網路服務系統崩潰的問題探討與解決方案

Network Service System Crashes and the Effective Solution Design

盧亞德(Ya-Te Lu)

指導教授 ： 林永松

共同指導教授 ： 游張松(Chang-Sung Yu)

國立臺灣大學/管理學院/資訊管理組/碩士(2016年)

https://doi.org/10.6342/NTU201700181

摘要

科技不斷推陳出新，企業於是引進電腦與網路，實施營運的電腦化，以電腦的速度同時服務更多客戶，創造更大的營收業績。現在，更成為重要業務推展以及行銷販售的主要管道。網路服務雖然是企業營運的關鍵核心，然而，網路服務崩潰卻會導致營運中斷、服務作業停擺。在媒體上常聽到網路服務崩潰的發生。災難型態的網路服務崩潰，除了導致商業損失之外，更造成企業形象受損，客戶流失等嚴重後果，必須事先防治、解決後患。

解決網路服務崩潰問題的傳統方法，熟知的增購大型主機擴大服務容量，或擴大各個服務模組的設備元件，全面提升系統容量，或者是使用雲端運算技術，建制容錯備援機制等等…，這些手段似乎都無法有效解決網路服務系統的崩壞問題。

本論文透過崩潰率的數學理論，分析各種網路服務強化技術，並探討其解決網路服務崩潰問題的限制。本論文再透過崩潰率的數學模型，提出TQMCR Model(Ticketing, Queuing, Monitoring, Control, Report 模型)的管理模型，構建商業服務資源管理系統(Business Service Resources Management System)解決方案，有效解決網路服務系統崩潰問題。也用高鐵訂票系統為案例，實際驗證BSRM成效。

未來，網路服務系統崩潰已經成為必須面對的經常性問題。例如，駭客針對分散式服務阻斷攻擊，或以目標持續性的應用程式威脅攻擊，甚至透過應用程式機器人攻擊，都會讓網路服務系統崩潰不斷地發生。面對網路服務面臨的挑戰日新月異，未來可以再延伸本論文提出的TQMCR Model，提出相對應的解決方案。

關鍵字

網路交易 ； 網路服務崩潰 ； 崩潰率 ； 隊列 ； 負載平衡 ； 商業服務資源管理系統 ； TQMCR模型

並列摘要

Internet service is booming, and many business practices prefer online marketing and online transactions as key business promotion or sales channels. However, the media often hear the collapse of network services, resulting in business disruption and even business interruption. In addition to business losses, but also caused damage to corporate image, customer churn and other serious consequences.
Typical approaches to solve the problem include the use of cloud computing technology to expand the capacity of services, or the various components of the service system to fully enhance the service capacity, or the establishment of fault-tolerant backup system, these methods seem to be unable to effectively solve the network service system Collapse of the problem.
In order to solve the problem of service collapse effectively, this paper explores various network service enhancement technologies by means of the mathematical method of crash rate. At the same time, TQMCR Model (Ticketing, Queueing, Monitoring, Control, Report) is proposed to build a “Business Service Resource Management” system, an effective solution to network service system crash. High-speed rail booking system is also used as a case, the actual test results BSRM.
In response to the challenges of Internet services, hackers for distributed services to block attacks, or target persistent application threat, or even through the application of robot attacks, resulting in network service system crash, the future can be extended this paper TQMCR Model, to strengthen the integrity of the solution and continuous improvement.

並列關鍵字

Network Service System Crashes ； TQMCR Model ； Crash Rate ； Queueing ； Load Balance

參考文獻

1. C.M. Assi ; Yinghua Ye ; Sudhir Dixit ; M.A. Ali (2003), Dynamic bandwidth allocation for quality-of-service over Ethernet PONs

2. Roy D. Williams (1991), Performance of dynamic load balancing algorithms for unstructured mesh calculations

3. Yuan-Chieh Chow ; Kohler (1979), Models for Dynamic Load Balancing in a Heterogeneous Multiple Processor System

4. Ritwik Banerjee; Abbas Razaghpanah; Luis Chiang; Akassh Mishra; Vyas Sekar◦ Yejin Choi; Phillipa Gill (2013), Internet Outages, the Eyewitness Accounts: Analysis of the Outages Mailing List

5. Deep Inside a DNS Amplification DDoS Attack.