| 摘要: | 從二十世紀末網際網路開始蓬勃發展至今，人與人之間的溝通得以更緊密的結合。這帶來了網路流量迅速成長，卻也造成網路供應商最棘手的問題，即業者該如何對其網路進行更有效率的品質管理。 網路資源及電路頻寬是有限的，而客戶的需求卻永遠不可能被滿足，這樣的狀況常造成網路頻寬壅塞，進而影響服務品質。因此，網路供應商一則思考如何開源，另一則思考該如何進行節流。開源的作法不外乎增加聯外頻寬、添購網路設備及新增機房等，但供應商不可能永無止盡的將所有資源投資於此。 節流首重網路品質管理，其目的為提供更好的傳輸服務品質及提高網路的執行效能，尤其需針對已簽署服務等級協議（Service Level Agreements）的客戶，更須針對其網路品質提供頻寬（Bandwidth）、網路可用度（Availability）、傳輸延遲（Latency）以及封包遺失率（Packet Loss Ratio）等特定項目之服務品質保障。 但一個大型的網路服務供應商有著數以千計且不同廠牌的網路設備，包含Cisco、Foundry、Juniper、Alcatel等，共同串接成環島網路，且各式各樣的路由協定運行其上，在如此複雜的網路環境中，只要有些微的變化不察，就足以影響網路服務品質。 本論文之研究動機主要為探討業界網路品質指標監控之實務運作狀況，希望能藉由訪談的結果探討個案公司如何對自己的網路品質進行監控，包含使用哪些程式、應用方式、監控流程及實際運作後所帶來的效益，最後分析整理結果，期望能歸納一套較為完整的網際網路品質指標建構觀。 Internet has been prospering since the end of the 21st century making the Communication between people more close. It brings the rapid growth of the network traffic and also a most knotty problem for internet service providers, that is, how to control their own network quality more efficiently. Network resources and the circuit bandwidth are limited, whereas users’ demand never be met. Such condition often causes the network bandwidth congestion thereby lowering the service quality. Therefore, internet service providers try to find means to bring more network resources for their customer and reduce the expenditure for businesses itself. Ways to produce resources include buying the new network equipments, building the new control rooms, etc.. However, network suppliers aren’t able to invest all resources on these endlessly. The primary task reducing the expenditure is to control the network quality more strictly in order to provide better quality of the packet transportation and improve the network performance for business, especially who signed the Service Level Agreements which will guarantee four service items: circuit bandwidth, network availability, packet latency, packet loss ratio. However, a large-scale service provider has thousands of different network facilities of varied brands including Cisco, Foundry, Juniper, Alcatel, etc. with divergent routing protocol runs on it. In such a complex network environment, once there are any slight changes happened without noticing, it will affect the quality of the network service. This research through interviews is to investigate the actual working condition of how an enterprise can monitor its network infrastructures with network quality indicators. We hope to understand the rules of how a network supplier monitors their own network, which includes uses of programs, applications, control flows, and benefits after implementing. Finally, we analyze the case and draw a conclusion to establish a more complete evaluation model of the network quality. |
