DiffServ-aware TE

DiffServ is a model that provides differentiated QoS guarantees based on class of service. MPLS TE is a traffic engineering solution that focuses on optimizing network resources allocation.

DiffServ-aware TE (DS-TE) combines DiffServ and TE to optimize network resources allocation on a per-service class basis. DS-TE defines different bandwidth constraints for class types. It maps each traffic class type to the CRLSP that is constraint-compliant for the class type.

The device supports the following DS-TE modes:

• Prestandard mode—Proprietary DS-TE mode of Hewlett Packard Enterprise.

• IETF mode—Complies with RFC 4124, RFC 4125, and RFC 4127.

• CT—Class Type. DS-TE allocates link bandwidth, implements constraint-based routing, and performs admission control on a per-class type basis. A given traffic flow belongs to the same CT on all links.

• BC—Bandwidth Constraint. BC restricts the bandwidth for one or more CTs.

• Bandwidth constraint model—Algorithm for implementing bandwidth constraints on different CTs. A BC model contains two factors, the maximum number of BCs (MaxBC) and the mappings between BCs and CTs. DS-TE supports two BC models, Russian Dolls Model (RDM) and Maximum Allocation Model (MAM).

• TE class—Defines a CT and a priority. The setup priority or holding priority of an MPLS TE tunnel for a CT must be the same as the priority of the TE class.

The prestandard and IETF modes of DS-TE have the following differences:

• The prestandard mode supports two CTs (CT 0 and CT 1), eight priorities, and a maximum of 16 TE classes. The IETF mode supports four CTs (CT 0 through CT 3), eight priorities, and a maximum of eight TE classes.

• The prestandard mode does not allow you to configure TE classes. The IETF mode allows for TE class configuration.

• The prestandard mode supports only RDM. The IETF mode supports both RDM and MAM.

• A device operating in prestandard mode cannot communicate with devices from some vendors. A device operating in IETF mode can communicate with devices from other vendors.

A device takes the following steps to establish an MPLS TE tunnel for a CT:

1. Determines the CT.

   A device classifies traffic according to your configuration:

   • When configuring a dynamic MPLS TE tunnel, you can use the mpls te bandwidth command on the tunnel interface to specify a CT for the traffic to be forwarded by the tunnel.

   • When configuring a static MPLS TE tunnel, you can use the bandwidth keyword to specify a CT for the traffic to be forwarded along the tunnel.

2. Verifies that bandwidth is enough for the CT.

   You can use the mpls te max-reservable-bandwidth command on an interface to configure the bandwidth constraints of the interface. The device determines whether the bandwidth is enough to establish an MPLS TE tunnel for the CT.

   The relation between BCs and CTs varies by BC model.

   • In RDM model, a BC constrains the total bandwidth of multiple CTs, as shown in Figure 27:

     • BC 2 is for CT 2. The total bandwidth for CT 2 cannot exceed BC 2.

     • BC 1 is for CT 2 and CT 1. The total bandwidth for CT 2 and CT 1 cannot exceed BC 1.

     • BC 0 is for CT 2, CT 1, and CT 0. The total bandwidth for CT 2, CT 1, and CT 0 cannot exceed BC 0. In this model, BC 0 equals the maximum reservable bandwidth of the link.

     In cooperation with priority preemption, the RDM model can also implement bandwidth isolation between CTs. RDM is suitable for networks where traffic is unstable and traffic bursts might occur.

     Figure 27: RDM bandwidth constraints model

     

   • In MAM model, a BC constrains the bandwidth for only one CT. This ensures bandwidth isolation among CTs no matter whether preemption is used or not. Compared with RDM, MAM is easier to configure. MAM is suitable for networks where traffic of each CT is stable and no traffic bursts occur. Figure 28 shows an example:

     • BC 0 is for CT 0. The bandwidth occupied by the traffic of CT 0 cannot exceed BC 0.

     • BC 1 is for CT 1. The bandwidth occupied by the traffic of CT 1 cannot exceed BC 1.

     • BC 2 is for CT 2. The bandwidth occupied by the traffic of CT 2 cannot exceed BC 2.

     • The total bandwidth occupied by CT 0, CT 1, and CT 2 cannot exceed the maximum reservable bandwidth.

     Figure 28: MAM bandwidth constraints model

     

3. Verifies that the CT and the LSP setup/holding priority match an existing TE class.

   An MPLS TE tunnel can be established for the CT only when the following conditions are met:

   • Every node along the tunnel has a TE class that matches the CT and the LSP setup priority.

   • Every node along the tunnel has a TE class that matches the CT and the LSP holding priority.