Providing High Availability and Dynamic Bandwidth Allocation for Triple-play Services over Carrier Ethernet

• Main Authors: Kun-Hua Cheng, 鄭昆樺
• Other Authors: Shang-Juh Kao
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/37971250892264201920

碩士 === 國立中興大學 === 資訊科學與工程學系所 === 97 === Typically, voice, video, and data are named the triple-play services. With the rapid growth of triple-play services, how to construct a service network architecture with stable transmission quality is an urgent research topic. In this thesis, a triple-play service network architecture with dynamic bandwidth management and high network availability over Carrier Ethernet is proposed. The proposed architecture consists of access networks and the IP/MPLS core network. In the access network, the traffics of triple-play services in the same truck are virtually isolated using Virtual LAN (VLAN) technology. To efficiently use the network bandwidth, a monitoring and control mechanism using policy decision router (PDR) to dynamically adjust the bandwidth allocation for each service is adopted. For improving the network availability, Multi-Chassis Link Aggregation (MC-LAG) mechanism is introduced in the uplink route of PDR to prevent the disconnection problem due to a single node failure. In addition, by setting up multiple VPLSs (Virtual Private LAN Services) in the IP/MPLS core network and equipping with bidirectional forwarding detection (BFD) mechanism, we are able to promptly discover the network disconnection. And, through the application of both node-level and pw-level (pseudowire-level) protection schemes, a fast network recovery can be provided. To verify the applicability of proposed architecture, several experiments were conducted. Since the discovery of route disconnection in the IP/MPLS core network is not a trivial task and takes time, various routing mechanisms were tested. For instances, RIP needs 180 seconds and OSPF needs 30 seconds to discover a network disconnection. A detection mechanism is developed by combining static routing protocol and Bidirectional Forwarding Detection (BFD) mechanism. With this mechanism, a disconnection in the IP/MPLS core network could be found in one second. Besides, through the MC-LAG mechanism, the traffic could be redirected from inactive node to the standby node in two seconds. The simulation results show the proposed architecture is suitable for triple-play services.