This feature combines traditional TE tunnels with the DiffServ model to provide QoS guarantee based on service types.
A static CR-LSP is easy to configure. Labels are manually allocated, and no signaling protocol is used to exchange control packets. The setup of a static CR-LSP consumes only a few resources, and you do not need to configure an IGP TE extension or CSPF for the static CR-LSP. However, static CR-LSP application is quite limited. A static CR-LSP cannot dynamically adapt to network changes and is limited in applications.
Traditional MPLS TE tunnels (non-standard DS-TE tunnels) cannot transmit services based on service types in compliance with the quality of service (QoS). For example, when a TE tunnel carries both voice and video flows, video flows may have more duplicate frames than voice flows. Therefore, video flows require higher drop precedence than the voice flows. The same drop precedence, however, is used for voice and video flows on MPLS TE tunnels.
To prevent services over a tunnel from interfere with each other, establish a tunnel for each type of service in a VPN instance or for each type of non-VPN service. This solution wastes resources because a large number of tunnels are established when many VPN instances carry various services.
In the preceding MPLS TE tunnel scenarios, the DS-TE tunnel solution is optimal. An edge node in a DS-TE domain classifies services and adds service type information in the EXP field in packets. A transit node merely checks the EXP field to select a proper PHB to forward packets.
A DS-TE tunnel classifies services and reserves resources for each type of services, which improves network resource use efficiency. A DS-TE tunnel carries a maximum of eight types of services.
Before configuring DS-TE, complete the following tasks: