CN112612603A - Cloud configuration method and system applicable to multi-frame micro-service application of financial business - Google Patents
Cloud configuration method and system applicable to multi-frame micro-service application of financial business Download PDFInfo
- Publication number
- CN112612603A CN112612603A CN202011464293.0A CN202011464293A CN112612603A CN 112612603 A CN112612603 A CN 112612603A CN 202011464293 A CN202011464293 A CN 202011464293A CN 112612603 A CN112612603 A CN 112612603A
- Authority
- CN
- China
- Prior art keywords
- task
- configuration
- model
- task type
- type
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000005457 optimization Methods 0.000 claims abstract description 15
- 238000013526 transfer learning Methods 0.000 claims abstract description 11
- 238000012360 testing method Methods 0.000 claims abstract description 9
- 238000012545 processing Methods 0.000 claims abstract description 7
- 238000004364 calculation method Methods 0.000 claims abstract description 6
- 238000007637 random forest analysis Methods 0.000 claims description 5
- 239000013598 vector Substances 0.000 claims description 5
- 238000004458 analytical method Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 claims description 3
- 238000013459 approach Methods 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000013508 migration Methods 0.000 description 2
- 230000005012 migration Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 239000000565 sealant Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a cloud configuration method and a system for multi-frame micro-service application suitable for financial business, which comprises the following steps: s1, receiving a task processing request, and calculating the similarity between the task in the request and a known task; s2, judging whether the task type in the request belongs to a known task type, a similar task type or an unknown task type according to the similarity calculation result; s31, for the known task types, generating a new task model by using the matched known initial model and performing online optimization; s32, for similar task types, loading the initial model of the most similar task by the online optimizer, and using transfer learning to guide the establishment of a new task model of a new task; s33, for the unknown task type, constructing a new task model based on benchmark test; and S4, discarding the configuration with the predicted execution time longer than the optimal predicted execution time of the current optimal configuration by using the new task model, constructing a configuration candidate set, and selecting the optimal cloud configuration predicted by the model.
Description
Technical Field
The invention relates to the technical field of software, in particular to a cloud configuration method and a cloud configuration system for multi-frame micro-service application suitable for financial business.
Background
Microservices are currently the primary implementation form of internet applications, and many are typically deployed in cloud computing infrastructures. A wide variety of microservice applications execute in a computing ecosystem comprised of multiple frames connected together in a graph-like structure to form a computing cluster. A common way to deploy these computing clusters is to acquire resources in a cloud environment and acquire resources on demand when tasks need to be performed repeatedly. When performing tasks in the cloud, it is important to ensure that the correct resources are allocated, in order to meet the strict quality of service and cost-effective needs of task completion time. Thus, deploying computing clusters in the cloud requires solving the cloud configuration problem, determining how many instances to use, and which type of instance to use. Selecting the correct configuration is critical because when such a decision is missed, the task cannot meet its expiration date or execution cost multiplied.
The existing method of selecting a Cloud configuration (CherryPick: adaptive approximation the Best Cloud Configurations for Big Data analysis. in the Proceedings of the 14th Usenix Symposium on networked Systems Design and Implementation,2017.) does not consider the presence of multiple frames in a compute cluster, but focuses on configuring a single frame. In a multi-frame computing cluster-oriented environment, existing approaches have limitations. The first method (mix: A sealant alternative for selecting closed entities. in the Proceedings of 11th IEEE International Conference on Cloud Computing, pages 409-416. IEEE,2018.) optimizes each frame individually, however, coupling between different frames and unreasonable pre-allocation of resources makes it difficult to improve performance as a whole. In the second method (Arrow: Low-level estimated Bayesian Optimization for filing the Best Cloud VM. in the Proceedings of 38th IEEE International Conference on Distributed Computing Systems, pages 660-670. IEEE,2018.) the whole Computing cluster is regarded as a single large system, and the configuration of all frames is modified at the same time, which supports joint Optimization, but the size of the configured search space is doubled with the increase of the number of frames. Exploring larger spaces naturally takes longer, resulting in expensive delays before users deploy their clusters in production.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides a cloud configuration method and a cloud configuration system for multi-frame micro-service application suitable for financial business, which select optimized cloud configuration for the multi-frame micro-service application deployed in a cloud computing environment, so that the performance of executing the service application is improved, namely the request processing time is reduced, and the utilization rate of cloud resources is improved. The technical scheme is as follows:
in one aspect, the invention provides a cloud configuration method for multi-frame microservice application suitable for financial business, comprising the following steps:
s1, receiving a task processing request, and calculating the similarity between the task in the request and a known task;
s2, according to the similarity calculation result, judging whether the task type in the request belongs to a known task type, a similar task type or an unknown task type, if the task type belongs to the known task type, executing S31 and S4, if the task type belongs to the similar task type, executing S32 and S4, and if the task type belongs to the unknown task type, executing S33 and S4;
s31, for the known task types, generating a new task model by using the matched known initial model and performing online optimization;
s32, for similar task types, loading the initial model of the most similar task by the online optimizer, and using transfer learning to guide the establishment of a new task model of a new task;
s33, for the unknown task type, constructing a new task model based on benchmark test;
s4, utilizing the new task model, discarding the configuration with the predicted execution time longer than the best predicted execution time of the current best configuration, constructing a configuration candidate set, selecting one configuration for the next execution, and selecting the best cloud configuration predicted by the model from the rest candidates.
Further, step S31 includes the following three stages:
an initialization stage: the optimizer first performs an analysis task using a pre-specified configuration to obtain an initial performance sample;
and a resource increasing stage: analyzing the running sequence, wherein the number of the instances distributed to each frame is increased according to the utilization rate of the monitored CPU and the monitored memory; if the sum of the average utilization rate of the CPU and the memory utilization rate exceeds a threshold value, the number of the instances is doubled, otherwise, the number of the instances is increased by a constant; when any one of the two conditions occurs, the optimizer checks whether a plurality of valid configurations are found, if so, the configuration with the best execution time is returned to terminate, otherwise, the optimizer enters a fine-grained resource adjustment stage;
and a resource adjusting stage: if the CPU and the memory utilization rate are smaller than the threshold value, the number of the instances is halved; if the CPU and the memory utilization rate are larger than the threshold value, the number of the instances is doubled; in other cases, the number of instances is subtracted by a constant; if the utilization index exceeds a preset high critical value, the number of the examples is halved, and if the utilization index is lower than a preset low critical value, the number of the examples is doubled.
Further, step S33 further includes:
selecting a previously executed configuration as a centroid, taking as a neighbor a valid configuration with an execution time within a current minimum execution time, extending the selection to the previously executed configuration that satisfies the execution time constraint using a scoring function, and representing an optimal selection according to its execution cost.
Further, step S33 further includes:
a higher score is assigned to configurations that satisfy the cost constraint than to configurations that do not satisfy the cost constraint, the assigned score being proportional to the number of exceeded constraints, such that in configurations that satisfy the time constraint, a neighborhood of valid configurations with lower execution times is preferentially searched.
Further, for unknown task types, constructing the configuration candidate set in step S4 includes:
forming a set of untested candidate configurations based on a set of centroids that vary the drawn configuration from the neighborhood of each centroid and by the instance type of the current best configuration; for each centroid, each configuration within a distance d from the centroid is selected, and a distance metric is defined based on differences in CPU and memory capacity.
Further, step S32 further includes:
when a new task arrives, executing on the initial configuration, and scoring according to the similarity with the previously executed task; and if the similarity of the task and the existing task reaches a preset similarity threshold, transmitting the task to an online optimizer, loading a performance model of the most similar task by the online optimizer, and using transfer learning to guide the establishment of the performance model of the new task, wherein the model is used as an initial state of the online optimizer to search for the proper configuration of the new task.
Further, the similarity between the two tasks in step S1 is that the corresponding two configuration vectors are smaller than the number of CPUs and the total memory amount is greater than the vector distance between the two tasks.
In another aspect, the present invention provides a cloud configuration system for multi-frame microservice application for financial transactions, comprising the following modules:
the task similarity calculation module is used for receiving a processing request of a task and calculating the similarity between the task in the request and a known task;
the known task type configuration module is used for responding to the fact that the task type in the request belongs to the known task type, using the matched known initial model and performing online optimization to generate a new task model;
the similar task type configuration module is used for responding to the fact that the task type in the request belongs to the similar task type, loading an initial model of the most similar task by the online optimizer, and guiding and establishing a new task model of the new task by using transfer learning;
the unknown task type configuration module is used for responding to the fact that the task type in the request belongs to the unknown task type, and then a new task model is built based on benchmark test;
and the optimal cloud configuration module is used for utilizing the new task model, discarding the configuration with the predicted execution time longer than the optimal predicted execution time of the current optimal configuration, constructing a configuration candidate set, selecting one configuration for the next execution, and selecting the optimal cloud configuration predicted by the model from the rest candidates.
Further, the system builds a performance model based on an online random forest approach to fine-tune the configuration, the model being continuously updated with each successive run of repeated tasks.
The technical scheme provided by the invention has the following beneficial effects:
a. by analyzing the similarity between tasks and reusing the knowledge in the previous task by using transfer learning, the benchmark test operation can be improved, and the time for finding better configuration in the initial stage is reduced;
b. the configuration is finely adjusted by constructing a performance model based on an online random forest method, and the configuration can be continuously updated to gradually search the optimized configuration, so that the problem of low efficiency of a multi-frame large search space is solved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a flowchart of a cloud configuration method of a multi-framework microservice application suitable for financial transactions according to an embodiment of the present invention.
Detailed Description
In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, apparatus, article, or device that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or device.
The invention relates to a cloud configuration method for multi-frame microservice application combined with off-line optimization. In the benchmark phase, an index-based optimizer is used to quickly determine the initial configuration. At this stage, the similarity between tasks is analyzed, and the knowledge in previous tasks is reused using migration learning. And then, constructing a performance model based on an online random forest method to fine-tune configuration, wherein the model is continuously updated along with each continuous operation of repeated tasks to gradually improve configuration so as to solve the performance limitation of performing rapid reference search in a large search space, and thus the efficiency of searching and optimizing cloud configuration is improved.
In an embodiment of the present invention, there is provided a cloud configuration method for a multi-framework microservice application suitable for financial transactions, as shown in fig. 1, the cloud configuration method including the steps of:
s1, receiving a task processing request, and calculating the similarity between the task in the request and a known task;
s2, according to the similarity calculation result, judging whether the task type in the request belongs to a known task type, a similar task type or an unknown task type, if the task type belongs to the known task type, executing S31 and S4, if the task type belongs to the similar task type, executing S32 and S4, and if the task type belongs to the unknown task type, executing S33 and S4;
s31, for the known task types, generating a new task model by using the matched known initial model and performing online optimization;
s32, for similar task types, loading the initial model of the most similar task by the online optimizer, and using transfer learning to guide the establishment of a new task model of a new task;
s33, for the unknown task type, constructing a new task model based on benchmark test;
s4, utilizing the new task model, discarding the configuration with the predicted execution time longer than the best predicted execution time of the current best configuration, constructing a configuration candidate set, selecting one configuration for the next execution, and selecting the best cloud configuration predicted by the model from the rest candidates.
The principle of the invention is as follows: a sufficiently good configuration is quickly found in the benchmarking phase and incrementally optimized as the task is repeated during execution. Specifically, in the benchmarking phase, an index-based optimizer is used to quickly determine the initial configuration. At this stage, the similarity between tasks is analyzed, and the knowledge in previous tasks is reused using migration learning. Then, to address the performance limitations of performing fast baseline searches in a large search space, a performance model is built based on an online random forest approach to fine-tune the configuration, which is continuously updated with each successive run of repeated tasks to gradually improve the configuration.
In one embodiment of the invention, the implementation steps of the cloud configuration method of the multi-frame microservice application combined with the off-line optimization are as follows:
the method comprises the following steps of firstly, calculating the similarity of a known task and judging whether the task type is known, similar or unknown. The similarity is the vector distance between the two configurations < number of CPUs, total amount of memory > current task and known task.
Each time a task request arrives, the following three cases are divided: the task type is known to be the same as the task previously performed; similar task types are new tasks not encountered before but similar to tasks performed before; the unknown task type is a new task that has not been encountered before and is dissimilar to any task that was performed before.
And secondly, generating a task model by using the initial model and performing online optimization on the known task type. The step is divided into three stages of initialization, resource increase and resource adjustment.
An initialization stage: the optimizer first performs an analysis task using a pre-specified configuration to obtain an initial performance sample.
And a resource increasing stage: the order of execution is analyzed and the number of instances assigned to each frame is increased based on the monitored CPU and memory utilization. If the sum of the average CPU utilization and the memory utilization exceeds a threshold, the number of instances is doubled. Otherwise, a lower sum of CPU and memory utilization indicates that resource doubling would be over-allocated. In this case, the number of instances increases by a constant each time the runtime is analyzed. Since the search space limits the number of instances of a particular size, a larger size of instances will be reached if the number of instances reaches the limit. If the execution time of the current configuration is slow compared to the execution time of the previous configuration, or the previous configuration is valid, but the current configuration is not valid. When either of these two conditions occurs, the optimizer checks whether it has found a number of valid configurations. If this is the case, it will be terminated by returning the configuration with the best execution time. Otherwise, the optimizer will go to a fine-grained resource adjustment phase to find a better configuration.
And a resource adjusting stage: the optimizer finds an efficient configuration by reducing the overall execution cost by reducing resources. During an analysis run, the framework reduces its resource allocation when any CPU and memory utilization does not exceed a certain threshold. This operation terminates when the current configuration is invalid and the previous configuration is valid. The number of examples was adjusted as follows: when the CPU and the memory utilization rate are smaller than a threshold value, the number of the instances is halved; when the CPU and the memory utilization rate are larger than a threshold value, the number of the instances is doubled; in other cases, the number of instances is reduced by a constant. When the utilization index is too high or too low, an adjustment is made to halve or double the number of instances. Otherwise, fine-grained adjustments are enabled, incremented or decremented to the number of instances to avoid over-or under-allocated resources.
Thirdly, constructing an initial model for the unknown task type based on a benchmark test; and for similar task types, constructing an initial model based on the transfer learning, and performing online optimization to generate a task model. Selecting a previously executed configuration as a centroid, taking as a neighbor a valid configuration with an execution time within e of the current lowest execution time to provide a better possible configuration, extending the selection to the previously executed configuration that satisfies the execution time constraint using a scoring function, and expressing an optimization selection according to its execution cost. The higher score is assigned to configurations that satisfy the cost constraint and the lower score is assigned to configurations that do not satisfy the cost constraint, in proportion to the number of exceeded constraints. It is ensured that in configurations satisfying the time constraint, the neighborhood of valid configurations with lower execution time is preferentially searched. Candidate set selection: a set of untested candidate configurations is formed based on a set of centroids that vary the drawn configuration from the neighborhood of each centroid and by the instance type of the current best configuration. For each centroid, each configuration within a distance d from the centroid is selected, and a distance metric is defined based on differences in CPU and memory capacity.
And fourthly, generating cloud configuration according to the generated task model. Using the candidate set, one of the configurations is selected for the next execution. The candidate set is filtered, the configuration with the prediction execution time excessively higher than the optimal prediction execution time of the current optimal configuration is discarded, and the optimal configuration predicted by the model is selected from the remaining candidates.
The cloud configuration method for the multi-frame microservice application combined with online optimization in the embodiment of the invention has the flow divided into two stages:
the first stage, based on the fast optimized benchmarking stage, runs benchmarking using an offline optimizer to determine a good enough initial configuration.
The second phase, in the production optimization phase, uses an online optimizer, each run is actually performed, for incrementally updating the performance model to improve the configuration of the next iteration task.
Whenever a task request arrives, the following three cases are distinguished: knowing the task type, the task being the same as the previously performed task; a similar task type, a new task not encountered before, but similar to a previously executed task; similar to the task type, new tasks have not been encountered before, and are not similar to any tasks performed before.
For known task types: and the requested running task is circulated again and directly transmitted to the online optimizer. And the online optimizer selects a new cloud configuration for testing by using the performance model of the task, and updates the performance model after the operation is finished.
For similar task types: when a new task arrives, it is executed on the initial configuration and scored for similarity to the previously executed task. If the task has sufficient similarity to the existing task, the task is passed to an online optimizer. The online optimizer loads the performance model of the most similar task and uses transfer learning to guide the building of the performance model of the new task. The model serves as an initial state for the online optimizer to search for the appropriate configuration for the new task.
For unknown task types: when the new task does not satisfy the similarity filtering rule, the task request is passed to an offline optimizer. The optimizer finds a good enough configuration to satisfy the user constraints, then uses that configuration to initiate execution, and the subsequently submitted task is optimized using an online optimizer.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.
Claims (9)
1. A cloud configuration method of multi-frame micro-service application suitable for financial business is characterized by comprising the following steps:
s1, receiving a task processing request, and calculating the similarity between the task in the request and a known task;
s2, according to the similarity calculation result, judging whether the task type in the request belongs to a known task type, a similar task type or an unknown task type, if the task type belongs to the known task type, executing S31 and S4, if the task type belongs to the similar task type, executing S32 and S4, and if the task type belongs to the unknown task type, executing S33 and S4;
s31, for the known task types, generating a new task model by using the matched known initial model and performing online optimization;
s32, for similar task types, loading the initial model of the most similar task by the online optimizer, and using transfer learning to guide the establishment of a new task model of a new task;
s33, for the unknown task type, constructing a new task model based on benchmark test;
s4, utilizing the new task model, discarding the configuration with the predicted execution time longer than the best predicted execution time of the current best configuration, constructing a configuration candidate set, selecting one configuration for the next execution, and selecting the best cloud configuration predicted by the model from the rest candidates.
2. The cloud configuration method for multi-framework microservice application for financial transaction as claimed in claim 1, wherein step S31 comprises the following three phases:
an initialization stage: the optimizer first performs an analysis task using a pre-specified configuration to obtain an initial performance sample;
and a resource increasing stage: analyzing the running sequence, wherein the number of the instances distributed to each frame is increased according to the utilization rate of the monitored CPU and the monitored memory; if the sum of the average utilization rate of the CPU and the memory utilization rate exceeds a threshold value, the number of the instances is doubled, otherwise, the number of the instances is increased by a constant; when any one of the two conditions occurs, the optimizer checks whether a plurality of valid configurations are found, if so, the configuration with the best execution time is returned to terminate, otherwise, the optimizer enters a fine-grained resource adjustment stage;
and a resource adjusting stage: if the CPU and the memory utilization rate are smaller than the threshold value, the number of the instances is halved; if the CPU and the memory utilization rate are larger than the threshold value, the number of the instances is doubled; in other cases, the number of instances is subtracted by a constant; if the utilization index exceeds a preset high critical value, the number of the examples is halved, and if the utilization index is lower than a preset low critical value, the number of the examples is doubled.
3. The cloud configuration method for multi-framework microservice application for financial transaction as claimed in claim 1, wherein the step S33 further comprises:
selecting a previously executed configuration as a centroid, taking as a neighbor a valid configuration with an execution time within a current minimum execution time, extending the selection to the previously executed configuration that satisfies the execution time constraint using a scoring function, and representing an optimal selection according to its execution cost.
4. The cloud configuration method for multi-framework microservice application for financial transaction as claimed in claim 3, wherein the step S33 further comprises:
a higher score is assigned to configurations that satisfy the cost constraint than to configurations that do not satisfy the cost constraint, the assigned score being proportional to the number of exceeded constraints, such that in configurations that satisfy the time constraint, a neighborhood of valid configurations with lower execution times is preferentially searched.
5. The cloud configuration method for multi-framework microservice application of financial transaction as claimed in claim 3, wherein the step of building a configuration candidate set in S4 for unknown task types comprises:
forming a set of untested candidate configurations based on a set of centroids that vary the drawn configuration from the neighborhood of each centroid and by the instance type of the current best configuration; for each centroid, each configuration within a distance d from the centroid is selected, and a distance metric is defined based on differences in CPU and memory capacity.
6. The cloud configuration method for multi-framework microservice application for financial transaction as claimed in claim 1, wherein the step S32 further comprises:
when a new task arrives, executing on the initial configuration, and scoring according to the similarity with the previously executed task; and if the similarity of the task and the existing task reaches a preset similarity threshold, transmitting the task to an online optimizer, loading a performance model of the most similar task by the online optimizer, and using transfer learning to guide the establishment of the performance model of the new task, wherein the model is used as an initial state of the online optimizer to search for the proper configuration of the new task.
7. The cloud configuration method for multi-frame microservice application of financial transaction as claimed in claim 1, wherein the similarity between two tasks in step S1 is such that the corresponding two configuration vectors are smaller than the number of CPUs and the total memory is larger than the vector distance between two tasks.
8. A cloud configuration system for multi-frame microservice applications for financial transactions, comprising the following modules:
the task similarity calculation module is used for receiving a processing request of a task and calculating the similarity between the task in the request and a known task;
the known task type configuration module is used for responding to the fact that the task type in the request belongs to the known task type, using the matched known initial model and performing online optimization to generate a new task model;
the similar task type configuration module is used for responding to the fact that the task type in the request belongs to the similar task type, loading an initial model of the most similar task by the online optimizer, and guiding and establishing a new task model of the new task by using transfer learning;
the unknown task type configuration module is used for responding to the fact that the task type in the request belongs to the unknown task type, and then a new task model is built based on benchmark test;
and the optimal cloud configuration module is used for utilizing the new task model, discarding the configuration with the predicted execution time longer than the optimal predicted execution time of the current optimal configuration, constructing a configuration candidate set, selecting one configuration for the next execution, and selecting the optimal cloud configuration predicted by the model from the rest candidates.
9. The cloud configuration system for multi-framework microservice applications for financial transactions according to claim 8, characterised in that it is based on an online random forest method to build a performance model to fine tune the configuration, which model is continuously updated with each successive run of repeated tasks.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011464293.0A CN112612603B (en) | 2020-12-14 | 2020-12-14 | Cloud configuration method and system for multi-frame micro-service application suitable for financial business |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011464293.0A CN112612603B (en) | 2020-12-14 | 2020-12-14 | Cloud configuration method and system for multi-frame micro-service application suitable for financial business |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112612603A true CN112612603A (en) | 2021-04-06 |
| CN112612603B CN112612603B (en) | 2024-05-10 |
Family
ID=75233611
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011464293.0A Active CN112612603B (en) | 2020-12-14 | 2020-12-14 | Cloud configuration method and system for multi-frame micro-service application suitable for financial business |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112612603B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116776748A (en) * | 2023-08-18 | 2023-09-19 | 中国人民解放军国防科技大学 | Throat bolt type variable thrust engine throat bolt spray pipe configuration design knowledge migration optimization method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018015779A1 (en) * | 2016-07-20 | 2018-01-25 | Worldline | Multi-criteria adaptive scheduling for a market-oriented hybrid cloud infrastructure |
| CN110443364A (en) * | 2019-06-21 | 2019-11-12 | 深圳大学 | A kind of deep neural network multitask hyperparameter optimization method and device |
| CN110727506A (en) * | 2019-10-18 | 2020-01-24 | 北京航空航天大学 | SPARK parameter automatic tuning method based on cost model |
| CN110855578A (en) * | 2019-11-06 | 2020-02-28 | 西安电子科技大学 | Similarity-based cloud micro-service resource scheduling optimization method |
-
2020
- 2020-12-14 CN CN202011464293.0A patent/CN112612603B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018015779A1 (en) * | 2016-07-20 | 2018-01-25 | Worldline | Multi-criteria adaptive scheduling for a market-oriented hybrid cloud infrastructure |
| CN110443364A (en) * | 2019-06-21 | 2019-11-12 | 深圳大学 | A kind of deep neural network multitask hyperparameter optimization method and device |
| CN110727506A (en) * | 2019-10-18 | 2020-01-24 | 北京航空航天大学 | SPARK parameter automatic tuning method based on cost model |
| CN110855578A (en) * | 2019-11-06 | 2020-02-28 | 西安电子科技大学 | Similarity-based cloud micro-service resource scheduling optimization method |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116776748A (en) * | 2023-08-18 | 2023-09-19 | 中国人民解放军国防科技大学 | Throat bolt type variable thrust engine throat bolt spray pipe configuration design knowledge migration optimization method |
| CN116776748B (en) * | 2023-08-18 | 2023-11-03 | 中国人民解放军国防科技大学 | Throat bolt type variable thrust engine throat bolt spray pipe configuration design knowledge migration optimization method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112612603B (en) | 2024-05-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10031774B2 (en) | Scheduling multi-phase computing jobs | |
| US7082606B2 (en) | Dedicated heterogeneous node scheduling including backfill scheduling | |
| CN110096362B (en) | Multitask unloading method based on edge server cooperation | |
| US10437648B2 (en) | Guided load balancing of graph processing workloads on heterogeneous clusters | |
| CN112948079B (en) | Task scheduling method, device, equipment and computer storage medium | |
| US20080281986A1 (en) | Parameter adjusting device | |
| CN112703512A (en) | Quantum circuit design specific to application or algorithm | |
| CN113282409B (en) | Edge calculation task processing method and device and computer equipment | |
| CN112612603A (en) | Cloud configuration method and system applicable to multi-frame micro-service application of financial business | |
| Ahmed et al. | RALB‐HC: A resource‐aware load balancer for heterogeneous cluster | |
| Tian et al. | JMSNAS: Joint model split and neural architecture search for learning over mobile edge networks | |
| WO2016197706A1 (en) | Data migration method and device | |
| CN112130927B (en) | Reliability-enhanced mobile edge computing task unloading method | |
| Herrera et al. | A grid-oriented genetic algorithm | |
| CN114650321A (en) | Task scheduling method for edge computing and edge computing terminal | |
| CN112882805A (en) | Profit optimization scheduling method based on task resource constraint | |
| Tuba et al. | Web intelligence data clustering by bare bone fireworks algorithm combined with k-means | |
| CN115629883A (en) | Resource prediction method, resource prediction device, computer equipment and storage medium | |
| CN113010288B (en) | Scheduling method and device of cloud resources and computer storage medium | |
| US20180210762A1 (en) | Apparatus, method, and program medium for parallel-processing parameter determination | |
| Yang et al. | Resource reservation for graph-structured multimedia services in computing power network | |
| CN113065641A (en) | Neural network model training method and device, electronic equipment and storage medium | |
| US20220342711A1 (en) | Allocation of heterogeneous computational resource | |
| Jiang et al. | PVBTS: A novel task scheduling algorithm for heterogeneous computing platforms | |
| Niyom et al. | An energy-efficient process clustering assignment algorithm for distributed system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |