KR101256918B1 - Method for enhancing scalability and availability of cloud services and the system thereby - Google Patents

Abstract

The method for improving the scalability and availability of the cloud service according to the present invention includes the following steps. In the first step, the management server receives the service quality and resource information of the mobile node and the resource information of the station node. In the second step, the management server determines whether the quality management process is performed using the collected information. In the third step, the management server selects at least one service quality management activity of a service migration process and a service replication process to be performed on a variable component among target services for which the quality management process is determined. In the fourth step, the service quality management activity selected by the management server is performed.
Through the service scalability management framework and method according to the present invention, all resources of the service system can be dynamically registered and managed, and the service system can be dynamically routed according to the load to maximize the scalability of the service system.

Description

Method for enhancing scalability and availability of cloud services and the system thereby}

The present invention relates to a method and system for improving the scalability and availability of a cloud service, and more particularly, to a method for improving quality by proactively responding to a cause of deterioration of a cloud service by integrating and managing resources of all nodes. It's about that system.

Mobile computing constitutes independent systems such as central processing unit (CPU) and memory, and is connected to other computer systems through a network with high portability and mobility. However, as a result of miniaturization for portability and mobility, performance is lower than that of general desktop computers or server computers.

Recently, in consideration of the performance limitations of mobile computing, researches on server-based mobile applications and service-based mobile applications that place some functions of mobile applications outside of mobile devices have been conducted.

These studies can be divided into the areas of measuring network and service quality, the technology of analyzing measured quality, and the technology of improving network and service quality, ie, service scalability, based on the negative results. have.

In service-oriented computing, the frequency of service user groups or service requests is not known in advance, so an unexpectedly large number of service requests can cause service server resource exhaustion and network congestion problems, thereby degrading the overall quality of the service. This problem occurs when service scalability is not high enough. Services with a low level of scalability lower user preferences. Therefore, in service-oriented computing, research on 'service scalability' for minimizing deterioration of service quality due to the increase of consumers and providing services to more consumers is required.

Scalability studies have been conducted in many fields, including networks, databases, and distributed computing. In the existing research, dynamic routing manages scalability by distributing load to the same components running on multiple nodes connected to the network, load balancing through replication and synchronization of a database, or by splitting tasks in parallel. However, since the service targets an unspecified number of consumers, it is difficult to predict the amount of load generated by the use of the service of the consumer, and there is a peak time in which the amount of load rapidly increases. Due to the characteristics of such services, the definition of service scalability, related metrics, and scalability management techniques must be newly defined. In other words, the scalability study required in each field provides the foundation for service scalability in service-oriented computing, but there is a lack of techniques for securing service level scalability by associating this.

The traditional approach to scalability is to add new hardware to increase the scalability of the system. In general, the scale-up method is to install additional hardware on one node to improve the physical processing capacity of the node, and the scale-out method is to add a new node to the system to target the system. It is to improve the processing power of. However, the hardware scalability improvement method needs to stop the system for hardware installation and operation, and the problem of decreasing the actual scalability improvement rate for additional resources as new hardware is added.

Wang's research suggests a load balancing model using agents in a service-based grid environment. The presented model consists of six components, each of which is a Global Scheduler (GS), Local Scheduler (LS), Balancer Agent (BA), Fault Tolerance Agent (FTA), History Information Database (HID), and Load Monitoring Agent ( LM). GS and LS serve to distribute the load, and GS covers the realm of LS. The BA is mounted on the GS and LS to perform load balancing, while the FTA performs fault tolerant. The HID records data about the load and the data is stored in GS. The LM monitors the load on the service instance. Along with the six components, load measurement and load balancing techniques are presented to improve performance and scalability. However, since the scalability of the service node is not considered, only limited nodes should be used. Accordingly, a solution is needed to respond to service requests that occur beyond expectations. Wang's work is described in Wang, J., Wu, Q., Zheng, D., and Jia, Y., "Agent based Load Balancing Model for Service based Grid Applications," In proceedings of Computational Intelligence and Security (CIS). 2006), pp 487-491, Guangzhou, China, Nov. 3-6, 2006. '.

Chawathe's work is building a service network to provide scalability and fault avoidance. The SNS (Scalable Network Services) architecture presented in this paper provides load balancing, node management and defect avoidance through a centralized manager. The SNS consists of a central manager who manages all nodes, a worker who performs the functions of a service, and a worker driver for connecting the central manager as a library included in the worker. The central manager collects data about the load that the node bears from the worker driver of the node, and plans and executes load balancing. The increase in load involves idle nodes from the central manager as workers. SNS understands the situation of all workers due to centralized management and promotes effective load balancing and scalability improvement in the configured network. However, centralized configurations are likely to cause bottlenecks. Since all requests from service users are concentrated and worker interactions are also passed through the central manager, there is a high possibility of delay in the service relay process. Chawathe's work is described in Chawathe, Y. and Brewer, EA, "System Support for Scalable and Fault Tolerant Internet Services," In proceedings of the IFIP International Conference on Distributed Systems Platforms and Open Distributed Processing, pp 71-88, London, UK, Sep. 1-1, 2009. '.

As such, many studies on scalability improvement and defect avoidance using the network have been conducted, but there are limitations in considering static node management and service characteristics.

As to solve the above problems,

An object of the present invention is to provide a dynamic management method and means for managing all nodes and services for overcoming the limitations of static node management, which is a problem of existing hardware techniques.

It is also an object of the present invention to provide a method and means for ensuring not only service scalability but also overall quality attributes provided by a service.

The method for improving the scalability and availability of the cloud service according to the present invention includes the following steps.

In the first step, the management server receives the service quality and resource information of the mobile node and the resource information of the station node.

In the second step, the management server determines whether the quality management process is performed using the collected information.

In the third step, the management server selects at least one service quality management activity of a service migration process and a service replication process to be performed on a variable component among target services for which the quality management process is determined.

In the fourth step, the service quality management activity selected by the management server is performed.

In addition, the selection of the service quality management activity may include selecting a destination station node and a target station node.

Also, in the third step, the service quality management activity may further include a method of updating a load balancing table.

In addition, performing the service duplication process may include the following steps.

In step 4a-1, the management server refers to the resource information transmitted from the station node, and selects the station node having the maximum available resource as the destination station node.

In step 4a-2, the management server transmits the information of the destination station node to the target station node whose quality of service is a problem.

In step 4a-3, the target station node transmits a variable component of the target service to the target station node.

In addition, performing the service migration process may include the following steps.

In step 4b-1, the target station node, which is a quality control target, transmits the variable component to the management server.

In step 4b-2, the target station node removes the shape of the service to which the transmitted component belongs.

In step 4b-3, the management server refers to the resource information transmitted from the target station node, and selects the station node having the maximum available resource as the destination station node.

In step 4b-4, the management server transmits the transmitted variable component to the destination station node.

Furthermore, the service may further include the following steps after the service migration or replication process. In the fifth step, the destination station node dynamically installs and arranges the transmitted variable component. In the sixth step, the destination station node transmits the result of performing the service migration or replication process to the management server.

Furthermore, in the fifth step, the variable component can be dynamically installed and arranged using a class loading technique.

Meanwhile, the method may further include a seventh step of measuring service scalability by comparing throughput before and after the service migration or replication process.

In addition, the variable component may include a service implementation and a service interface, which is a set of classes implementing a service function.

In addition, in the third step, when the physical distance between the target station node that needs management and the mobile node using the service provided by the station node is far, when the response time is long, when the traffic of the target station node increases, and when the target station node The service migration process can be selected when at least one of the resources is insufficient.

In addition, in the third step, the service replication process may be selected when the use of the service provided by the managed station node increases rapidly or the quality of service (QoS) is in a degraded state.

In a cloud service system having a plurality of mobile nodes and a station node providing a service and a management server, the management server according to the present invention includes the following configuration.

The monitoring means receives the service quality and resource information of the mobile node and the resource information of the station node to determine whether to perform the quality management process.

The quality management planning means selects at least one service quality management activity of a service migration process and a service replication process to be performed on the variable components among the target services for which the quality management process is determined.

The quality control activity executing means executes the quality control activity according to the target node, the type of service quality management activity, and the target node selected by the quality control plan establishment means.

In addition, the variable component may include a service implementation and a service interface, which is a set of classes implementing a service function.

On the other hand, in the cloud service system having a plurality of mobile nodes and a station node for providing a service and a management server, the station node according to the present invention includes the following configuration. The quality information transmitting means collects resource information of the station node and transmits it to the management server.

The quality control means receives the execution command of any one of the service migration process and the service replication process from the management server and the information about the target destination station node to perform the quality management activity.

In addition, the quality control unit may transmit a variable component of the quality control target service to the management server when receiving a service migration process execution command from the management server.

On the other hand, when the quality control means receives a service replication process execution command and destination station node information from the management server, it may transmit the variable component to the destination station node.

According to the present invention, the service scalability improvement method is proposed in terms of software rather than the conventional hardware method, thereby increasing resource utilization even in the same resource situation and consequently improving service scalability.

In addition, according to the present invention, by adding a new node dynamically during service operation and suggesting a management method for deploying and operating a service to a specific node, the limitation of static service addition, that is, the problem of stopping a service in order to add a new node, is a problem of the conventional technique. Can overcome.

In addition, according to the present invention, by effectively distributing the service request of the consumer to each node registered in the management server, the ratio of improving the actual scalability of the added resources is increased.

That is, through the service scalability management framework and method according to the present invention, all resources of the service system can be dynamically registered and managed, and the new service calls can be dynamically routed according to the load to maximize the scalability of the service system. have.

1 is a flow chart showing the overall flow of a quality of service (QoS) management process of the cloud service system.
2 is a block diagram illustrating a service quality management system of a cloud service and a mobile computing environment according to an exemplary embodiment of the present invention.
3 is a block diagram illustrating a mobile node agent according to one embodiment.
4 is a block diagram illustrating a station node agent according to an embodiment.
5 is a block diagram illustrating a management server according to an exemplary embodiment.
6 is a graph illustrating a classification state of service level quality and shape level quality according to an exemplary embodiment.
7 is a block diagram illustrating a conceptual service quality management activity according to an embodiment.
8 is a block diagram illustrating a specific service quality management activity according to an embodiment.
9 is a flowchart schematically illustrating an overall process of a method for improving scalability and availability of a cloud service according to an exemplary embodiment.
10 is a flowchart illustrating a service replication process according to an embodiment.
11 is a flowchart illustrating a service migration process according to an embodiment.

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. Unless otherwise defined or mentioned, terms indicating directions used in the present description are based on the states shown in the drawings. In addition, the same reference numbers indicate the same members throughout the embodiments.

Meanwhile, hereinafter, the service type refers to one service itself including a plurality of instances, and the shape refers to a plurality of service types and mobile apps (applications) to be managed entirely.

Figure 1 shows the overall flow of the quality of service (QoS) management process of the cloud service system. The service quality autonomous management process largely evaluates the quality of the configuration and determines whether there is a need for quality control. It analyzes the overall quality of the cloud service system, the quality of each service type, and the quality of service instances of each type, and establishes activities and execution plans to improve service quality. According to the established quality improvement plan, the corresponding node or management server performs quality improvement activities. Finally, according to the result of the quality improvement activity performed, the current configuration of the changed cloud service system is transmitted to the management server, and the management server performs quality measurement activity based on the structure of the changed cloud service system. In particular, the present invention has major features in the planning of quality improvement activities and the performance of quality improvement activities.

Referring to Figure 2 will be described the overall appearance of the service quality management system of the cloud service and mobile computing environment according to an embodiment of the present invention. 2 is a block diagram illustrating a service quality management system of a cloud service and a mobile computing environment according to an exemplary embodiment of the present invention.

The management server 100, the mobile node 200, and the station node 300 are provided on the network 10 provided with the cloud service. The mobile node 200 refers to a node provided with a cloud service, and includes a service app 220 for using the cloud service and a mobile node agent 210 for service quality management. The station node 300 has a role of providing a cloud service, and includes a service providing means 320 including a plurality of instances for providing a service, and a station node agent 310 for quality of service management. Meanwhile, the management server 100 includes a management manager 110 for managing the quality of the overall shape including the mobile node 200 and the station node 300 and a data storage unit 130 storing various quality data. do. Hereinafter, specific service quality management methods of cloud services and mobile computing environments will be described with the corresponding configurations.

A mobile node agent according to an embodiment of the present invention will be described with reference to FIG. 3. 3 is a block diagram illustrating a mobile node agent according to one embodiment.

The mobile node agent 210 is provided with measuring means 211 for measuring quality. As a measuring means 211 of the mobile node agent 210, a resource measuring means 2111 and a response time measuring means 2112 are provided. The resource measuring means 2111 measures a resource use state of the corresponding node device. Resource measuring means measures the CPU usage, memory usage, network bandwidth, battery level of the node. Response time measuring means 2112 measures the response time of the service.

On the other hand, the state of the dynamic mobile ecosystem (DME) means the quality of the system and a quality model is needed to measure it. Therefore, the quality of the software can be determined objectively and quantitatively and the monitoring quality model is determined using ISO 9126 [7], which is recognized as a standard for quality models.

 In an embodiment of the present invention, the efficiency, which is closely related to performance, is defined as a quality model, and the time consumption and resource utilization attributes are defined. In particular, response time (RT) was selected as an attribute of consumption time. Response time means the time required from the time of requesting the service required by the application to the completion of service execution and the transmission of the result to the application. Metrics of the respective attributes are shown in Table 1 below. Resource efficiency includes CPU usage, memory usage, battery usage, and network bandwidth.

TABLE 1

The quality information transmitting unit 212 periodically transmits the service quality information and the resource information measured from the measuring unit 211 to the management server 100. The quality control means 214 serves to mediate the performance of quality control activities in accordance with the quality management activity plan established at the management server.

A station node agent will be described with reference to FIG. 4. 4 is a block diagram illustrating a station node agent according to an embodiment.

Station node agent 310 functions similarly to mobile node agent. However, since it belongs to the service providing side, it is not necessary to measure the response time, and the resource measuring means 311 is provided to measure the resource of the corresponding station node. In addition, the quality information transmitting means 312 and the quality control means 314 play a similar role as the quality information transmitting means and the quality control means 314 of the mobile node agent described above.

The management server 100 will be described. As described above, the management server 100 includes a management manager 110 and a data storage unit 130. The management manager 110 receives the service quality and resource information from the mobile node agent and the station node agent described above, and monitors 111 to evaluate the quality and manages to establish and execute a quality control plan corresponding to the evaluated quality. Means 112 are provided.

The monitoring means 111 will be described with reference to FIGS. 5 and 6. 5 is a block diagram illustrating a management server according to an embodiment, and FIG. 6 is a graph illustrating a classification state of service level quality and shape level quality according to an embodiment.

The quality information collecting unit 1111 receives service quality information and resource state information from the mobile node and the station node.

The quality evaluating means 1112 calculates the quality of the service instance level, the quality of the service type level, and the quality of the shape level from the service quality and the resource state information transmitted from the mobile node for each evaluation item. In addition, the quality of the configuration level is evaluated to determine whether the quality control process is performed.

Quality assessment items can be subdivided into three types: efficiency at the service instance level, efficiency at the service type level, and efficiency at the overall configuration level.

For example, service instance level efficiency can be divided into response time (RT) and throughput (TP) attributes. Evaluation items having an attribute of response time (RT) are AVG_RT _{ServiceInstance} (S _j ^k ) and SD_RT _{ServiceInstance} (S _j ^k ). AVG_RT _{ServiceInstance} (S _j ^k ) refers to the average time taken for a service instance to perform a function, and SD_RT _{ServiceInstance} (S _j ^k ) refers to the standard deviation taken for a service instance to perform a function. The RT attribute has a value greater than or equal to zero, and higher values indicate lower efficiency. Evaluation items having a property of throughput are AVG_TP _{ServiceInstance} (S _j ^k ) and SD_TP _{ServiceInstance} (S _j ^k ). AVG_TP _{ServiceInstance} (S _j ^k ) means the average of function calls that the instance successfully completed within the given time, and SD_TP _{ServiceInstance} (S _j ^k ) means the standard deviation of the function call successfully completed within the given time. . Throughput attributes range from 0 to 1, with higher values indicating higher efficiency.

Next, the efficiency at the service type level is also evaluated in terms of response time and throughput. Evaluation items for response time include AVG_RT _ServiceType (S _j ) and SD_RT _ServiceType (S _j ). AVG_RT _ServiceType (S _j ) represents the average of the response times of all service instances implementing the service type, and SD_RT _ServiceType (S _j ) represents the standard deviation of the response times of all service instances implementing the service type. The endpoint of the processing amount is AVG_TP (S _{j) ServiceType} to mean the average of the throughput of all service instances that implement the service type and SD_TP _ServiceType (S _j) to mean the standard deviation of the amount of all service instances that implement the service type There is. The value that each item has and its effective meaning are the same as described above.

The efficiency of the overall shape level can likewise be evaluated according to the nature of the response time and throughput. Rate this item to mean the standard deviation for AVG_RT _ME, AVG_TP _ME and current geometry throughput, which means SD_RT _ME, averaging throughput of current shape, which means the standard deviation of the current shape the response time, which means an average response time of the current shape There is an SD_TP _ME .

Prior to establishing a service quality management plan, the types of service management activities to be performed according to the service quality management plan and a method of determining whether or not to perform the management process will be described.

Service quality management activities can be divided into four categories according to the purpose and quality level combination as shown in Table 2 below. First of all, there are quality improvement activities at the quality of environment level (QoE level) and service quality improvement activities and service quality stabilization activities. Second, there are quality control activities at the service quality level, and service quality improvement activities and service quality stabilization activities.

<Table 2. Types of QoS Management Activities>

Here, EnhanceEff _SME () is a QoS quality management activity performed when the quality is degraded at the QoE level, and aims to improve QoS. EnhanceEff _ServiceType () is a QoS quality management activity performed when quality is degraded at the QoS level. The purpose is to improve QoS. In addition, BalanceEff _SME () is a QoS quality control activity performed when there is an excess of quality at the QoE level. In addition, BalanceEff _servicetype () is a QoS quality control activity performed when there is an excess of quality at the QoS level, and aims to stabilize QoS. For example, EnhanceEffServiceType () aims to improve the quality of a service type in question, which can be implemented by replicating or migrating some components of a service instance implementing the service type to another node. This will be described in detail later.

A method of determining quality will be described with reference to FIG. 6. The real-time quality judgment technique for service type and overall shape is determined as follows. First, the real-time quality judgment technique at the quality of environment level (QoE level) is based on the two user-defined thresholds, the lowest quality and highest quality thresholds. state, a poor state and an exceeding state. Steady state means that no quality improvement activities need to be taken because the current quality is in the normal range. The degraded state indicates that the current quality is deteriorating because the current quality is lower than the user-defined minimum quality threshold. In this case, activities to improve the current quality through shape reconstruction should be performed. The excess quality implies that the current quality is too good, because the current quality is below the user-defined high quality threshold. In some of these cases, the overuse of resources results in excess quality, so additional testing should be used to determine if resources are wasted. On the other hand, even if the overall quality is within the normal range, some services may be in a degraded state. In order to solve this problem, quality assessment at the service type level is performed in addition to the QoE level.

The highest and lowest thresholds for AVG_RT _ME and AVG_TP _{ME, which} are quality values for the overall geometry, are determined by the service manager, and the highest and lowest thresholds for AVG_RT _ServiceType (Ti) and AVG_TP _ServiceType (Ti), which are quality values for service types, are determined. Is determined based on the value specified by the service provider in the SLA. That is, in the QoS level quality determination, quality of each service is evaluated by comparing the quality of each service type constituting the current configuration with a threshold specified in a service level agreement (SLA). Quality determination at the QoE level and the QoS level is determined using both the response time (metric related to RT) and throughput (metric related to TP), which are previously defined quality models.

In the area A1 located within the threshold of both throughput and response time, it can be determined that the quality is good, and in the area A2 where the response time is small and the throughput is large, it can be determined that the quality is excessive. Can be. In addition, the area A3 having a large response time and a low throughput can be determined as a quality deficient state. The area A4 around the area A1 may determine the state of quality through further investigation.

On the other hand, the quality information display means 1113 shows the data transmitted from each node in a user interface in a form that is easy for the user to understand, and the quality shape management means 1114 stores the monitored and collected quality data or upon request. It serves to search.

The management means 112 will be described with reference to FIGS. 5, 7 and 8. 7 is a block diagram illustrating a conceptual service quality management activity according to an embodiment, and FIG. 8 is a block diagram illustrating a concrete service quality management activity according to an embodiment.

The quality management plan establishment means 1121 selects a quality management process to be performed by the service type and shape determined to perform the quality management process, and selects a target node and a destination node of the process. The quality management plan establishment means 1121 establishes a quality control plan through a two-step process of establishing a conceptual service quality management plan and a specific service quality management plan.

In the conceptual quality control planning step, service quality management activities to be performed are determined as shown in FIG. At this time, one or more service quality management activities may be included. Next, when a plurality of service quality management activities are included, the order between service quality management activities is defined.

In the specific QoS management planning step, as shown in FIG. 8, a problem service type and service instance are first determined, and then a method of implementing service quality management activities is determined. Service quality management activities include service component migration, service component replication, and load balancing table updates. Finally, the target node, that is, the target target, necessary for the implementation of each service quality management activity is determined.

Specifically, a service migration is a node that calls multiple services at the same time when the physical distance between the client and the service is too long, or when the response time is slow, such as network congestion, or when the node is no longer able to handle the load of the service or provides the service. Effective when resources are scarce Service replication is applied when the service usage spikes and the QoS is less than the expected SLA. In other words, it is applied to reduce the increase in the frequency of service use or the loss of QoS degradation.

These criteria are summarized according to their attributes and shown in Table 3 below.

TABLE 3; Decision Criteria for Migrating and Cloning Service Components>

If there is no defect in the performance or resources of the existing node itself in which the problematic service is deployed, the service is replicated to the node that is determined using the shortest path algorithm, and migrated otherwise. Through this process, new shapes and plans for building them are established. As above, service component migration and service component duplication are decided according to the current situation.

QoS management activities determined in conceptual QoS management planning are substantiated by specific procedures as shown in Tables 4 to 7 below.

TABLE 4; Realization Technique of EnhanceEff _ME ()>

Referring to Table 4, in order to improve the QoS quality at the overall configuration level, QoS quality improvement management activities for each service type are sequentially performed.

TABLE 5; Realization Technique of EnhanceEff _ServiceType (S _i )>

On the other hand, QoS quality improvement management activities at each service type level are determined by whether to migrate or replicate for each instance as shown in Table 5, and then determine the destination node, and then determine the quality according to the determined method and destination node. Perform management activities.

TABLE 6; Realization Technique of BalanceEff _ME ()>

Similarly, as shown in Table 6, QoS stabilization is sequentially performed for each service type in order to stabilize or balance QoS quality at the overall shape level.

TABLE 7; Realization Technique of BalanceEff _ServiceType (S _i )>

In addition, QoS quality stabilization management activity at each service type level determines whether to migrate or replicate for each instance as shown in Table 7, and after determining the destination node, quality management according to the determined method and destination node. Perform the activity.

In the present embodiment, instead of copying or migrating a binary package file including all components of a service, an Open-Closed Principle (OCP) is applied so that each node agent maintains a common structure between services and uses the service. Migrate and duplicate only the variable parts between them, that is, the variable components. The basic structure of a cloud service consists of a service implementation and a service interface. The service implementation is a collection of classes that actually implement the functions provided by the service, and the service interface plays a role of exposing the functions provided by the service to the service user. In other words, the service user can access the actual service implementation through the API exposed through the service interface. Since such a service interface and a service implementation vary depending on the type of cloud service, the service interface and service implementation are classified into a variable service structure in the present invention. The common structure of cloud services consists of abstract classes of service interfaces, service adapters, and communication stubs. The abstract class of the service interface provides a connection to the actual interface of the service, on which the user of the service can access the actual service. The service adapter is a component that enables dynamic service installation and deployment, and connects the variable structure of the service with the common structure. The service adapter makes the service implementation callable through the network, and the service user can call the service function implemented in the service implementation. That is, the variable component that is the object of migration and replication in this embodiment means such a service interface and a service implementation. This method is more efficient than migrating or duplicating entire components in view of the recent trend of increasing service size.

In addition, such a variable component can be dynamically placed in a corresponding station node by applying a class loading scheme by utilizing a service common structure provided in each node.

The quality management activity executing means 1122 executes the quality management activities according to the target node, the quality management process, and the target node selected by the quality management plan establishing means 113, that is, executing information related to the quality management plan to the corresponding node. To pass. The detailed process of the quality control activity performed by the quality control activity executing means 1122 will be described in detail in the description of each process below.

The configuration manager 1123 updates data regarding the shape stored in the data storage unit 130 including the migrated and replicated services.

Hereinafter, an outline of a process for a method of improving the scalability and availability of a cloud service according to an embodiment of the present invention, and a process for a specific quality control method will be described.

First, an overview of the overall process will be described with reference to FIG. 9. 9 is a flowchart schematically illustrating an overall process of a method for improving scalability and availability of a cloud service according to an exemplary embodiment.

First, the management server receives the service quality and resource information of the mobile node and the resource information of the station node (S10). Next, the management server determines whether the quality management process is performed using the collected information.

Next, the management server selects a quality management activity to be performed on the variable components of the target service is determined to perform the quality management process (S30). In this case, the quality control activity may select a service migration process and / or a service replication process as described above, and may further include a method of updating a load balancing table as necessary. Subsequently, the service quality management activity selected by the management server is performed (S50).

A service duplication process (S42) will be described with reference to FIG. 10 is a flowchart illustrating a service replication process according to an embodiment.

First, candidate node search and selection are performed as a first step (S421). That is, the management manager searches for a node having a currently available resource sufficient to install and deploy a corresponding service from the problem station node by referring to the resource information received from the station node. Among the searched target nodes, the highest node is selected as the destination node for service replication based on the available resources and network bandwidth.

The candidate node notification step is performed as the second step (S422). The management manager transmits information about the station node selected in the first step to the station node agent of the station node having the quality of service problem.

A service transmission step is performed as a third step (S423). The station node agent of the station node in question receives information about the destination station node to transmit the corresponding service from the management manager, and transmits a variable component among the corresponding services to the destination station node.

As a fourth step, the service dynamic installation and deployment step is performed (S424). The station node agent of the received station node dynamically installs the received service using the service adapter component and arranges the access through the network.

As a fifth step, a service replication result notification step is performed (S425). That is, when the service replication activity is completed, the station node agent of the target station node delivers the result of the replication activity such as whether the service replication is normally performed and additional information to the management manager of the management server.

As a sixth step, the cloud service structure update step is performed (S426). In other words, it updates the entire cloud service structure that has changed as a result of service replication activities. The updated information makes it possible to connect user calls to the newly replicated service and to monitor the quality of service.

The service migration process S43 will be described with reference to FIG. 11 is a flowchart illustrating a service migration process according to an embodiment.

A service transmission step is performed as a first step (S431). The station node agent of the target station node having the quality of service problem transmits a variable component of the target service that needs quality control to the management manager.

As a second step, the shape removal step of the target service is performed (S432). Remove the shape of the target service being migrated from the current shape of the cloud service system that provides the service. By removing the service geometry, the service is no longer exposed to the user through the target station node.

As a third step, a candidate node search and selection step is performed (S433). Search for a station node with enough available computing resources to install and deploy the target service to migrate the target service that is currently experiencing quality problems. The highest node among the discovered station nodes is selected as the destination node for service replication based on the available resources and network bandwidth.

A service delivery step is performed as a fourth step (S434). The manager delivers the service for migration to the station node agent of the selected target node.

As a fifth step, the service dynamic installation and deployment step is performed (S435). The destination station node agent that receives the service dynamically installs the received service by using the service adapter component and arranges the access through the network.

As a sixth step, a service migration result notification step is performed (S436). When the service migration activity is completed, the station node agent of the destination station node delivers the result of the service migration activity to the management manager.

As a seventh step, a cloud service structure update step is performed (S437). That is, update the entire cloud service structure that changed as a result of service migration activities. The updated information makes it possible to connect user calls to the newly migrated service and to monitor the quality of service.

On the other hand, the update of the load distribution table is to update the load distribution table provided in the existing load balancing node provided separately so that the load distribution is done in hardware. Detailed description thereof will be omitted. The update of the load distribution table can be provided or omitted in the system as needed. That is, in the case of the present invention, if a load balancing is performed by simply providing a separate hardware node, a management area should be set, and so on, but a management area is narrow. In the case of using only service replication and migration, all nodes connected to the network such as the Internet can be managed.

In the service scalability measurement step (S50), the scalability of the corresponding service is measured by comparing the quality before and after applying the scalability improvement technique according to the present invention. Service scalability may be calculated as in Equation 1 below.

Equation (1):

The denominator is the service throughput of the problem before applying the service scalability improvement technique. Molecule is the throughput measured after application of the technique. If the technique is successfully applied, the throughput of the service must be increased to meet the minimum quality. That is, scalability must have a value of 1 or more.

Although the preferred embodiments of the present invention have been described above, the technical spirit of the present invention is not limited to the above-described preferred embodiments, and various cloud services can be expanded within the scope not departing from the technical spirit of the present invention specified in the claims. Implemented as a method and system for improving performance availability.

100: management server 110: management manager
111: monitoring means 1111: quality information collection means
1112: quality evaluation means 1113: quality information display means
1114: quality shape management means 112: management means 112
1121: Means for establishing quality control plan 1122: Means for executing quality control activities
1123: CM manager 130: data storage unit
200: mobile node 211: measuring means
2111,311: resource measuring means 2112: response time measuring means
212, 312: quality information transmission means 214, 314: quality control means
300: station node 310: station node agent

Claims (16)

A first step of receiving, at the management server, service quality and resource information of the mobile node and resource information of the station node;
A second step of determining, by the management server, whether to perform a quality control process using the collected information;
A third step of selecting, by the management server, at least one of a service migration process and a service duplication process to be performed on a variable component among target services for which the quality management process is determined; And
And a fourth step of performing a service quality control activity selected by the management server.
The method of claim 1,
The selection of the service quality management activity includes selecting a target station node and a target station node.
The method of claim 1,
The service quality management activity in the third step further comprises a method of updating the load balancing table.
The method of claim 1,
Performing the service replication process,
Step 4a-1 of the management server selecting a station node having the maximum available resources as a destination station node by referring to the resource information transmitted from the station node;
Step 4a-2, wherein the management server transmits the information of the destination station node to the target station node whose quality of service is a problem;
And a step 4a-3 of transmitting, by the target station node, a variable component of a target service to the destination station node.
The method of claim 1,
Performing the service migration process,
Step 4b-1 transmitting the variable component to a management server by a target station node which is a target of quality management;
Step 4b-2 of removing the shape of a service to which the target station node belongs to the transmitted component;
Step 4b-3, the management server selecting a station node having the maximum available resources as a destination station node by referring to the resource information transmitted from the target station node;
And (4b-4) transmitting, by the management server, the transmitted variable component to the destination station node.
The method according to claim 4 or 5,
A fifth step of dynamically installing and making available the transmitted variable component after the service migration or replication process; And
And a sixth step of sending, by the target station node, the result of performing the performed service migration or replication process to the management server.
Claim 7 has been abandoned due to the setting registration fee. The method according to claim 6,
In the fifth step, a method of improving scalability and availability of a cloud service for dynamically installing and deploying the variable component using a class loading technique.
The method according to claim 6,
And a seventh step of comparing service throughput before and after the service migration or replication process to measure service scalability.
The method of claim 1,
The variable component is a method of improving the scalability and availability of a cloud service including a service implementation and a service interface, which is a set of classes that implement a service's functionality.
The method of claim 1,
In the third step, when the physical distance between the target station node that needs management and the mobile node using the service provided by the station node is far, when the response time is long, when the traffic of the target station node increases, and when How to improve the scalability and availability of cloud services that select a service migration process when at least one of the resources is scarce.
The method of claim 1,
The third step is to improve the scalability and availability of the cloud service to select a service replication process in case of a sudden increase in the use of the service provided by the managed station node or the quality of service (QoS).
In the cloud service system having a station node and a management server for providing a plurality of mobile nodes and services,
Monitoring means for determining whether to perform a quality management process by receiving service quality and resource information of a mobile node and resource information of a station node;
Quality control plan establishment means for selecting at least one service quality control activity of a service migration process and a service replication process to be performed on the variable components among the target services for which the quality management process is determined to be performed:
Quality management activity execution means for performing quality management activities according to the target node selected by the quality management plan establishment means, the type and quality of service quality management activities: Management server to improve the scalability and availability of the cloud service, including .
The method of claim 12,
The variable component is a management server that improves the scalability and availability of the cloud service including a service implementation and a service interface, which is a set of classes implementing the function of the service.
In the cloud service system having a station node and a management server for providing a plurality of mobile nodes and services,
Quality information transmitting means for collecting resource information of the station node and transmitting the collected resource information to the management server; And
Enhance the scalability and availability of the cloud service, including; quality control means for performing a quality management activity by receiving the execution command of any one of the service migration process and service replication process from the management server and information on the target destination station node; Station node.
15. The method of claim 14,
The quality control means,
Station node for improving the scalability and availability of the cloud service for transmitting a variable component of the quality management target service to the management server when receiving a command to perform a service migration process from the management server.
15. The method of claim 14,
The quality control means,
When receiving a service replication process execution command and destination station node information from the management server, the station node to improve the scalability and availability of the cloud service for transmitting a variable component to the destination station node.

Images