The available
bandwidth on each link is also advertised using the TE-enhanced IGP (i.e., OSPF-TE
or ISIS-TE) and will be stored in the TED at each LSR. When an LSR runs CSPF, it
will only select a path that has sufficient bandwidth to allow the LSP to be established.
However, LSP reservations and IGP advertisements run asynchronously to each other,
and the LSP setup may still fail due to insufficient bandwidth at one of the LSRs in
the path as described above. In that case, the ingress LSR will run CSPF again to find
another path. The process of checking whether resources are available before attempting
to reserve them is known as Connection Admission Control (CAC) and required to
provide hard QoS.
LDP does not reserve bandwidth for LSPs, and control plane QoS is not applicable
for LDP-signalled LSPs.
Forwarding Plane QoS As mentioned previously, there are two QoS models for MPLS
LSPs:
?– E-LSP model In this model, QoS is inferred solely from the value of the EXP
bits in the MPLS label shim. Traffic from two LSPs with the same EXP value
will receive identical QoS treatment. This is an implementation of the soft QoS
model??”even if bandwidth is reserved per LSP in the control plane, there is no
434 Chapter 14
way to enforce this in the forwarding plane.
Pages:
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978