Search

In QoS routing, load balancing routing algorithm for long-lived traffic can be used efficiently to minimize the number of routing update, control traffic overhead and computational overhead which mainly come from network traffic fluctuation. The congestion problem from short-lived traffic is however difficult to handle. A simple and robust algorithm is needed for short-lived traffic to reduce the traffic loss probability. In the paper, we proposed a new hybrid routing schemes for short-lived traffic and long-lived traffic. An alternate routing algorithm is applied for the short-lived traffic when congestion occurs. We restricted the number of additional hops in the path to minimize the negative effect of alternate routing algorithm. The proposed algorithm reduces packet loss probability from 0.6% to 1.7% under various link utilizations. We also compare our proposed algorithms with and without crank-back.

For QoS control, traffic engineering (TE) of large IP backbone networks becomes a critical issue. However, provisioning network resource efficiently through TE is very difficult for ISPs because the traffic volume usually fluctuates widely over time. The congestion especially from short-lived traffic is difficult to handle due to its bursty arrival process. In the paper, we propose a detour routing schemes for short-lived traffic when congestion occurs. Our study shows that additional hops in detour paths should be carefully restricted to avoid network resource waste under heavy load. The proposed algorithm has less packet loss probability and less resource waste because we restricted the hop count by one or two.

Label Switched Paths (LSPs) in a MPLS network sometimes fail to deliver user traffic in the condition of the connection fault. In this case, control messages are exchanged among nodes, and the fault is notified. Nevertheless, there is a need that this fault such as traffic black holes or misrouting should be detected and isolated by the users in MPLS network. For this purpose, the ping and the traceroute especially for MPLS were proposed in the IETF. In this paper, we describe how this proposal is implemented in our test bed and the detail algorithm of this proposal is introduced. Finally, we show how our algorithm test is conducted in our network.