阅读说明：本技术 基于业务动态画像的资源调整方法和系统 (Resource adjustment method and system based on business dynamic portrait ) 是由 张继东 曹靖城 周帅 秦臻 于 2020-06-19 设计创作，主要内容包括：本申请涉及基于业务动态画像的资源调整方法和系统。该方法包括：基于预先收集基础数据建立业务画像标签库；获取业务基本信息,通过与标签库进行关联映射形成标准业务静态标签；采集业务运行时的指标数据,将所采集到的指标数据进行建模分析,并通过与标签库进行关联映射形成标准业务动态标签；结合业务静态标签和业务动态标签形成业务画像；基于业务画像制定业务资源动态调整方案。该系统包括业务画像标签库模块、业务静态标签模块、业务运行数据采集模块、业务动态标签模块、业务动态画像模块,和业务资源调整模块。该方法和系统可提升资源分配的合理性,提高综合利用率,提升资源调整的智能化水平。(The application relates to a resource adjusting method and system based on a business dynamic portrait. The method comprises the following steps: establishing a service portrait label library based on pre-collected basic data; acquiring basic service information, and performing association mapping with a tag library to form a standard service static tag; acquiring index data during service operation, performing modeling analysis on the acquired index data, and performing associated mapping on the index data and a tag library to form a standard service dynamic tag; forming a service portrait by combining the service static label and the service dynamic label; and formulating a dynamic business resource adjusting scheme based on the business portrait. The system comprises a service portrait label library module, a service static label module, a service operation data acquisition module, a service dynamic label module, a service dynamic portrait module and a service resource adjustment module. The method and the system can improve the rationality of resource allocation, improve the comprehensive utilization rate and improve the intelligent level of resource adjustment.)

1. A resource adjustment method is characterized by comprising the following steps:

establishing a service portrait label library based on pre-collected basic data, wherein the basic data comprises service basic information and service characteristics;

acquiring basic service information, and performing associated mapping with the tag library to form a standard service static tag;

acquiring index data during service operation, performing modeling analysis on the acquired index data, and performing associated mapping on the index data and the label library to form a standard service dynamic label;

combining the service static label and the service dynamic label to form a service image; and

and formulating a service resource dynamic adjustment scheme based on the service portrait.

2. The method of claim 1, wherein the step of building a library of business representation tags comprises:

collecting the basic service information to form a basic service information label;

acquiring resource configuration information to form a service resource configuration label;

forming service resource use level labels according to the use conditions of the hierarchical resources;

constructing a service resource use change trend classification label;

and obtaining a service portrait label library based on the service basic information label, the service resource configuration label, the service resource use level label and the service resource use change trend classification label.

3. The method of claim 2, wherein the step of forming a standard business dynamic label comprises:

preprocessing the time sequence data of the index data;

counting the preprocessed index data, and judging the variation trend;

identifying the state of the service according to the judged change trend;

analyzing the future resource utilization level of the service to form a characteristic label corresponding to the classification of the service resource utilization condition; and

and performing correlation mapping on the characteristic label and the label library to form a standard business dynamic label.

4. The method of claim 1, wherein the service basic information includes a service class, and a service belonging party, wherein the service class is classified into at least three classes.

5. The method of claim 1, wherein the steps of forming a standard business static label, forming a standard business dynamic label, and forming a business representation are repeated periodically, and the business representation is updated continuously to form a business dynamic representation.

6. The method of claim 3, wherein identifying the service status according to the determined trend of change comprises: and constructing a service trend characteristic classification model by using an SVM classification algorithm, and analyzing whether the service condition of the resources is ascending, descending or stable, thereby identifying whether the service is in a growing state, a stable state or an atrophied state.

7. The method of claim 3, wherein analyzing future resource usage levels for a service comprises a DeepAR algorithm modeling historical resource usage for the service and predicting resource usage levels for the service over a fixed period of time in the future, and forming a characteristic label corresponding to a ranking of service resource usage that is preset by the service resource usage level label.

8. The method of claim 1, further comprising: and searching the low-utilization-rate collapsed service according to the service dynamic representation, and sequencing according to the service grade to generate a service list to be recovered so as to reduce or recover service resources.

9. A resource adjustment system is characterized by comprising a business portrait label library module, a business static label module, a business operation data acquisition module, a business dynamic label module, a business dynamic portrait module and a business resource adjustment module,

the service dynamic portrait module integrates the service static label generated by the service static label module and the service dynamic label generated by the service dynamic label module to form a service dynamic portrait, and the service resource adjusting module makes a service resource dynamic adjusting strategy according to the service dynamic portrait.

10. The system of claim 9, wherein the service resource adjustment module further formulates a service resource dynamic adjustment strategy by constructing a service trend characteristic classification model using an SVM classification algorithm, and modeling service historical resource utilization using a deep ar algorithm and predicting a future development trend of a service to form a resource demand prediction model.

Technical Field

The invention relates to the field of Internet, in particular to a resource adjusting method and system based on a business dynamic portrait.

Background

With the continuous development of internet technology, more and more services are developed by using network service resources, which are commonly called as 'going to the cloud'.

In the actual use process of network resources of an internet service provider, especially in the stage of online and resource allocation of a new service, in order to guarantee the stability of the new service and good customer experience in a future peak period, the allocated resources are usually much higher than actual requirements, and if the resources cannot be adjusted in time, a large amount of resources are left unused and wasted, so that the use cost of the resources is high.

For the problem, the traditional solution is that the operation and maintenance personnel pre-collect monitoring indexes such as the access amount, the utilization rate, the operation load and the like of each service resource in a manual mode or an automatic script mode, and in combination with the operation and maintenance experience, preset a fixed adjustment threshold, and adjust the allocation of the resources when the collected monitoring indexes reach the preset fixed adjustment threshold.

The conventional method for fixing the threshold mainly has the following two problems:

first, the conventional method lacks objective dynamic quantitative indicators to identify the current stage of business development and the characteristics of resource demand. In fact, when any service has different development stages, such as growth stage, stationary stage and atrophy stage, the demands for resources in different stages are greatly different. And the fixed threshold is set based on the monitoring index collected at the initial stage of the new service starting. With the continuous development of the service, the resource requirements of the service may change greatly, and whether the dynamic development requirements of the service cannot be met by adjusting the resources is simply judged based on a fixed threshold preset at an initial stage, so that the dynamic resource requirements of the service need to be accurately matched.

Secondly, the traditional method for fixing the threshold mainly refers to a resource utilization rate index of a fixed time period in the past to allocate resources to the whole service, and due to the lack of intelligent data analysis and intelligent prediction means, the operation of a service system is possibly unstable after adjustment, so that negative effects are generated. For example, the average resource utilization rate of a certain service is not high in a fixed time period (for example, three months) that has passed, but actually an exponential increase trend appears in the last two weeks, but the traditional method simply adjusts the allocated resources to be low according to the low average resource rate of the certain time period in the past, and fails to make any prejudgment through the increase trend of the last two weeks, and as a result, the actual resource demand suddenly increases, which causes a bottleneck, and generates a poor user experience.

Therefore, a method for dynamically adjusting resources is needed, which can greatly improve the rationality of resource allocation, improve the comprehensive utilization rate, avoid idle waste of resources, and improve the intelligence level of resource adjustment while ensuring safe and stable operation of services.

Disclosure of Invention

This summary is provided to introduce a selection of concepts in a simplified form that are further described below in the detailed description. This summary is not intended to identify key features or essential features of the claimed subject matter; nor is it intended to be used as an aid in determining or limiting the scope of the claimed subject matter.

The invention aims to greatly improve the comprehensive utilization rate of resources and the intelligent level of resource adjustment while ensuring the safe and stable operation of the service. Based on an intelligent model algorithm, a service dynamic portrait is established, a resource dynamic adjustment strategy is set, the service state and the resource demand change trend can be intelligently identified, the service and the resources corresponding to each module of the service can be dynamically and accurately adjusted in advance, the service is ensured to operate safely and stably, and meanwhile the reasonability of resource allocation and the utilization rate of the resources are improved.

The network resource dynamic adjusting system comprises a service portrait label library module, a service static label module, a service operation data acquisition module, a service dynamic label module, a service dynamic portrait module, a service resource adjusting module and the like. Firstly, a service portrait label library is established based on basic data, a service static label is generated by collecting basic information such as service grades and service belongings, a dynamic label of index data analysis service such as access amount and resource utilization rate generated during service operation is collected, and then the service static label and the dynamic label are integrated to form a service dynamic portrait. Constructing a service trend characteristic classification model by using an SVM classification algorithm, and identifying the development state of a service; and modeling the service historical resource utilization rate by using a DeepAR algorithm and predicting the future development trend of the service to form a resource demand prediction model. And dynamically adjusting the services and the resources corresponding to the modules in advance by intelligently setting a resource adjustment strategy.

The invention discloses a resource adjusting method, which comprises the following steps: establishing a service portrait label library based on pre-collected basic data, wherein the basic data comprises service basic information and service characteristics; acquiring basic service information, and performing associated mapping with a tag library to form a standard static service tag, wherein the basic service information comprises service grades, service categories and service belonged parties, and the service grades are divided into at least three grades; acquiring index data during service operation, performing modeling analysis on the acquired index data, and performing associated mapping with a tag library to form a standard service dynamic tag; combining the business static label and the business dynamic label to form a business image; and formulating a business resource dynamic adjustment scheme based on the business portrait. Further comprising: and searching the low-utilization-rate collapsed service according to the service dynamic image, and sequencing according to the service level to generate a service list to be recovered so as to reduce or recover service resources. And regularly repeating the steps of forming a standard business static label, forming a standard business dynamic label and forming a business portrait, and continuously updating the business portrait to form a business dynamic portrait.

The step of establishing a service portrait label library comprises the following steps: acquiring basic service information to form a basic service information label; acquiring resource configuration information to form a service resource configuration label; forming service resource use level labels according to the use conditions of the hierarchical resources; constructing a service resource use change trend classification label; and obtaining a service portrait label library based on the service basic information label, the service resource configuration label, the service resource use level label and the service resource use change trend classification label.

Wherein the step of forming a standard business dynamic label comprises: preprocessing time sequence data of the index data; counting the preprocessed index data, and judging the variation trend; identifying the state of the service according to the judged change trend, comprising the following steps: constructing a service trend characteristic classification model by using an SVM classification algorithm, and analyzing whether the service condition of resources is ascending, descending or stable so as to identify whether the service is in a growing state, a stable state or an atrophied state; analyzing the future resource utilization level of the service to form a characteristic label corresponding to the grade of the service resource utilization condition, wherein the characteristic label comprises that a deep AR algorithm models the service historical resource utilization rate and predicts the resource utilization level of the service in a future fixed time period, and forms a characteristic label corresponding to the grade of the service resource utilization condition preset by the service resource utilization level label; and carrying out association mapping on the characteristic label and the label library to form a standard business dynamic label.

The invention discloses a resource adjusting system which comprises a business portrait label library module, a business static label module, a business operation data acquisition module, a business dynamic label module, a business dynamic portrait module and a business resource adjusting module.

The service dynamic portrait module integrates a service static label generated by the service static label module and a service dynamic label generated by the service dynamic label module to form a service dynamic portrait, and the service resource adjusting module makes a service resource dynamic adjusting strategy according to the service dynamic portrait. The business resource adjusting module also constructs a business trend characteristic classification model by utilizing an SVM classification algorithm, models the business historical resource utilization rate by utilizing a DeepAR algorithm and predicts the future development trend of the business to form a resource demand prediction model so as to formulate a business resource dynamic adjusting strategy.

Drawings

The present invention will now be described more fully hereinafter with reference to the accompanying drawings, in which specific embodiments of the invention are shown.

FIG. 1 is a flow chart of a method for dynamically adjusting resources according to the present invention;

FIG. 2 is a flowchart of sub-steps of step S1 in FIG. 1;

FIG. 3 is a flowchart of sub-steps of step S4 in FIG. 1;

fig. 4 is a schematic diagram of the service resource dynamic adjustment policy of step S6 in fig. 1;

FIG. 5 is a block diagram of a dynamic resource adjustment system of the present invention;

FIG. 6 is a schematic flow chart of resource allocation and recovery by using the dynamic resource adjustment system of the present invention.

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s).

Detailed Description

The present invention will now be described more fully hereinafter with reference to the accompanying drawings, in which specific embodiments of the invention are shown. Various advantages and benefits of the present invention will become apparent to those of ordinary skill in the art upon reading the following detailed description of the specific embodiments. It should be understood, however, that the present invention may be embodied in various forms and should not be limited to the embodiments set forth herein. The following embodiments are provided so that the invention may be more fully understood. Unless otherwise defined, technical or scientific terms used herein shall have the ordinary meaning as understood by those of skill in the art to which this application belongs.

The dynamic resource adjustment method of the present invention is described below with reference to fig. 1 to 4.

The resource dynamic adjustment method of the present invention as shown in fig. 1 comprises the following steps:

firstly, a service portrait label library is established, wherein the label library comprises information such as service basic information, hardware configuration information, use condition grading standards, change trend classification labels, use period classification labels and the like. The tag library has certain data specifications, i.e., standards for tags, such as resource usage levels that use words such as idle, extremely low, medium, high, extremely high, and insufficient.

As in step S1 in fig. 1: and establishing a business portrait label library based on basic data such as business basic information, business characteristics and the like collected in advance. It comprises the following sub-steps, see fig. 2, in which the sub-steps of step S1 are illustrated:

s1-1: collecting basic service information such as various different service levels, service categories and service belongings to form basic service information labels, wherein the service levels are divided into at least A, B, C levels from high to low (as can be understood by those skilled in the art, the division of the three levels is only an example, and more levels can be adopted for actual situations);

s1-2: acquiring hardware resource configuration information such as a CPU (Central processing Unit), a memory, a magnetic disk and the like to form a service resource configuration label;

s1-3: grading the resource use conditions, such as idle (0-0.2%), extremely low (0.2% -0.5%), low (0.5% -2%), low (2% -30%), medium (30% -60%), high (60% -70%), extremely high (70% -80%), insufficient (80% -100%), forming service resource use level labels (those skilled in the art can understand that the eight-level classification and the specific percentages thereof are only examples, and different numbers of level and percentage standards can be adopted for different services);

s1-4: constructing service resource use change trend classification labels, such as growth, stability, atrophy and the like, and constructing service resource use periodic classification labels, such as period, non-period and the like;

s1-5: and (5) arranging the information obtained from the S1-1 to S1-4 to obtain a service portrait label library.

Next, returning to fig. 1, at step S2: the basic information of the service, including but not limited to the inherent information of the resource such as the service affiliated party, the service level, the configuration condition of the hardware resource (CPU, memory, storage resource, etc.), is obtained, and is associated and mapped with the tag library, so that the tag library conforms to the data specification of the tag library, namely becomes a standard tag, namely a standard static tag of the service is established.

Subsequently, at step S3: collecting various index data, namely service load data, including but not limited to service access volume, service data packet, and hardware resource utilization rate data of each service sub-module (such as web page, middleware, database, etc.) during service operation in a near fixed time period, wherein the index data is used for establishing a service dynamic label in step S4.

And step S4, modeling and analyzing various index data such as the service load and the resource utilization rate data collected in the step S3, and establishing a standard service dynamic label. It comprises the following sub-steps, see fig. 3, in which the sub-steps of step S4 are illustrated:

s4-1: preprocessing collected time sequence data of various indexes such as service load data, resource utilization rate data and the like, such as smoothing, null filling and the like;

s4-2: counting the characteristics of the preprocessed index data, such as: statistical characteristics such as mean value, maximum value, median, minimum value and the like of the granularity at the time of nearly 7 days, 1 month, 3 months and the like; decomposing the trend items of each index by using a time sequence decomposition algorithm, and judging the change trend of the index;

s4-3: according to the output of the step S4-2, a service trend characteristic classification model is constructed by utilizing an SVM classification algorithm, and the resource use conditions (ascending, descending or stable) of modules such as a service web page, a middleware and a database are analyzed, so that whether the service is in an increasing state, a stable state or an atrophied state is identified;

s4-4: modeling the service historical resource utilization rate by using a DeepAR algorithm, predicting the resource utilization level of the service in a fixed time period (for example, 3 months) in the future, analyzing the future resource utilization level of the service, and forming a characteristic label corresponding to the grade of the service resource utilization condition preset in the step S1-3;

s4-5: and performing correlation mapping on the analysis result of the S4-3 and the characteristic label obtained in the S4-4 and the label library to enable the label library to conform to the data specification of the label library, namely to become a standard label, thereby forming a standard business dynamic label.

In step S5: combining the service static label and the dynamic label obtained in the previous step to form a service image, and periodically repeating the steps S2-S4 to continuously update the service image to form the service dynamic image.

Finally at step S6: based on the service dynamic representation, dynamically analyzing the service resource adjustment scheme, and making a service resource dynamic adjustment policy including recycling, decreasing allocation, and expanding capacity, see fig. 4, which shows an example of the detailed policy of step S6:

the traffic resource is reclaimed at step S6-1 in accordance with the following:

a-level service: the service portrait is matched with any one group of characteristics in the 'idle + stable' and 'idle + atrophy';

b-level service: the service portrait is matched with any group of characteristics of idle + stable ', ' idle + atrophy ', ' extremely low + stable ', ' extremely low + atrophy ';

c-level service: the service portrait is matched with any one group of characteristics of idle + stable ', ' idle + atrophy ', ' extremely low + stable ', ' extremely low + atrophy ', ' lower + stable ', ' lower + atrophy ';

the service resources are reallocated to the service module presenting the same characteristics as the service overall resource usage at step S6-2 in accordance with the following:

a-level service: the service portrait is matched with any one group of characteristics in the characteristics of extremely low and steady and extremely low and atrophy;

b-level service: when the service portrait is matched with any one group of characteristics of lower, stable and lower and atrophy;

c-level service: when the service portrait is matched with any one group of characteristics of low, steady and low and atrophy;

the service module capacity expansion service resource presenting the same characteristics as the service overall resource usage in step S6-3 is satisfied with the following condition:

a-level service: the service portrait is matched with any one group of characteristics in the terms of 'deficiency + growth', 'deficiency + steady', 'high + growth', 'high + steady', 'high + growth' and 'high + steady';

b-level service: the service portrait is matched with any one group of characteristics in the categories of 'deficiency + growth', 'deficiency + stability', 'high + growth' and 'high + stability';

c-level service: the service portrait is matched with any one group of characteristics in the conditions of deficiency, growth and deficiency and stability.

The dynamic resource adjustment system of the present invention is described below with reference to fig. 5.

The dynamic resource adjustment system 500 of the present invention includes: the system comprises a service portrait label library module 501, a service static label module 502, a service operation data acquisition module 503, a service dynamic label module 504, a service dynamic portrait module 505 and a service resource adjusting module 506.

The service portrait label library module 501 collects basic service information, service characteristics and the like to establish a service portrait label library, which includes attribute values related to a service static label and a service dynamic label. The static label refers to inherent attributes of the service, including a service affiliated party, a service level, service resource configuration conditions and the like; dynamic labels refer to dynamic characteristics of a business when running, such as: the service load situation, the service resource use trend, the maximum, the average, the minimum use rate and the like in different time periods such as the past three months, the past one month, the past one week, the current, the predicted future one week, the predicted future one month and the like.

The service static tag generation module 502 obtains service resource allocation and service basic attributes (such as service level, service party, etc.), and forms a standard static tag by performing associated mapping with the tag library established by the service image tag library module 501.

The service operation data acquisition module 503 acquires data such as access volume, service data packet, service and resource utilization rate of each module under the service during service operation in real time, and supplies the data to the service dynamic tag generation module 504.

The service dynamic label generation module 504 constructs a classification algorithm model based on the service running index data collected by the service running data collection module 503, analyzes the service resource usage level, the long-term change trend and other characteristics, and forms a standard dynamic label by performing associated mapping with the label library module 501 for the service portrait of the label library.

The service dynamic representation generation module 505 integrates the service static label generation module 504 and the service dynamic label generation module 502 to form a dynamic label, and forms a service dynamic representation.

The service resource adjustment module 506 forms a service dynamic image based on the service dynamic image generation module 505, and analyzes and formulates a service resource adjustment scheme.

The following describes a process of resource allocation and recovery using the dynamic resource adjustment system of the present invention with reference to fig. 6.

First, in step S601, a service with a long-term resource utilization rate lower than a certain percentage (e.g., 2%) and a reduced type is retrieved according to the service dynamic image.

Then, in step S602, a list of services to be recovered is generated according to the service level and sorted according to the low-usage collapsed services retrieved in step S601.

In step S603, it is determined whether the service to be recovered can perform targeted resource reallocation operations according to the service module; and in the case that the downgrading is judged to be impossible, the service resource is recycled in step S604, otherwise, the downgrading or other adjustment operation is performed in step S605.

Since the service resource recovery belongs to high-risk operations, for the sake of safety, the service resource recovery may generally add a step of manual intervention by the operation and maintenance personnel, that is, before step S604, the system pops up a reminder (shown by a dashed oval in the figure), notifies the operation and maintenance personnel that the condition for recovering the service resource has been met, waits for manual judgment, and recovers the service resource after the manual judgment is confirmed.

The invention analyzes and obtains the business static label and the business dynamic label based on the business basic information and various index data in operation, forms a business dynamic portrait and a change trend prediction model, can effectively improve the utilization rate of resources, solves the problem of idle waste in the resource using process and effectively optimizes the resource using cost.

The method sets a resource adjustment strategy based on the service dynamic portrait, dynamically adjusts the service and the resources corresponding to each module according to the identified service form in advance, accurately matches the dynamic requirements of the resources, and the adjustment strategy is more intelligent and accurate. The method can ensure the safety and stability of the service system while dynamically adjusting the resources, and realize the balance of resource adjustment and service safety.