應用於ATCA接取設備之電信等級IGMP代理伺服器

Abstract

如何提高系統與服務的可用度，以達到電信等級的水準，對電信設備供應商與服務供應商來說都是一個令人困擾的難題。本論文採用先進通訊電腦架構(ATCA)的硬體平台，並以乙太被動光纖網路(EPON)的接取設備為例，探討在IPTV的服務架構中，如何提升ATCA-based局用接取設備上IGMP代理伺服器的可用度。目前常用的乙太網路恢復技術，如：Rapid Spanning Tree Protocol (RSTP)及Resilient Packet Ring (RPR)等，仍有過於複雜與拓墣結構的限制等問題，並不適用於ATCA-based接取設備。 本論文提出以負載分配、快速失效備援與無縫復原的機制，來提升ATCA-based IGMP代理伺服器的可用度。尤其是針對ATCA結構及multicast 封包的獨特性，提出較簡易又快速的方法來達到備援同步與封包導向的目的，這些機制都大幅增進系統失效備援的效能。在本論文最後，也提出IGMP代理伺服器失效備援的速度的量測原理與方法，其模擬結果平均約45ms，符合電信等級要求的50ms。

How to increase system or service availability and to reach carrier grade is always a grand challenge for telecom equipment venders and service providers. The thesis adopts Advanced Telecom Computing Architecture (ATCA) as a hardware platform and Ethernet Passive Optical Network (EPON) as an example to research how to improve the availability of IGMP proxy server on an ATCA-based central office access equipment for IPTV application. There are lots of Ethernet resiliency technologies had been developed, such as Rapid Spanning Tree Protocol (RSTP) and Resilient Packet Ring (RPR) etc. They still have an over complication and topology limitation problems for an ATCA-based access equipment. The thesis proposes using load sharing, fast failover, and seamless resuming mechanisms to improve the availability of IGMP proxy server. The proposed methods take the characteristics of ATCA-based topology and multicast traffic to simplify and to speed up the standby synchronization and traffic redirection. These improvements immediately shorten the failover time of IGMP Proxy and improve system availability. Finally, the thesis also proposes the failover time measurement principle and method. The average simulation result is around 45ms. It is better than 50ms, which is the carrier-grade failover requirement specified by SONET/SDH.