computed metric to the tunnel endpoint. Any changes to the IGP route metric to the tunnel endpoint is
automatically reflected on the autoroute destination route too.
In a typical VPN deployment over a TE core network, an operator creates a mesh of TE tunnels between PE
routers and then configures autoroute announce to these tunnels. This leads to a mix of default VRF and
VPNv4 traffic on the same tunnel connecting the PE routers. An operator my want to segregate their VPNv4
traffic on different tunnels. This can be achieved by creating multiple tunnels to the egress PE(s). The limitation
of this approach is that the static routes are added with zero metrics. The VRF Redirection to MPLS TE
Tunnels feature is a solution to resolve this limitation. Multiple VRFs can be mapped on the same tunnel by
adding multiple autoroute destination addresses (BGP next-hops) to the same tunnel.
Routes added by static route are always added with zero cost metric. This results in traffic that is mapped on
multiple tunnels to always load-balance due to ECMP. This may be undesirable when some of those tunnels
have sub-optimal paths (have higher underlying cost to the endpoint). With autoroute destination, only the
tunnel whose IGP cost to its endpoint is lowest will be considered for carrying traffic.
VRF redirection over TE tunnels feature supports:
•
Automatic redirection of VRF traffic over TE tunnels.
•
Multiple autoroute destinations under one tunnel to aggregate VRF traffic. If two VRFs are to be mapped
on same tunnel, then two autoroute destination prefixes (BGP next-hops) will be configured under the
tunnel.
•
One autoroute destination under multiple tunnels to enable ECMP load-balance of VRF traffic.
•
Implicit /32 mask for each route. Only host addresses residing on the tunnel endpoint are supported.
•
High availability, RP failover, and non-stop forwarding (NSF) scenarios by proving hitless to traffic
mechanisms.
MPLS TE Extended Admin Groups
The MPLS TE extended admin groups (EAG) configuration assigns EAG/AG name to bit-position and
associates affinity-names with TE links. The configuration extends to assign names, up to 256, to TE links
over the selected interface and assigns 32 names per attribute-set and index.
Use the
affinity-map map-name bit-position value
command to assign EAG/AG name to bit-position. Use
the
attribute-names attribute-name1 attribute-name2
... and
attribute-names index index-number
attribute-name1 attribute-name2
... commands to assign up to 32 names per attribute-set and index value.
Stateful Path Computation Element
The stateful path computation element (PCE) describes a set of procedures by which a path computation client
( PCC) can report and delegate control of head-end tunnels sourced from the PCC to a PCE peer. The PCE
peer can request the PCC to update and modify parameters of label switched paths (LSPs) it controls. The
stateful model also enables a PCC to allow the PCE to initiate computations allowing the PCE to perform
network-wide orchestration.
The transfer of LSP state and computation constraints is independent from the computation request, such that
a PCE may see how state changes over time, without a computation request ever taking place. This allows
the PCE to have better visibility into network state, as well as improve the efficiency of computation requests,
as these can rely on state present on the PCE.
Cisco IOS XR MPLS Configuration Guide for the Cisco CRS Router, Release 5.1.x
208
Implementing MPLS Traffic Engineering
MPLS TE Extended Admin Groups
