KR20100071613A - Usn middleware query processing load balancing systems and the method - Google Patents

Abstract

PURPOSE: A processing load balancing system and method thereof of the USN(Ubiquitous Sensor Network) middleware quality are provided to be proceed the load balancing of the requirement of the excessive quality. CONSTITUTION: According to the apparatus is the configuration information, it registers in the corresponding USN middleware system and it interlinks. The apparatus transfers the information of the quality about the request of the quality from USN application service through the evaluation process from suitable USN middleware system. The apparatus transfers the request finish result of the quality delivered from USN middleware system to the corresponding USN application service.

Description

USS middleware query processing load balancing systems and the method

The present invention provides a system that applies a weighted load balancing algorithm capable of distributing the USN sensor information query in the USN (hereinafter referred to as "USN") middleware POOL, thereby improving the performance of the existing system and The present invention relates to a USN middleware query processing load balancing system and a method for smoothing sensor information query request distribution processing that can simultaneously solve a scalability problem.

Computer service provides computing service by actively monitoring environment information of user. Recently, various types of sensors including RFID are being deployed everywhere to monitor user environments required for ubiquitous services. These sensors periodically monitor environmental information and deliver data to the computing environment. In this case, the data received in the computing environment is called a data stream because it is very large and has a characteristic of being updated continuously and periodically.

Data collected from various sensors represents a form of stream data that is continuously generated and transmitted through a network, and there is a growing interest in developing a stream data processing system for efficiently processing such stream data.

On the other hand, the query for stream data in the stream data processing system is to be distinguished from the one-time query in which the data set to be queried is queried in a predetermined environment as in the existing database management system to obtain the result. Is registered first, and it is called continuous query because it searches continuously applying the query to the stream data that is continuously generated and input.

In this case, in the case of continuous query, since one or more (possibly dozens or hundreds) continuous query processes are performed on each stream data collected by one or more sensors, the data throughput is enormous.

   Currently developed and used service platforms are Web Service, J2EE, and OSGi. These platforms provide component-based standardized computing environments for distributed services. However, they do not consider the data stream that occurs in the ubiquitous environment as input data. They provide services by managing data primarily by reading and writing to a static data store, which is constantly updated and has a performance problem in reading and writing large amounts of data streams.

 The proposed stream processing engine for data stream processing supports very efficient data processing only for new SQL-based query models for data streams and limited database operations such as selection and aggregation.

  However, the stream processing engine is limited in developing and operating more general and complex forms of service logic.

   Most useful high value-added services are written in program code that can't be expressed in SQL.

  As the algorithm for finding the optimal route in real time based on the road information updated in real time by the prior art, the data used by Dijkstra algorithm and the like cannot be represented in SQL.

As the USN application service expands or the USN sensor network grows, the number of requests for sensor information queries can increase, and thus, many studies have been conducted to solve this problem, which inevitably decreases the response speed of USN middleware. .

For example, Korean Patent Laid-Open Publication No. 10-0786504 discloses a dynamic session management apparatus for knowledge-based home network service in a ubiquitous environment, wherein a sensor value (user recognition information) according to a user's movement is received. A session managing engine for checking a current location of a user and whether a service is registered using an identifier of a value; A service control engine for selecting a service suitable for a user according to user location information from the session management engine, extracting collision information and error information by performing policy situation verification on the selected service; After generating a query using the location information of the user extracted from the session management engine and the service, collision information, and error information extracted from the service management engine, the query is generated and transmitted to the inference engine, and the service provision information received from the inference engine is provided. A query generation engine for delivering to the management engine (Query Generating Engine); And a service command management engine for extracting information corresponding to each service bundle from the service providing information received from the query generation engine and transmitting the information corresponding to each service bundle to the corresponding service bundle, and managing the corresponding service. The dynamic session management device of knowledge-based home network service is disclosed in

Korean Patent Publication No. 10-0793057 discloses an information service providing system based on a heterogeneous sensor network, comprising: sensing means for configuring a sensor network and sensing information of a surrounding environment; Sensor network data transmission means for collecting data sensed in real time from the sensing means and delivering the message to a USN middleware means as a message; Sensor network abstraction means for extracting sensing data from the message transmitted through the sensor network data transmission means; Sensor network intelligence means for analyzing the sensing data extracted by the sensor network abstraction means to extract things and environment information, and to generate events (knowledge content) and environment information from the extracted things and environment information; And a service platform management means for providing the application program with the event (knowledge content) and environmental context information generated by the sensor network intelligence means in a form desired by the application program. Described,

Publication No. 783,679 A middleware system that facilitates the development, deployment, and provision of services based on data streams. In particular, it is a large quantity and continuous such as a radio frequency identification (RFID) system or a ubiquitous sensor network (USN). An apparatus and method for supporting a service using a data stream updated as an input, the method comprising: selecting a service unit and a stream source description language file designated by a service unit activation language file; Sending a query to the stream provider using the query made by the service unit activation language file and the address of the stream provider written in the selected stream source description language file; Preparing a buffer according to schema information on a result of the transmitted query; Creating an instance of the selected service unit and initializing the service unit in an execution mode designated by the service unit activation language file; Updating in real time a data stream stored in the prepared buffer with a continuous data stream provided from the stream provider; In response to a user request, a data stream mediation processing method is disclosed, which comprises processing data stored in the buffer in a given execution mode and delivering a request result to a user.

In the publication number 813,000 A stream data processing system, comprising: evaluation result storage means for updating and storing a query condition evaluation result; Window evaluation means for performing window evaluation on the stream data to select data to be subjected to next query condition evaluation; Data separating means for separating the data received from the window evaluating means into new data and redundant input data according to whether the previous query condition evaluation result can be reused; Reuse result extraction means for extracting a query condition evaluation result for the duplicate input data received from the data separation means from the evaluation result storage means; Query condition evaluation means for generating a query condition evaluation result by performing query condition evaluation upon receiving new data from the data separation means; And a result structuring means for collecting and outputting query condition evaluation results received from the reuse result extracting means and the query condition evaluating means, respectively, and delivering the result to the evaluation result storage means.

The sensor information query request requested by the USN application service is received by the USN middleware, analyzed, and then converted to a command suitable for the corresponding sensor network and transmitted to the USN sensor network. USN sensor network executes instructions from USN middleware and passes the result to USN middleware. USN middleware converts the results from USN sensor network to the appropriate data format required by USN application service and ends the process. do.

  However, as the USN application service is expanded or the size of the USN sensor network increases, the number of sensor information query requests may increase, and thus, the response speed of the USN middleware may inevitably decrease.

  As a solution to this problem, there is a method of upgrading the hardware specification of USN middleware that has already been introduced, and a method of solving the problem through the introduction of USN middleware. But both methods present a problem.

  The first method can cause the suspension of USN application services and backup and restore of existing data during the upgrade of hardware specifications of USN middleware.

  The second method is a problem to be solved by the present invention, which causes a complicated situation in which the USN application service program needs to be partially modified to accommodate all of the introduced middleware.

The present invention is a technical configuration applying a weighted load balancing algorithm capable of distributing the USN sensor information query in front of the USN middleware in order to solve the above problems, for a query processing request to the USN middleware system, load balancing system Means for storing an evaluation result for this query request and a query response; Means for selecting a USN middleware system capable of performing a query request in a short time using the stored evaluation result; Send a query to a corresponding USN middleware system; It is a problem solving means of the present invention to provide a USN middleware query processing load balancing system and a method configured as a means for receiving a query processing result and storing processing performance.

According to the present invention, the distributed processing of the USN sensor information query request enables load balancing of excessive query requests that may come from the development stage of the USN application service. In other words, a totally new type of application service can come out in various forms through the convergence between USN application services, including USN application service, and the size of the same sensor network that can be seen in u-City business is also rapidly increasing. According to the present invention, distributed processing of query requests is essential, and through the present invention, it is possible to provide a stable and convenient application service, and to prepare in advance in the upcoming u-City business. have

In order to achieve the above object, the present invention provides a means for loading the configuration information for the middleware system connection in the system startup process; Means for registering and connecting to the corresponding USN middleware system according to the configuration information; Means for conveying query information to a suitable USN middleware system through an evaluation process for a query request from a USN application service; Means for delivering a query request completion result delivered from the middleware system to a corresponding USN application service; Means for storing a query evaluation result from the query request completion result; Consists of means for unregistering and disconnecting during system shutdown

USN middleware query processing load balancing system,

 Identifying query processing status information which is performed by the USN middleware to perform USN sensor information query distribution processing by using the system; A query request management step requested by the application service; Analyzing the query request information; The present invention relates to a USN middleware query processing load balancing method including a distributed application step considering query processing state information.

  As shown in FIG. 1, the overall USN middleware query processing load balancing diagram of the present invention includes n application services and n USN middleware, for distributed application of query requests therebetween. Consists of one load balancing server,

2 is USN middleware query processing load balancing of the present invention As a detailed view,

 It can be seen that a LoadBalancer for identifying and managing USN sensor information, an EstimationManager connected to the LoadBalancer, an PerformanaceMonitor connected to the EstimationManager, a QueryManager connected to the other side of the EstimationManager, and a QueryExecutionLogManager connected to the QueryManager. For reference, in the process of starting up, USN middleware Proxy object corresponding to agent for each USN middleware system is created by Spring Framework environment, and in case of PerformanceMontor, it runs in clone form by Quartz scheduler.

therefore, EstimationManager includes: evaluation means for evaluating query requests through QueryManager and performance resources of middleware system through PerformanaceMonitor; A comprehensive evaluation means considering both query request evaluation and performance resource evaluation, and selects a high performance USN middleware system.

   QueryManager is a total time calculation function considering the completion time for each query request for each middleware system for the currently executing query request; And average completion time calculation for past completed request queries; It is responsible for calculating the comprehensive completion estimation time for each middleware system,

  PerfomanceMonitor is a USN middleware query processing load balancing system that is responsible for evaluating the available performance of middleware systems through performance resources (CPU information, CPU utilization, memory information, memory utilization, network traffic information, network traffic utilization). It is

Korean Registered Patent Registration No. 783679 Even when developing a new stream-based service, there is no need to code complex stream connection and processing parts, so that the development of the service is very easy and the development time is shortened. By applying a balancing algorithm, it is a technique for improving the response speed of the USN middleware, it can be seen that the present invention is different in the way of processing by distributed to increase the query response speed in the USN middleware,

  Korean Registered Patent Registration No. 813,000, which speeds up the query response speed, selects a method that prevents data from being duplicated by using a load balancing server, but the present invention uses the Telecommunications Technology Association Standard TTAS.KO-06.0170 According to the "USN middleware platform reference model", as a basic function of the USN middleware system, it has a data redundancy processing function for query requests, so the data balancing processing control function for query requests is not processed by the load balancing system, but USN sensor information query By applying a weighted load balancing algorithm capable of distributed processing, it can be seen that the method of applying the distributed method to each USN is a technique for improving the response speed of the USN middleware.

2 is USN middleware query processing load balancing of the present invention The detailed diagram will be described in more detail as follows.

  In the step of identifying query processing status information performed by the USN middleware, it is a step of identifying information such as query request contents, query processing results, and query application considering the query processing time.

  In the query request management step requested by an application service, a query request from various application services is stored and stored in a queue for information about request time, request query, priority of query quality, and response request time of a request for each application service.

  In the query request information analysis step, the priority is determined by analyzing the query information requested by each application service and the queries stored in the queue, and reordering the query processing order.

  Refers to a middleware server having excellent availability status by identifying the availability status of USN middleware in the distributed application stage, and applying and distributing query contents according to the rearranged query processing order.

The grasping step of collecting query processing status information performed by the USN middleware refers to a step of collecting information such as query request contents, query processing results, and query application considering the query processing time.

 In the query request management step requested from an application service, a query request from various application services is stored and stored in a queue for information related to request time, request query, priority of query quality, and request completion time for each application service.

 In the query request information analysis step, the priority is determined by analyzing the query information requested by each application service and the queries stored in the queue, and reordering the query processing order.

  LoadBalancer loads middleware configuration information (config.properties) during startup and creates a Proxy object corresponding to USN middleware by loadBalancerContext.xml and authenticates and registers each Proxy object with each middleware system; A function for canceling registration and disconnection from each middleware system during shutdown; And a management function for sessions created during authentication with each USN application service system; And selecting a suitable middleware through a result evaluated from EstimationManagger for a plurality of middleware systems with respect to the requested query request, and transmitting the query; It is in charge of delivering the result of query request to the corresponding USN application service.

Table 1 config configuration file

Table 2 loadbalancer context.XML

EstimationManager performs a performance evaluation on a query request requested from the USN application service system, and then selects an appropriate middleware system and delivers the query request. Obtaining an estimated completion time for a query request being executed for each middleware system and an average completion time for an already completed query request from the QueryManager for performing a performance evaluation on the query request; A function of obtaining average performance information on performance resources of each middleware system from the PerfomanceMonitor; and selecting a suitable middleware system for a corresponding query request through a comprehensive performance evaluation process; It is responsible for passing this to LoadBalancer.

 The QueryManager stores and manages query request information requested from the USN application service system as a queue data structure type, checks whether a new query request has not existed before, assigns a query ID to a new query request, and provides an existing query. A function that saves the QueryExecutionLog data through the QueryExecutionLogManager when the query is completed and utilizes an already assigned ID if it matches the request; Calculate the sum of estimated execution time for each middleware for the current query request, and if the query request is not in progress, search the past execution result from QueryExecutionLogManager and include the estimated execution time in the calculation, but there is no past execution result from QueryExecutionLogManager. In this case, the function does not calculate the estimated execution time. Query execution management information being executed is as follows. Query start time, estimated execution time of query execution, query ID, and query execution middleware ID.

  QueryExecutionLogManager provides a function of storing query request information and query processing results requested from the USN application service system, and provides a plurality of QueryExecutionLog data for a specific query request from the QueryManager. The attributes that make up QueryExecutionLog are as follows. Query start time, query completion time, middleware unique ID, query ID given by QueryManager, and log data validity.

 The PerformanceMonitor has a function of collecting collection of performance resources (CPU information, CPU utilization, memory information, memory utilization, network traffic information and network traffic utilization) for each USN middleware system at regular intervals; It collects by utilizing SNMP service installed in USN middleware system. When requesting a query, the middleware system should be selected through evaluation of performance resources.

  It can be seen that the evaluation method of each system is evaluated as described in the following table.

Table 3 Calculation of query request evaluation

Table 4 Evaluation calculation results by performance resources

Table 5 Overall Evaluation Calculation for Specific Query Requests

Table 6 Calculation of appropriate middleware selection

1 is a general diagram of a US middleware query processing load balancing system of the present invention.

Figure 2 is a detailed diagram of the US middleware query processing load balancing system of the present invention.

Claims (12)

In the USN middleware query processing load balancing system,   Means for loading configuration information for connecting a middleware system in a system startup process; Means for registering and connecting to the corresponding USN middleware system according to the configuration information; Means for conveying query information from a suitable USN middleware system through an evaluation process on a query request from a USN application service; Means for delivering a query request completion result delivered from the middleware system to a corresponding USN application service; Means for storing a query evaluation result from the query request completion result; USN middleware query processing load balancing system, characterized in that it comprises a means for registering and disconnecting during system shutdown. The method of claim 1, wherein the query information is transmitted from a suitable USN middleware system through an evaluation process for a query request from the USN application service.  Performance evaluation is carried out in EstimationManager in part, evaluation means for the evaluation of the query request through the QueryManager and the performance resources of the middleware system through the PerformanaceMonitor; A comprehensive evaluation means that considers query request evaluation and performance resource evaluation together; and selects a high performance USN middleware system. USN middleware query processing load balancing system. The method of claim 1, wherein the query information is transmitted from a suitable USN middleware system through an evaluation process for a query request from the USN application service.  The performance evaluation is partially performed in the QueryManager, and includes a total time calculation function considering a completion time of each query request for each middleware system with respect to a running query request; An average completion time calculation function for the previously completed request query; And the USN middleware query processing load balancing system, which is responsible for calculating a comprehensive completion estimation time for each middleware system. The method of claim 1, wherein the query information is transmitted from a suitable USN middleware system through an evaluation process for a query request from the USN application service.  Performance evaluation is partly performed in PerfomanceMonitor, and is used to evaluate the usable performance of middleware systems through performance resources (CPU information, CPU utilization, memory information, memory utilization, network traffic information, network traffic utilization). USN middleware query processing load balancing system. In the USN middleware query processing load balancing method,  Identifying query processing status information which is performed by each USN middleware; A query request management step requested by the application service; Analyzing the query request information; USN middleware query processing load balancing method comprising a distributed application step taking into account the current state of the query processing. The method of claim 5, wherein the analyzing of the query request information comprises: EstimationManager,  In QueryManager, QueryExecutionLogManager, and PerformanceMonitor Is performed, EstimationManager performs performance evaluation on the query request requested from the USN application service system, selects an appropriate middleware system, delivers the query request, and performs performance evaluation on the query request from the QueryManager. Obtaining an estimated completion time for a query request being executed for each system and an average completion time for an already completed query request; A function of obtaining average performance information on performance resources of each middleware system from the PerfomanceMonitor; and selecting a suitable middleware system for a corresponding query request through a comprehensive performance evaluation process; A suitable middleware system will then forward it to LoadBalancer,    The QueryManager queues and manages query request information requested from the USN application service system, checks whether a new query request has not existed before, assigns a query ID to a new query request, and matches with an existing query request. A case that utilizes an already assigned ID and stores QueryExecutionLog data through QueryExecutionLogManager when the query is completed; The sum of the estimated execution time for each middleware for the current query request is calculated. If the query request is not in progress, the previous execution result is retrieved from the QueryExecutionLogManager and the estimated execution time is included in the calculation. If not, it is responsible for not estimating the estimated execution time,   QueryExecutionLogManager has a function of storing query request information and query processing results requested from the USN application service system; Provides multiple QueryExecutionLog data for a specific query request from the and QueryManager.  The PerformanceMonitor has a function of performing collection on a periodic basis for performance resources (CPU information, CPU utilization rate, memory information rate, memory utilization rate, network traffic rate and network traffic rate) for each USN middleware system; It collects by utilizing SNMP service installed in USN middleware system. A USN middleware query processing load balancing method characterized by selecting a middleware system by evaluating a performance resource when a query is requested. The method of claim 5, wherein the analyzing of the query request information comprises: EstimationManager,  In QueryManager, QueryExecutionLogManager, and PerformanceMonitor The request query calculation calculation for the middleware is performed by the following equation.  Formula N number of query requests executing in kth middleware

m number of executed queries in k middleware

Sum of query evaluations in the k middleware

The method of claim 5, wherein the analyzing of the query request information comprises: EstimationManager,  In QueryManager, QueryExecutionLogManager, and PerformanceMonitor USN middleware query processing load balancing method, characterized in that the performance evaluation performed by the performance resources for the middleware is evaluated by the following equation.  Formula for k middleware  CPU resources: number of CPUs, CPU processing speed, CPU utilization

Memory resources: memory capacity, memory processing speed, memory utilization

Network resources: network speed, network traffic utilization

 Overall system performance evaluation

: Weights Determined by the Load Balancing System Administrator The method of claim 5, wherein the analyzing of the query request information comprises: EstimationManager,  In QueryManager, QueryExecutionLogManager, and PerformanceMonitor The overall evaluation calculation calculation considering the query evaluation result and the performance evaluation result for the middleware is calculated by the following equation.  Formula for k middleware

: Weights Determined by the Load Balancing System Administrator The method of claim 5, wherein the analyzing of the query request information comprises: EstimationManager,  In QueryManager, QueryExecutionLogManager, and PerformanceMonitor A suitable middleware selection calculation calculation is performed by the following equation: USN middleware query processing load balancing method. Formula

The middleware system with the lowest value among the overall evaluation results is selected as a suitable middleware system for query delivery. The method of claim 5, further comprising: identifying query processing state information performed by each of the USN middlewares; The query request management step requested by the application service is  Performed in LoadBalancer, the function of authenticating and registering each middleware system after loading middleware configuration information during startup; A function that is responsible for unregistering and disconnecting from each middleware system in the shutdown process, and managing a session created during authentication with each USN application service system; And selecting a suitable middleware through a result evaluated from EstimationManagger for a plurality of middleware systems with respect to the requested query request, and transmitting the query; A USN middleware query processing load balancing method characterized by delivering a query request completion result to a corresponding USN application service. 12. The method of claim 11, further comprising: identifying query processing state information performed in each of the USN middleware; The query request management step requested by the application service is  Load balancing method performed by LoadBalancer that loads the config.properties configuration information file, creates N USNMiddlware objects, and then registers and manages N USNMiddleware objects in the LoadBalancer object. <? xml version = "1.0" encoding = "UTF-8"?> <beans xmlns = "http://www.springframework.org/schema/beans" xmlns: xsi = "http://www.w3.org/2001/XMLSchema-instance"        xsi: schemaLocation = "http://www.springframework.org/schema/beans http://www.springframework.org/schema/beans/spring-beans-2.5.xsd">     <!-First USN Middleware->     <bean id = "middleware1" class = "net.uspider.USNMiddleware">         <property name = "id" value = "$ usnmw1.id" />         <property name = "host" value = "$ usnmw1.host" />         <property name = "userid" value = "$ usnmw1.userid" />         <property name = "port" value = "$ usnmw1.port" />         <property name = "proxyport" value = "$ usnmw1.proxyport" />         <property name = "password" value = "$ usnmw1.password" />         <property name = "snmpUserId" value = "$ usnmw1.snmp.userid" />         <property name = "snmpCommunity" value = "$ usnmw1.snmp.community" />         <property name = "snmpVersion" value = "$ usnmw1.snmp.version" />     </ bean>     ................     <!-Nth USN Middleware->     <bean id = "middlewareN" class = "net.uspider.USNMiddleware">         <property name = "id" value = "$ usnmwN.id" />         <property name = "host" value = "$ usnmwN.host" />         <property name = "userid" value = "$ usnmwN.userid" />         <property name = "port" value = "$ usnmwN.port" />         <property name = "proxyport" value = "$ usnmwN.proxyport" />         <property name = "password" value = "$ usnmwN.password" />         <property name = "snmpUserId" value = "$ usnmwN.snmp.userid" />         <property name = "snmpCommunity" value = "$ usnmwN.snmp.community" />         <property name = "snmpVersion" value = "$ usnmwN.snmp.version" />     </ bean>     <!-Register N USN middleware with Load Balancer Manager->     <bean id = "loadBalancerManager" class = "net.uspider.LoadBalancerManagerImpl">         <property name = "middlewares">         <list>         <ref bean = "middleware1" />         ..........         <ref bean = "middlewareN" />         </ list>            </ property>          </ bean> </ beans>

Images