and minimum normalized values ​​and weight coefficients corresponding to the allocatable computing power of each resource in each target computing node, an equivalent computing power evaluation is performed to obtain a second computing power evaluation result of each target computing node.

[0010] According to a diversified computing power configuration system provided by the present invention, the quantization layer is further used for: According to the historical records of the computing task to be executed, a load trend model of the computing task to be executed is obtained; the load trend model is used to characterize the relationship between the application load, service quality and resource capacity corresponding to the computing task to be executed; Determine an elastic scaling mode corresponding to each of the target computing nodes according to the load trend model and the plurality of computing power evaluation results; According to the elastic scaling mode, computing power resources are configured for each of the target computing nodes.

[0011] According to a diversified computing power configuration system provided by the present invention, the orchestration layer is further used for: According to the computing power balancing scheduling result and the computing power resource configuration result, obtaining at least one node to be configured in at least one of the target computing nodes; According to the operating environment and operating tools of each node to be configured, the source code of the application corresponding to the computing task to be executed is cross-compiled across nodes to obtain the binary code corresponding to each node to be configured; Containerizing and packaging the binary code corresponding to each of the nodes to be configured and the dependency information of each of the binary codes; According to the packaging format of the containerized package, the binary code corresponding to each of the nodes to be configured is deployed in a standardized manner.

[0012] The present invention also provides a method for configuring diversified computing power, which is applied to any of the above-mentioned diversified computing power configuration systems, and the method includes: Based on the perception layer, the resource information of each computing node is perceived and acquired, and according to the resource information and the image architecture label in the resource encapsulation creation request, at least one target computing node is acquired from the multiple computing nodes, and a mapping matrix between the architecture information and the function information of each target computing node is encapsulated and established, and the action information and the architecture feature information corresponding to the resource encapsulation creation request are parsed and acquired; Based on the quantization layer, according to the mapping matrix, the action information and the architecture feature information, a computing power evaluation is performed on each of the target computing nodes to obtain a plurality of computing power evaluation results corresponding to each of the target computing nodes, and computing power balancing scheduling and computing power resource allocation are performed on each of the target computing nodes according to the plurality of computing power evaluation results; Based on the orchestration layer, according to the computing power balancing scheduling results and computing power resource configuration results of each of the target computing nodes, the application programs corresponding to the computing power tasks to be executed are orchestrated and configured.

[0013] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the diversified computing power configuration method as described above is implemented.

[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described diversified computing power configuration methods.

[0015] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned diversified computing power configuration methods.

[0016] The diversified computing power configuration system and method provided by the present invention collects computing power resource information of computing nodes using different architectures through resource perception capability; then uses computing power quantification to evaluate the equivalence relationship between the computing powers of different computing nodes, and implements computing power balanced scheduling and computing power elastic scaling based on the computing power evaluation results. Finally, standardized orchestration and deployment of application programs are implemented through orchestration, which can effectively solve the problems of uneven computing power between regions and low resource utilization. Through computing power perception, computing power quantification and computing power orchestration, diversified computing powers are integrated and unified, and standardized computing power services are provided to users, which provides strong support for the aggregation and efficient utilization of diversified computing power, and effectively improves the rationality of diversified computing power configuration, computing power service performance and practicality. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 It is a structural schematic diagram of the diversified computing power configuration system provided by the present invention.

[0019] Figure 2 It is a structural schematic diagram of the perception layer provided by the present invention.

[0020] Figure 3 It is a schematic diagram of the process of orchestration deployment of the orchestration layer provided by the present invention.

[0021] Figure 4 It is a flowchart of the diversified computing power configuration method provided by the present invention.

[0022] Figure 5 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] In the era of digital economy, the scale of computing power continues to grow, but it also faces problems such as uneven distribution of computing power, diversified computing power types, and imbalance between supply and demand, which makes it difficult to improve computing power utilization. Therefore, innovative computing network services for diversified computing power grid connection are the future trend. This innovative technology system and service model can give full play to the advantages of computing networks, widely gather computing power from multiple parties, and promote universal computing power and efficient services.

[0025] In the related technology, diversified computing power configuration is unified through the integration of diversified computing power. Specifically, the performance of each computing power device is uniformly modeled through a data-driven method to obtain a unified performance model; the performance data reflected by the hardware device layer is mapped to the performance events recorded in the kernel state layer of the operating system; the performance event layering is implemented in the kernel state layer of the operating system to obtain a platform-related event layer and a platform-independent event layer; the performance events of the platform-independent event layer are aggregated to the user state layer of the operating system; when the computing power device undergoes iterative changes, dynamic changes are achieved by adjusting the mapping relationship between the performance data reflected by the hardware device layer and the performance events of the platform-related event layer.

[0026] However, this computing power configuration method only uses a data-driven approach to transform the performance behavior description and classification of heterogeneous computing power devices into performance data collection and fusion problems. However, in the process of evaluating the overall computing power, there is a lack of process design for computing power quantification and a lack of method design for computing power test evaluation. It is difficult to accurately evaluate and compare the performance of heterogeneous computing power devices. After the aggregation of diversified computing power, how various applications run in diversified computing power environments is unclear. There is a lack of method design to ensure the stable operation of upper-level applications, which makes it difficult to ensure the rationality of computing power configuration.

[0027] In this regard, the present embodiment provides a diversified computing power configuration system and method, which collects computing power resource information of computing nodes using different instruction sets (different architectures) through resource perception capabilities; then uses a computing power quantification design method to evaluate the equivalence relationship between the computing power of different computing nodes, and implements computing power balanced scheduling and computing power elastic scaling based on the computing power evaluation results. Finally, standardized orchestration and deployment of application programs are implemented through orchestration, which makes up for the shortcomings of the prior art that lacks computing power quantification evaluation design and lacks design for how to orchestrate and deploy upper-level applications, solves the problems of uneven computing power and diverse computing power technology routes, and integrates and unifies diversified computing power through computing power perception, computing power quantification and computing power orchestration, and provides standardized computing power services to users, providing strong support for the aggregation and efficient utilization of diversified computing power, and effectively improving the rationality of diversified computing power configuration and computing power service performance.

[0028] Figure 1 Schematic diagram of the structure of the diversified computing power configuration system provided by the present invention; the system can be applied to various diversified computing power application scenarios, such as computing power grid connection, one cloud and multiple core scenarios, etc., and has good versatility. Figure 1 As shown, the diversified computing power configuration system 100 includes a resource layer 110 , a perception layer 120 , a quantization layer 130 and an orchestration layer 140 .

[0029] The resource layer 110 includes a plurality of computing nodes with different architectures.

[0030] Optionally, the resource layer 110 is located at the bottom layer and is used to provide a variety of bottom layer infrastructure, including computing nodes of multiple different architectures, to provide computing power support for upper layer applications. The computing nodes of multiple different architectures here include but are not limited to computing nodes of x86 architecture, computing nodes of ARM architecture, and computing nodes of RISC-V architecture, which are not specifically limited in this embodiment.

[0031] The perception layer 120 is used to perceive and obtain the resource information of each of the computing nodes, and according to the resource information and the image architecture label in the resource encapsulation creation request, obtain at least one target computing node from the multiple computing nodes, and encapsulate and establish a mapping matrix between the architecture information and function information of each of the target computing nodes, and parse and obtain the action information and architecture feature information corresponding to the resource encapsulation creation request.

[0032] The perception layer 120 is located between the resource layer and the quantization layer, and is the basis for the entire computing lifecycle management. It is a key part of the diversified computing power to achieve node scheduling and adaptive display of interface functions, and is the basis for supporting the operability of the program and realizing resource encapsulation lifecycle management. The perception layer has functions such as resource collection, resource interception, resource encapsulation and resource analysis.

[0033] It should be noted that the perception layer may use a processor to uniformly implement functions such as resource collection, resource interception, resource encapsulation and resource analysis, or may configure different functional modules for different functions to implement corresponding functions. This embodiment does not specifically limit this.

[0034] Optionally, the perception layer can perform resource collection on the resource layer to perceive and obtain resource information of each computing node in the resource layer in real time. The so-called resource information includes CPU architecture information and hardware feature information, etc.

[0035] After collecting and acquiring the resource information of each computing node, the perception layer may identify the image architecture tag in the resource package creation request, and perform interception and filtering based on the matching result between the resource information of each computing node and the image architecture tag in the resource package creation request to obtain at least one target computing node. The image architecture tag refers to metadata associated with the resource package, which is used to describe the hardware architecture used by the package.

[0036] The perception layer can also use the intercepted and filtered structural information for resource encapsulation to establish a mapping matrix (or functional matrix) between the architectural information and functional information that is encapsulated and dynamically expandable. The mapping matrix encapsulates the mapping relationship between the architectural information and functional information of each target computing node.

[0037] The perception layer can also parse and obtain the action information and architecture feature information corresponding to the resource encapsulation creation request to provide basic data for subsequent computing power evaluation.

[0038] The quantization layer 130 is used to perform computing power evaluation on each of the target computing nodes according to the mapping matrix, the action information and the architecture feature information, obtain multiple computing power evaluation results corresponding to each of the target computing nodes, and perform computing power balancing scheduling and computing power resource allocation for each of the target computing nodes according to the multiple computing power evaluation results.

[0039] Among them, the quantization layer 130 is located between the perception layer and the orchestration layer, and is a key method for characterizing the equivalent relationship of computing power.

[0040] Optionally, due to the different computing capabilities of computing nodes of different architectures, even if the same application uses resource encapsulation of the same specifications, such as the same CPU core, memory, etc., the performance of the same application running in a heterogeneous environment is different. To this end, this embodiment configures a quantization layer in the diversified computing power configuration system to characterize the computing power equivalence relationship and solve the problem of performance differences currently faced by the diversified computing power grid connection in the related technology, that is, the computing nodes of the ARM, x86 and RISC-V architectures of multiple manufacturers are different in instruction sets, core numbers, production processes, etc. The specific implementation steps are as follows: The quantization layer can evaluate the computing power of each target computing node based on the mapping matrix encapsulated by the perception layer, as well as the action information and architectural feature information obtained by the perception layer analysis, to obtain multiple computing power evaluation results corresponding to each target computing node. The computing power evaluation here can be quantitative calculations at multiple different levels for each target computing node, such as testing and evaluating the computing power of each target computing node from the server system resource level, testing and evaluating the resource encapsulation performance of the middleware from the platform level, and using real business load tests to evaluate the computing power of resource encapsulation from the application level.

[0041] After obtaining multiple computing power evaluation results, the quantization layer can use balanced scheduling technology to achieve balanced scheduling of computing power for each target computing node, or it can use elastic scaling technology to configure computing power resources for each target computing node.

[0042] For balanced computing power scheduling, the quantization layer can select a computing power evaluation result from multiple computing power evaluation results for equivalent computing power evaluation, or combine all computing power evaluation results for equivalent computing power evaluation, so as to perform balanced computing power scheduling for each target computing node based on the equivalent computing power evaluation, so as to ensure that all computing nodes can run in the optimal state, avoid overloading of some computing nodes and idleness of other computing nodes, and ensure the reasonable allocation of resources among the nodes.

[0043] For computing power resource configuration, the quantization layer can select one computing power evaluation result or multiple computing power evaluation results, or all computing power evaluation results from multiple computing power evaluation results, and dynamically adjust the resource allocation of each target computing node, such as increasing or decreasing memory, adjusting network bandwidth, etc., to support elastic scaling of resources for business peaks and troughs, and ensure that the service quality of the business scales linearly with the increase or decrease of resources.

[0044] The orchestration layer 140 is used to orchestrate and configure the application corresponding to the computing task to be executed according to the computing power balancing scheduling results and computing power resource configuration results of each target computing node.

[0045] Among them, the orchestration layer 140 is located at the top layer and is the key to ensuring the operation of the application. It can be based on the computing power balance scheduling results and computing power resource configuration results of each target computing node, and can orchestrate and configure the application corresponding to the computing power task to be executed to meet the business needs of different computing power tasks to be executed; the orchestration configuration here includes but is not limited to cross-architecture compilation and operation of the application, standard containerized packaging, and lightweight standard deployment.

[0046] The system provided in this embodiment collects computing power resource information of computing nodes using different architectures through resource perception capabilities; then uses computing power quantification to evaluate the equivalence relationship between the computing power of different computing nodes, and implements computing power balanced scheduling and computing power elastic scaling based on the computing power evaluation results. Finally, standardized orchestration and deployment of application programs are implemented through orchestration, which can effectively solve problems such as uneven computing power between regions and low resource utilization. Through computing power perception, computing power quantification and computing power orchestration, diversified computing power is integrated and unified, and standardized computing power services are provided to users, providing strong support for the aggregation and efficient utilization of diversified computing power, and effectively improving the rationality of diversified computing power configuration, computing power service performance and practicality.

[0047] In some embodiments, the perception layer includes a collector, an interceptor, and a scheduler; The collector is used to sense and obtain resource information of each computing node; The interceptor is used to identify the image architecture tag in the resource encapsulation creation request, match the image architecture tag with the architecture information in the resource information of each computing node, and filter out at least one target computing node matching the image architecture tag from the plurality of computing nodes according to the matching result, and encapsulate and establish a mapping matrix between the architecture information and the function information of each target computing node; The scheduler is used to parse and obtain the action information and architecture feature information corresponding to the resource package creation request, and display the action information and the architecture feature information.

[0048] Figure 2 A schematic diagram of the structure of the perception layer provided by the present invention; Figure 2 As shown, the perception layer includes collector, interceptor and scheduler; Among them, the collector communicates with the collector configured on each computing node in the resource layer and has a resource collection function. Specifically, it communicates with the collector configured on each computing node to collect the resource information of each computing node in the resource layer from the collector configured on each computing node in real time, reports the resource information of each computing node, and establishes a host list containing architectural characteristics based on the resource information.

[0049] The interceptor has resource interception function and resource encapsulation function. Specifically, it can identify the image architecture label in the resource encapsulation creation request through the cascade architecture perception filter device, match the image architecture label with the architecture information in the resource information of each computing node, and filter out at least one target computing node matching the image architecture label from the plurality of computing nodes. In addition, the intercepted and filtered result information is used to establish a mapping matrix between the architecture information and function information of each target computing node encapsulated into a dynamically scalable state.

[0050] The scheduler has the function of resource parsing. Specifically, it parses and obtains the action information and architecture feature information corresponding to the resource encapsulation creation request, and displays the parsing results as feedback, so as to realize the automatic identification and dynamic expansion of the functional differences of different architectures, shield the underlying implementation differences, and provide a unified resource management view.

[0051] The system provided in this embodiment configures collectors, interceptors and schedulers at the perception layer to work together to achieve accurate perception and acquisition of computing node resource information, efficient and accurate resource matching and filtering, and automated resource scheduling and information display, thereby significantly improving resource utilization and computing efficiency. At the same time, it ensures that applications can be quickly and accurately deployed and run in a variety of computing environments, thereby providing strong support for the aggregation and efficient utilization of diversified computing power, and effectively improving the rationality of diversified computing power configuration and computing power service performance.

[0052] In some embodiments, the quantization layer is further used to: According to the mapping matrix, the action information and the architecture feature information, a multi-level computing power evaluation is performed on each of the target computing nodes to obtain a plurality of computing power evaluation results corresponding to each of the target computing nodes; the plurality of computing power evaluation results include a specification computing power evaluation result, an effective computing power evaluation result and a business computing power evaluation result, the specification computing power evaluation result is an evaluation result of a whole machine specification performance benchmark test on each of the target computing nodes, the effective computing power evaluation result is an evaluation result of a middleware resource encapsulation performance test on each of the target computing nodes, and the business computing power evaluation result is an evaluation result of a business load performance test on each of the target computing nodes; According to the task type and task requirement corresponding to the computing task to be executed, obtaining a first computing power evaluation result from the multiple computing power evaluation results; According to the first computing power evaluation result, perform equivalent computing power evaluation on each of the target computing nodes to obtain a second computing power evaluation result corresponding to each of the target computing nodes; Determining a scheduling priority level of each of the target computing nodes according to the second computing power evaluation result; According to the scheduling priority level, computing power balancing scheduling is performed on each of the target computing nodes.

[0053] Optionally, the quantization layer realizes the computing power balanced scheduling capability through computing power evaluation, weight adjustment and load scheduling. Specifically, based on the mapping matrix, that is, the mapping relationship between node architecture and function, and the action information and architecture feature information obtained by parsing, a multi-level computing power evaluation is performed for each target computing node to obtain multiple computing power evaluation results.

[0054] The multi-level computing power assessment here can be divided into standard computing power, effective computing power and business computing power.

[0055] Among them, for the specification computing power evaluation, it tests and evaluates the computing power from the server system resource level; specifically for the devices at the server system resource level, such as processors, memory, disks, and networks, the quantification layer uses the Standard Performance Evaluation Corporation (SPEC) CPU, Unix Benchmark Suite (UnixBench) and stress (Stress) testing tools to conduct quantitative evaluation through the system specification performance benchmark test to obtain the specification computing power evaluation results.

[0056] For the effective computing power evaluation, it tests and evaluates the resource encapsulation performance of the middleware from the platform level; specifically for middleware such as Redis, MySQL, Kafka, it uses testing tools such as the Redis benchmark tool (Redis-Benchmark) or the MySQL stress testing tool (Mysqlslap) to conduct a joint quantitative evaluation of indicators such as the concurrency performance, response delay, and throughput of the middleware to obtain the effective computing power evaluation results.

[0057] For business computing power assessment, it uses real business load tests to evaluate resource encapsulation computing capabilities from the application level; specifically for business applications such as customer relationship management (CRM), e-commerce, AI reasoning or training, the quantification layer uses open source performance testing tools such as Jmeter or LoadRunner testing tools to conduct quantitative assessments on indicators such as business transaction processing capability (TPS), response latency, and iteration time to obtain business computing power assessment results.

[0058] After obtaining multiple computing power assessment results, the first computing power assessment result that meets the task type and task requirements can be associated among the multiple computing power assessment results based on the task type and task requirements corresponding to the computing power task to be executed. For example, for highly computing-intensive tasks, the standard computing power assessment results may be used; while for businesses that rely on specific middleware, effective computing power assessment results are more critical.

[0059] Next, in order to compare the computing power between different nodes more fairly, the quantization layer will perform an equivalent computing power evaluation, that is, convert the first evaluation results of different computing node dimensions into a unified second computing power evaluation result, so as to ensure the objectivity and accuracy of the scheduling decision.

[0060] Finally, from the resource level, when selecting computing nodes for resource packaging, the quantization layer can evaluate the computing power of each target computing node based on the second computing power evaluation results of each target computing node, so as to use the balancing strategy to load schedule each target computing node and solve the constrained optimization problem with the goal of maximizing resource utilization; the specific quantization layer can be to evaluate the computing power of each target computing node based on the second computing power evaluation results, so as to assign scheduling priority to each target computing node based on computing power, and perform computing power balancing scheduling accordingly, so as to ensure that when resources are allocated, the computing nodes that are more suitable for the current task can be called first to perform tasks, while ensuring the balanced utilization of overall resources to avoid resource waste and overload. The balanced scheduling algorithm here acts after the cascade filter, and can select computing nodes with relatively small loads, that is, computing nodes with relatively strong computing power from the multiple filtered target computing nodes for priority load scheduling.

[0061] The system provided in this embodiment ensures the comprehensiveness and accuracy of the evaluation results through a multi-level computing power evaluation mechanism, providing a scientific basis for task scheduling. The introduction of equivalent computing power evaluation further enhances the fairness and flexibility of scheduling decisions. Ultimately, the computing power balance scheduling strategy based on scheduling priority not only improves the execution efficiency and success rate of tasks, but also optimizes the overall resource utilization, reduces operating costs, and maximizes the use of computing power resources, thereby improving the rationality of diversified computing power configuration and computing power service performance.

[0062] In some embodiments, the quantization layer is further used to: In a case where the first computing power evaluation result is the standard computing power evaluation result, obtaining the standard computing power coefficient of each resource in each of the target computing nodes according to the standard computing power evaluation result; Combined with the specification computing power coefficient, as well as the maximum and minimum normalized values ​​and weight coefficients corresponding to the allocatable computing power of each resource in each target computing node, an equivalent computing power evaluation is performed to obtain a second computing power evaluation result of each target computing node.

[0063] Optionally, the quantization layer can evaluate the specification computing power coefficients of multiple types of resources in each target computing node based on the specification computing power evaluation results; then combine the specification computing power coefficients, the maximum and minimum normalized values ​​corresponding to the allocable computing power of each resource, the fairness of the main resources and other numerical methods to perform equivalent computing power evaluation to calculate the second computing power evaluation results (or available computing power) of each target computing node.

[0064] The second computing power evaluation result of each target computing node is calculated based on the following formula: ; ; ; in, Compute the target node The second computing power evaluation result, which is the target computing node The sum of the computing power weight scores of r types of resources, such as the sum of the computing power weight scores of resources including CPU, memory, hard disk, etc. Compute the target node The computing power weight score of resource i; Compute the target node The maximum and minimum positive normalized values ​​corresponding to the allocable computing power of resource i; Compute the target node The weight coefficient of resource i can be adaptively adjusted according to the intensive type of load such as CPU, memory or IO to reflect the importance of each resource. Compute the target node The specification computing power coefficient of resource i; for example, the specification computing power quantization relationship between the CPU of the ARM computing node and the x86 computing node is 1:2, then the corresponding specification computing power coefficients are 1 and 2 respectively, so that when the number of allocable CPU cores is the same, the x86 node is scheduled with a higher priority, thereby realizing balanced scheduling based on computing power quantification in the scenario of diversified social computing power grid connection. Compute the target node The allocable computing power of resource i; and are the maximum and minimum values ​​of resource i respectively.

[0065] In the system provided in this embodiment, the quantization layer can more accurately evaluate the actual computing power of each target computing node and provide a scientific basis for resource allocation for the system, which helps to optimize the execution efficiency of computing tasks, reduce resource waste, and improve the performance and stability of computing power configuration.

[0066] In some embodiments, the quantization layer is further used to: According to the historical records of the computing task to be executed, a load trend model of the computing task to be executed is obtained; the load trend model is used to characterize the relationship between the application load, service quality and resource capacity corresponding to the computing task to be executed; Determine an elastic scaling mode corresponding to each of the target computing nodes according to the load trend model and the plurality of computing power evaluation results; According to the elastic scaling mode, computing power resources are configured for each of the target computing nodes.

[0067] Optionally, the quantization layer can also support elastic scaling for business peaks and troughs through data collection, intelligent analysis, and resource scheduling computing power elastic scaling capabilities, so as to achieve precise planning, rapid scheduling, and computing power equivalence of resource encapsulation, and ensure that application services are scaled correctly, quickly, and accurately, thereby ensuring that the system can run stably during business peaks while avoiding wasting resources during trough periods.

[0068] Optionally, in terms of resource planning, the quantization layer collects the historical probability distribution characteristics of the application load of the computing task to be executed within a specific period to obtain the historical data time series, and establishes the load trend model of the computing task to be executed based on the historical data time series, so as to characterize the load portrait and capacity portrait of the relationship between the application load, service quality and resource capacity corresponding to the computing task to be executed, thereby predicting the current load change trend through the load trend model, and determining the appropriate elastic scaling mode for each target computing node based on the current load change trend and multiple computing power evaluation results, and then configuring computing power resources for each target computing node based on the elastic scaling mode to ensure the reasonable allocation of resources. The elastic scaling mode here includes but is not limited to increasing / decreasing computing resources, adjusting resource allocation ratios, and migrating services.

[0069] In terms of fast scheduling, the quantization layer uses architecture awareness and balanced scheduling technology to quickly schedule resources to the best nodes according to the determined elastic scaling mode, such as selecting nodes with sufficient idle resources and low network latency, and pulling up corresponding service instances to ensure timely response of application services.

[0070] When elastic scaling triggers cross-architecture switching of resource encapsulation, the quantization layer characterizes the computing power of different architectures based on computing power quantization technology, and jointly calculates the equivalent relationship of resource encapsulation based on effective computing power and business computing power to ensure that the service quality of the business can be linearly scaled when resources increase or decrease. For example, if the load is expected to increase significantly, some services can be automatically migrated from a low-computing power architecture to a high-computing power architecture to ensure that service performance is not affected.

[0071] In some embodiments, the orchestration layer is further used to: According to the computing power balancing scheduling result and the computing power resource configuration result, obtaining at least one node to be configured in at least one of the target computing nodes; According to the operating environment and operating tools of each node to be configured, the source code of the application corresponding to the computing task to be executed is cross-compiled across nodes to obtain the binary code corresponding to each node to be configured; Containerizing and packaging the binary code corresponding to each of the nodes to be configured and the dependency information of each of the binary codes; According to the packaging format of the containerized package, the binary code corresponding to each of the nodes to be configured is deployed in a standardized manner.

[0072] Figure 3 It is a flow chart of the orchestration layer deployment provided by the present invention; the design of the orchestration layer for the diversified social computing power configuration system includes three aspects: first, realizing the cross-architecture compilation and operation of the application, second, building a standard containerized package, and finally realizing lightweight deployment through cloud resource orchestration management, such as Figure 3As shown in the figure, the steps to achieve standardized application deployment through cross-architecture compilation and containerized packaging capabilities include: Node selection: After the computing power balance scheduling is completed, the orchestration layer first selects at least one node to be configured from multiple target computing nodes based on the scheduling results and computing power resource configuration results. These nodes are selected as the environment to perform specific computing power tasks.

[0073] Cross-architecture compilation: The orchestration layer uses the operating environment and operating tools of each node to be configured to cross-compile the source code of the application corresponding to the computing task to be executed across nodes to generate executable programs for other architectures. Here, cross-compilation realizes the cross-platform binary code generation of the program through architecture-independent source code. Instruction set translation technology is used to solve the problem of cross-architecture porting of applications, which is specifically implemented through software-level binary translation and chip-level binary translation.

[0074] Containerized packaging: cross-platform languages ​​solve cross-architecture problems of applications; cross-compilation solves cross-architecture compilation problems. Containerized packaging focuses on solving dynamic library dependency problems at runtime. The specific implementation process is to use standardized container methods to containerize and encapsulate the binary code corresponding to each node to be configured and the dependency information of each binary code, as the basic resource encapsulation for implementing cross-architecture deployment and switching of applications. On the one hand, it ensures its own operability, and on the other hand, for modern complex applications, it takes into account the runtime it depends on.

[0075] Standardized deployment: Based on the standardized packaging format of containers, the binary code corresponding to each node to be configured is deployed in a standardized manner to provide a consistent operating environment, so that the applications corresponding to the computing tasks to be executed deployed on different nodes can run in the same deployment environment, so as to integrate the diverse computing power into one, effectively ensure the reliability and standardized operation of computing power tasks on different nodes, improve the standardization of diversified computing power services, and thus improve the service performance of diversified computing power.

[0076] The diversified computing power configuration method provided by the present invention is described below. The diversified computing power configuration method described below and the diversified computing power configuration system described above can be referenced to each other.

[0077] Figure 4 is a flow chart of the diversified computing power configuration method provided by the present invention; Figure 4 As shown, the method includes step 410, step 420 and step 430.

[0078] Step 410, based on the perception layer, perceive and obtain the resource information of each computing node, and according to the resource information and the image architecture label in the resource encapsulation creation request, obtain at least one target computing node from the multiple computing nodes, and encapsulate and establish a mapping matrix between the architecture information and function information of each target computing node, and parse and obtain the action information and architecture feature information corresponding to the resource encapsulation creation request.

[0079] Optionally, the perception layer can perform resource collection on the resource layer to perceive and obtain resource information of each computing node in the resource layer in real time. The so-called resource information includes CPU architecture information and hardware feature information, etc.

[0080] After collecting and acquiring the resource information of each computing node, the perception layer may identify the image architecture tag in the resource package creation request, and perform interception and filtering based on the matching result between the resource information of each computing node and the image architecture tag in the resource package creation request to obtain at least one target computing node. The image architecture tag refers to metadata associated with the resource package, which is used to describe the hardware architecture used by the package.

[0081] The perception layer can also use the intercepted and filtered structural information for resource encapsulation to establish a mapping matrix between the architecture information and functional information that is encapsulated into a dynamically expandable state, and the mapping matrix encapsulates the mapping relationship between the architecture information and functional information of each target computing node.

[0082] The perception layer can also parse and obtain the action information and architecture feature information corresponding to the resource encapsulation creation request to provide basic data for subsequent computing power evaluation.

[0083] Step 420, based on the quantization layer, according to the mapping matrix, the action information and the architecture feature information, the computing power of each target computing node is evaluated to obtain multiple computing power evaluation results corresponding to each target computing node, and based on the multiple computing power evaluation results, computing power balancing scheduling and computing power resource allocation are performed on each target computing node.

[0084] Optionally, the quantization layer can evaluate the computing power of each target computing node based on the mapping matrix encapsulated by the perception layer, as well as the action information and architectural feature information obtained by the perception layer analysis, to obtain multiple computing power evaluation results corresponding to each target computing node. The computing power evaluation here can be quantitative calculations at multiple different levels for each target computing node, such as testing and evaluating the computing power of each target computing node from the server system resource level, testing and evaluating the resource encapsulation performance of the middleware from the platform level, and using real business load tests to evaluate the computing power of resource encapsulation from the application level.

[0085] After obtaining multiple computing power evaluation results, the quantization layer can use balanced scheduling technology to achieve balanced scheduling of computing power for each target computing node, or it can use elastic scaling technology to configure computing power resources for each target computing node.

[0086] For balanced computing power scheduling, the quantization layer can select a computing power evaluation result from multiple computing power evaluation results for equivalent computing power evaluation, or combine all computing power evaluation results for equivalent computing power evaluation, so as to perform balanced computing power scheduling for each target computing node based on the equivalent computing power evaluation, so as to ensure that all computing nodes can run in the optimal state, avoid overloading of some computing nodes and idleness of other computing nodes, and ensure the reasonable allocation of resources among the nodes.

[0087] For computing power resource configuration, the quantization layer can select one computing power evaluation result or multiple computing power evaluation results, or all computing power evaluation results from multiple computing power evaluation results, and dynamically adjust the resource allocation of each target computing node, such as increasing or decreasing memory, adjusting network bandwidth, etc., to support elastic scaling of resources for business peaks and troughs, and ensure that the service quality of the business scales linearly with the increase or decrease of resources.

[0088] Step 430, based on the orchestration layer, according to the computing power balancing scheduling results and computing power resource configuration results of each of the target computing nodes, the application corresponding to the computing power task to be executed is orchestrated and configured.

[0089] Optionally, the orchestration layer is located at the top layer and is the key to ensuring the operation of the application. It can be based on the computing power balance scheduling results and computing power resource configuration results of each target computing node, and can orchestrate and configure the application corresponding to the computing power task to be executed to meet the business needs of different computing power tasks to be executed; the orchestration configuration here includes but is not limited to cross-architecture compilation and operation of the application, standard containerized packaging, and multiple items of lightweight standard deployment.

[0090] The method provided in this embodiment collects computing power resource information of computing nodes using different architectures through resource perception capabilities; then uses computing power quantification to evaluate the equivalence relationship between the computing power of different computing nodes, and implements computing power balanced scheduling and computing power elastic scaling based on the computing power evaluation results. Finally, standardized orchestration and deployment of application programs are implemented through orchestration, which can effectively solve problems such as uneven computing power between regions and low resource utilization. Through computing power perception, computing power quantification and computing power orchestration, diversified computing power is integrated and unified, and standardized computing power services are provided to users, which provides strong support for the aggregation and efficient utilization of diversified computing power, and effectively improves the rationality of diversified computing power configuration, computing power service performance and practicality.

[0091] The method provided by the present invention is executed based on the above-mentioned system embodiments. Please refer to the above-mentioned embodiments for the specific structure and detailed content, which will not be repeated here.

[0092] Figure 5 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 5 As shown, the electronic device may include: a processor (processor) 510, a communication interface (Communications Interface) 520, a memory (memory) 530 and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other through the communication bus 540. The processor 510 can call the logic instructions in the memory 530 to execute a diversified computing power configuration method, which includes: based on the perception layer, sensing and obtaining the resource information of each computing node, and according to the resource information and the image architecture label in the resource encapsulation creation request, obtaining at least one target computing node from the multiple computing nodes, and encapsulating and establishing a mapping matrix between the architecture information and the function information of each target computing node, and parsing and obtaining the action information and architecture feature information corresponding to the resource encapsulation creation request; based on the quantization layer, according to the mapping matrix, the action information and the architecture feature information, performing computing power evaluation on each target computing node, obtaining multiple computing power evaluation results corresponding to each target computing node, and performing computing power balanced scheduling and computing power resource configuration on each target computing node according to the multiple computing power evaluation results; based on the orchestration layer, according to the computing power balanced scheduling results and computing power resource configuration results of each target computing node, orchestrating and configuring the application corresponding to the computing power task to be executed.

[0093] In addition, the logic instructions in the above-mentioned memory 530 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0094] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the diversified computing power configuration methods provided by the above methods, the method including: based on the perception layer, sensing and obtaining resource information of each computing node, and according to the resource information and the image architecture label in the resource encapsulation creation request, obtaining at least one target computing node from multiple computing nodes, and encapsulating and establishing a mapping matrix between the architecture information and the function information of each target computing node, and parsing and obtaining the action information and architecture feature information corresponding to the resource encapsulation creation request; based on the quantization layer, according to the mapping matrix, the action information and the architecture feature information, performing computing power evaluation on each target computing node, obtaining multiple computing power evaluation results corresponding to each target computing node, and performing computing power balanced scheduling and computing power resource configuration on each target computing node according to the multiple computing power evaluation results; based on the orchestration layer, according to the computing power balanced scheduling results and computing power resource configuration results of each target computing node, orchestrating and configuring the application corresponding to the computing power task to be executed.

[0095] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented by a processor to execute the diversified computing power configuration method provided by the above-mentioned methods when executed, the method comprising: based on the perception layer, perceiving and obtaining resource information of each computing node, and according to the resource information and the image architecture label in the resource encapsulation creation request, obtaining at least one target computing node from multiple computing nodes, and encapsulating and establishing a mapping matrix between the architecture information and the function information of each target computing node, and parsing and obtaining the action information and the architecture feature information corresponding to the resource encapsulation creation request; based on the quantization layer, according to the mapping matrix, the action information and the architecture feature information, performing computing power evaluation on each target computing node, obtaining multiple computing power evaluation results corresponding to each target computing node, and performing computing power balanced scheduling and computing power resource configuration on each target computing node according to the multiple computing power evaluation results; based on the orchestration layer, according to the computing power balanced scheduling result and computing power resource configuration result of each target computing node, performing orchestration and configuration on the application corresponding to the computing power task to be executed.

[0096] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0097] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0098] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A diversified computing power configuration system, It is characterized in that include: Resource layer, perception layer, quantization layer and orchestration layer; The resource layer includes a plurality of computing nodes of different architectures; The perception layer is used to perceive and obtain resource information of each of the computing nodes, and obtain at least one target computing node from the multiple computing nodes according to the resource information and the image architecture tag in the resource encapsulation creation request, and encapsulate and establish a mapping matrix between the architecture information and function information of each of the target computing nodes, and parse and obtain action information and architecture feature information corresponding to the resource encapsulation creation request; The quantization layer is used to perform computing power evaluation on each of the target computing nodes according to the mapping matrix, the action information and the architecture feature information, obtain multiple computing power evaluation results corresponding to each of the target computing nodes, and perform computing power balancing scheduling and computing power resource allocation on each of the target computing nodes according to the multiple computing power evaluation results; The orchestration layer is used to orchestrate and configure the application corresponding to the computing task to be executed according to the computing power balancing scheduling results and computing power resource configuration results of each target computing node.

2. The diversified computing power configuration system according to claim 1, It is characterized in that The perception layer includes a collector, an interceptor and a scheduler; The collector is used to sense and obtain resource information of each computing node; The interceptor is used to identify the image architecture tag in the resource encapsulation creation request, match the image architecture tag with the architecture information in the resource information of each computing node, and filter out at least one target computing node matching the image architecture tag from the plurality of computing nodes according to the matching result, and encapsulate and establish a mapping matrix between the architecture information and the function information of each target computing node; The scheduler is used to parse and obtain the action information and architecture feature information corresponding to the resource package creation request, and display the action information and the architecture feature information.

3. The diversified computing power configuration system according to claim 1, It is characterized in that The quantization layer is further used for: According to the mapping matrix, the action information and the architecture feature information, a multi-level computing power evaluation is performed on each of the target computing nodes to obtain a plurality of computing power evaluation results corresponding to each of the target computing nodes; The multiple computing power evaluation results include a specification computing power evaluation result, an effective computing power evaluation result, and a business computing power evaluation result. The specification computing power evaluation result is an evaluation result of a whole-machine specification performance benchmark test on each of the target computing nodes. The effective computing power evaluation result is an evaluation result of a middleware resource encapsulation performance test on each of the target computing nodes. The business computing power evaluation result is an evaluation result of a business load performance test on each of the target computing nodes. According to the task type and task requirement corresponding to the computing task to be executed, obtaining a first computing power evaluation result from the multiple computing power evaluation results; According to the first computing power evaluation result, perform equivalent computing power evaluation on each of the target computing nodes to obtain a second computing power evaluation result corresponding to each of the target computing nodes; Determining a scheduling priority level of each of the target computing nodes according to the second computing power evaluation result; According to the scheduling priority level, computing power balancing scheduling is performed on each of the target computing nodes.

4. The diversified computing power configuration system according to claim 3, It is characterized in that The quantization layer is further used for: In a case where the first computing power evaluation result is the standard computing power evaluation result, obtaining the standard computing power coefficient of each resource in each of the target computing nodes according to the standard computing power evaluation result; Combined with the specification computing power coefficient, as well as the maximum and minimum normalized values ​​and weight coefficients corresponding to the allocatable computing power of each resource in each target computing node, an equivalent computing power evaluation is performed to obtain a second computing power evaluation result of each target computing node.

5. The diversified computing power configuration system according to claim 3, It is characterized in that The quantization layer is further used for: According to the historical records of the computing task to be executed, a load trend model of the computing task to be executed is obtained; the load trend model is used to characterize the relationship between the application load, service quality and resource capacity corresponding to the computing task to be executed; Determine an elastic scaling mode corresponding to each of the target computing nodes according to the load trend model and the plurality of computing power evaluation results; According to the elastic scaling mode, computing power resources are configured for each of the target computing nodes.

6. The diversified computing power configuration system according to any one of claims 1 to 5, It is characterized in that The orchestration layer is also used to: According to the computing power balancing scheduling result and the computing power resource configuration result, obtaining at least one node to be configured in at least one of the target computing nodes; According to the operating environment and operating tools of each node to be configured, the source code of the application corresponding to the computing task to be executed is cross-compiled across nodes to obtain the binary code corresponding to each node to be configured; Containerizing and packaging the binary code corresponding to each of the nodes to be configured and the dependency information of each of the binary codes; According to the packaging format of the containerized package, the binary code corresponding to each of the nodes to be configured is deployed in a standardized manner.

7. A diversified computing power configuration method, It is characterized in that The method is applied to the diversified computing power configuration system according to any one of claims 1 to 6, and the method comprises: Based on the perception layer, the resource information of each computing node is perceived and acquired, and according to the resource information and the image architecture label in the resource encapsulation creation request, at least one target computing node is acquired from the multiple computing nodes, and a mapping matrix between the architecture information and the function information of each target computing node is encapsulated and established, and the action information and the architecture feature information corresponding to the resource encapsulation creation request are parsed and acquired; Based on the quantization layer, according to the mapping matrix, the action information and the architecture feature information, a computing power evaluation is performed on each of the target computing nodes to obtain a plurality of computing power evaluation results corresponding to each of the target computing nodes, and computing power balancing scheduling and computing power resource allocation are performed on each of the target computing nodes according to the plurality of computing power evaluation results; Based on the orchestration layer, according to the computing power balancing scheduling results and computing power resource configuration results of each of the target computing nodes, the application programs corresponding to the computing power tasks to be executed are orchestrated and configured.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, It is characterized in that When the processor executes the program, the diversified computing power configuration method as described in claim 7 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by the processor, the diversified computing power configuration method as claimed in claim 7 is implemented.

10. A computer program product comprising a computer program, It is characterized in that When the computer program is executed by the processor, the diversified computing power configuration method as claimed in claim 7 is implemented.

Images