近年來在MPLS網路上的群播技術漸漸被重視，相關的協定與標準也陸續被提出，未來的網路將會是群播技術架構在MPLS網路上應用的時代，而針對MPLS上群播技術的故障回復機制卻沒有被相對地研究發展，使得群播技術一但發生了節點故障(Node Failure)或鏈路故障(Link Failure)之情形，也只能仰賴第三層的重新繞路(Layer 3 Rerouting)，會耗費大量時間在重覆做header的檢查，這對於需要即時傳輸的服務將是一大挑戰。 MPLS網絡上群播服務預先建立Backup Paths之故障回復機制可分為Global Restoration以及Local Restoration，Global Restoration為群播樹上之根節點與葉節點間任何一節點/鏈路之故障都使用相同之Backup Path，反之則為Local Restoration。 本論文提出一在MPLS骨幹網路上，為群播技術之節點故障/鏈路故障做MPLS 層的快速回復機制，並透過BR因子量化，以選擇在考量頻寬與故障回復時間下之最佳平衡點，其中BR因子之分子項為Backup Path之頻寬使用率，分母項為故障回復時間。經由模擬結果可以發所現提出之演算法經由BR因子量化後，可在現有之Local Restoration演算法中，取得最佳頻寬利用率－故障回復時間比。 Recently, the technology of multicast services in MPLS networks has been put importance gradually, and the related standards and researches have been proposed continually. In future network, applications of the multicast technology over MPLS networks architecture will be a mainstream, however, the fault recovery mechanism has not been developed relatively. Once nodes failure or links failure occurs, the technique of layer 3 rerouting is the only solution. Nevertheless, it takes too much time to repeat examining header, and this is an obstacle to real time services. The recovery model, which pre-establishes backup paths and be classified into global restoration and local restoration. In global restoration, recovery of node failures and link failures between the root and the leaf uses the same backup path. On the other hand, local restoration uses different backup paths. In this paper, we propose quick recovery mechanisms for node failures and link failures of multicasting services on the MPLS networks. In addition, through the transformation of BR factor, we select the balance point, which is faster and waste less bandwidth of pre-established backup paths. In BR factor, the denominator is the bandwidth utilization of backup paths, and the numerator of BR factor is recovery time. Simulation results show that our algorithms can obtain the best ratio between the bandwidth utilization and the recovery time in each local restoration algorithm.
