以WEB為基礎的應用系統效能監測管理系統之研究

Abstract

Internet的快速興起，對人們生活造成相當的大的影響力，也促成應用系統的變革，從過去採用主從式架構（Client-Server），轉變為大量採用Web的一股潮流。個案對象主要之Web-based applications，即以Java語言開發後在J2EE平台上運作。然因其應用系統規模日增及資源有限，致系統服務偶有不穩定、無法及時回應或停止等異常情形發生，而管理維護者在缺乏工具輔助下，對於異常事件獲知、問題診斷處理之時效，服務中斷之預防及資源使用之趨勢等方面，難以有效掌握。 企業或政府機關所提供服務之應用系統須具備良好的效能來滿足顧客需求，故應用系統的管理維護者有其必要性將所提供的這些應用系統服務加以監測管理。本研究即針對個案對象之Web-based應用系統運作情形，建置一套效能監測管理機制，其架構分為伺服主機資源效能監測、應用系統運作監測、異常通知等三個部分。 本研究結果為透過效能監測管理機制的運作下，對於Web-based應用系統發生服務異常之獲知時效及初步問題診斷上，可獲得主動、迅速及先期預警效果，並可有效降低異常情形發生率來提升其可用性，同時，藉由彙集資源使用趨勢將有限資源作有效之分配管理，對於確保資訊服務品質，降低資訊運作成本及提升顧客關係等面向，皆具有正面的助益。

The rapid proliferation of the Internet usage has created, besides substantial impacts on people's daily lives, dramatic changes in the building of application systems – from the once dominant client-server architecture to the widely adopted web-based application architecture. The case under study is a system of web-based applications developed using Java to run on the J2EE platform. As the application systems grow in scale but with limited computing resources, some of the system services occasionally run into abnormal situations such as degraded stability, failure to respond in timely manner, or unintended stop. Lack of management tools makes the system maintenance personnel hard to be notified of abnormal events on their occurrence, to be able to diagnose and handle the problems promptly, to prevent service interruption, and to grasp the trend of computing resource utilization. The application systems that provide enterprise or governmental services must have good performance to meet customers’ needs. Therefore, it is necessary for the system administrator to monitor and manage these application systems services. This research focuses on the web-based application system operation of the case under study to build a performance monitoring and management mechanism which is characterized by the architecture of three major components: unit to monitor the performance of server resources, unit to monitor the operation of the application systems, and unit for abnormal event notification. This research proves that a web-based application system can, via an integrated performance monitoring and management mechanism, attain active and prompt preliminary system diagnosis and maintenance when the systems encounter abnormal operations and early warning notification is given. It is possible for the application systems to achieve high availability by effectively lowering the occurrence rate of abnormal operations. Meanwhile, the accompanying report on the trend of system resource usage compiled from the data enables optimal allocation and management of limited computing resources that ensures high quality and low operation cost of the provided information services, as well as enhanced customer relation.