Professional Documents
Culture Documents
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 3
• Control Label Advertisement—Change the label distribution mode from Ordered Control, which is
the default, to Independent Control. A mode change triggers an LDP re-init, which results in recycling
all sessions and flushing all learned label bindings. See Change the Label Distribution Mode on
page 66.
• Inbound Label Binding Filtering—Choose the labels that an LSR will accept from a neighbor. See
Filter Incoming Label Bindings on page 67.
• LDP Show and Debug Commands— Display bindings, discoveries, interfaces, neighbors, and
parameters using the commands under the show mpls ldp command tree. Display binding, error,
event, and routing events and address, hello, init, keepalive, label and notification messages using the
commands under the debug mpls ldp command tree. See LDP show and debug Commands on
page 69.
• Log Neighbor Changes— The LDP session initialization state machine consists of five possible
states. The system reaches operational state when a TCP connection is established, LDP parameters
have been negotiated, and keepalives are being periodically transmitted. Log a state change whenever
a neighbor transitions to or from operational state. See Manage and Monitor your LDP Configuration
on page 68.
• Clear Sessions and Counters—Clear LDP sessions for a specified neighbor or all neighbors using the
command clear mpls ldp neighbor {ip-address | *}. Clear LDP statistics for a specified neighbor or all
neighbors using the command clear mpls ldp statistics neighbor {ip-address | *}. See Clear Sessions
and Counters on page 69.
• MD5 Authentication—MD5 Authentication is a method that each LSR may use to authenticates its
peer LSRs. Each LSR stores the MD5 password, and only after password verification during the TCP
handshake is the TCP session established. See MD5 Authentication on page 67.
4 New Features
• Resource Class— A resource class is a group of resources, in this case, links, that have the same
characteristics (or “colors”). The characteristics that are significant are decided by the administrator.
OSPF-TE provides 32 possible resource classes. See Resource Class Attribute on page 28.
• Verbatim Path Support—You can configure multiple paths for an LSP with the tunnel mpls
path-option command. The paths are configured with sequence numbers so that the most preferred
path is selected. If the preferred path is an explicit path, but CSPF cannot find a path to the tunnel
destination using the explicit path list, the first hop in the explicit path list may be resolved using a
routing table lookup so that RSVP signaling can be performed. See Verbatim Path Support on page 34.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 5
• Tunnel Reoptimization Make-Before-Break—Make-before-break Tunnel Rerouting is a method of
rerouting traffic from one LSP to another without disrupting traffic by establishing the new LSP before
tearing down the original. While establishing the new LSP, the old and new LSP share resources for
the links they have in common. Reoptimization uses this technique to switch to an optimal route. See
Make-before-break Tunnel Rerouting on page 38.
• RSVP-TE Graceful Restart Helper Mode—When RSVP-TE Graceful Restart is enabled, Graceful
Restart Helper is also enabled. See RSVP-TE Graceful Restart Helper on page 38.
• RSVP-TE Graceful Restart Recovery Mode—The graceful restart extensions add the ability for an
ingress router to recover Path state and for ingress and transit nodes to recover objects they previously
transmitted. Among the recoverable objects is EXPLICIT_ROUTE and SESSION_ATTRIBUTES.
See RSVP-TE Graceful Restart on page 36.
• Restart and Recovery Time Intervals—Graceful Restart uses two timers. Restart Time is the amount
of time a recovering node requires to bring up RSVP communication after a failure. Recovery Time is
the period after communication is re-established during which the recovering node and neighbor
resynchronize Path state. See Enable RSVP-TE Graceful Restart on page 37.
• MPLS over LAG Interfaces—Link aggregation group (LAG) virtual interfaces support MPLS traffic
engineering tunnels.
• LAG Hashing on MPLS Packets—FTOS includes a default CAM profile and microcode that treats
MPLS packets as follows: when MPLS IP packets are received, hashing is based on the label + IP 5
tuple (source IP address, destination IP address, IP protocol, and source and destination UDP or TCP
port number). If the packet is part of a VPN, hashing is based on the inner and outer labels.
6 New Features
• Forwarding Adjacency Per-Tunnel Configuration—A tunnel forwarding adjacency instructs OSPF
or IS-IS to treat the tunnel as a link and as directly connected to the head-end LSR. The head-end IGP
includes the tunnel in its advertisements. While Autoroute is available to only the head-end of a tunnel,
Forwarding Adjacency makes tunnels available to all area routers. The tunnel cost is the same as the
IGP calculated cost for the link. See Enable IGP Forwarding Adjacency on page 91.
• Forwarding Adjacency Hold Timer to Delay Tunnel Down Advertisements— Forwarding
adjacency hold time is the amount of time the local IGP waits before advertising a local tunnel-down
event to IGP neighbors. Delaying the advertisements avoids frequent updates in all routers in an area in
case of tunnel flapping. See Enable IGP Forwarding Adjacency on page 91.
• Autoroute Per-Tunnel Configuration—Autoroute enables OSPF or IS-IS on the head-end to use a
tunnel to reach a destination. The tunnel destination becomes the next hop for its prefix. All traffic
bound for the tunnel destination is routed through the tunnel. The tunnel may also be used for traffic
bound for a destination topologically behind the it if the cost of using it is the lowest compared to other
IGP routes. See Enable IGP Autoroute on page 89.
• Autoroute Absolute, Relative, and Explicit Metrics—The cost of using a tunnel to reach prefixes
beyond it is the tunnel cost plus the IGP link cost from the tunnel destination to the ultimate
destination. You can assign one of three types of metrics to a tunnel. Metric is used to reach the tunnel
destination; beyond the tunnel destination, the IGP metric is used. Absolute Metric is used to reach the
tunnel destination and any destination topologically behind the tunnel. Relative Metric is added or
subtracted from the IGP metric to the tunnel destination; beyond the tunnel destination, the IGP metric
is used. See Enable IGP Autoroute on page 89.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 7
• No-Ack Option—One of the refresh reduction extensions is an acknowledgement object. Each
message sent the ACK_Desired flag is set must be acknowledged using this object. You can choose to
not set this flag in Resv messages; received messages with the flag set are still acknowledged. See
Enable Signaling Refresh Reduction on page 79.
• Configurable Message Refresh Interval—RSVP stores reservations as soft state, which must be
periodically refreshed through the receipt of a matching path or reservation message. The default
refresh period (R) is 30 seconds; unrefreshed state is deleted after the local state lifetime (L) expires.
See Specify the Signaling Refresh Interval on page 79.
• Configurable Refresh Limit—Specify the number of hello packets that must be unacknowledged to
consider the neighbor down. See Enable RSVP Hello Signaling on page 77.
• Graceful-Shutdown—The RSVP graceful shutdown feature provides a user-initiated approach to
shutting down RSVP. By default, graceful shutdown is disabled since a large number of RSVP
sessions may take a long time to shut down, and a session timeout on neighbor routers may be
preferred. See Graceful Shutdown on page 77.
• TTL Check—The TTL Check feature enables checking the time-to-live (TTL) field in the header of
RSVP and IP packets. When enabled, only PATH messages are dropped if the TTL check fails. See
Enable TTL Checking on page 81.
• Bandwidth Holding with RSVP Path Messages—When an LSR receives a Path message indicating
a bandwidth requirement, it earmarks the bandwidth on the interface until a Reservation message is
received. You can configure the amount of time the LSR holds the bandwidth before releasing it for
other reservations in case a reservation is unacceptably delayed or never arrives. See Bandwidth
Holding for Path Messages on page 81.
Tunnel Management
• Tunnel Attributes: Bandwidth—Specify a tunnel bandwidth requirement. If none is configured then
the tunnel is assumed to have no specific bandwidth requirement.
• Tunnel Attributes: Multiple Path Options with Priority—Configure an ordered set of
traffic-engineering options, including the verbatim path option to skip CSPF computation. Path options
with the lowest sequence numbers are preferred.
• Tunnel Attributes: Record Route—Include the RECORD_ROUTE (RRO) object in Path messages
and Resv messages, which collects a list of nodes in the path to the tunnel destination. See Enable
Record Route on page 77.
• Tunnel Attributes: Setup and Hold Priorities—Setup and hold priorities are used when determining
whether a particular tunnel can receive or hold on to a bandwidth reservation. See Session Attributes
on page 28.
• Retry Interval—The retry interval sets the time, in seconds, between attempts to establish an LSP.
This feature is useful in cases when the primary LSP of a tunnel cannot be established; the primary
LSP is up, but its corresponding standby LSP is not up; or a make-before-break LSP is attempted due
to a resource change or reoptimization to determine the time interval between retries. See LSP Retry
Interval on page 51.
• Pre-Signaled Standby LSP Support—Every path option can be configured with a standby path
option that will be signaled when the primary LSP is established. If the primary goes down due to a
RSVP signaling failure, traffic fails over to standby. See Path Protection on page 49.
8 New Features
MPLS QoS
• MPLS-EXP Marking using Class Maps—create a class map match against the incoming DSCP
value and map the value to the MPLS-EXP bit for MPLS marking. See MPLS-EXP Marking using
Class Maps on page 71.
• MPLS-EXP Marking using QoS Policies—create a QoS policy to match against the incoming DSCP
value and map the value to the MPLS-EXP bit for MPLS marking. See MPLS-EXP Marking using
Qos Policies on page 72.
• MPLS-EXP Propagation with PHP—By default, MPLS-EXP is propagated from the top label to the
next label or to the DSCP value. Disable this behavior by entering no mpls ip propagate-exp from
CONFIGURATION mode. See Propagating MPLS-EXP when Penultimate Hop Popping on page 72.
• Explicit Null—The Explicit-Null label alerts the Egress LSR that it is the egress so that it can
immediately remove the top label, a process called Ultimate-Hop Popping. Ultimate-hop popping
ensures that any packets traversing an MPLS network include a label so that Class of Service is
consistent across the entire LSP. See Advertising Explicit-Null on page 73.
• Short-Pipe and Uniform Models—FTOS supports both the short-pipe model and the uniform model
for differentiated services with MPLS. The uniform model is the default; the IP packet DSCP value is
propagated to the MPLS experimental bits field. Optionally, the set-dscp command in an input policy
map can be used to mark the EXP bit field. Setting the DSCP value also sets the EXP bit value; there is
no separate command to set only the EXP bit. See Quality of Service on page 22.
MPLS ECMP
• 16 ECMPs with MPLS—The FTOS implementation of MPLS ECMP requires that the ECMP
number be a power of 2. For example, if 5 ECMPs are available, 8 ECMPs are written into hardware -
12345123. To ensure an even distribution of traffic, a bit mask, based on the number of ECMP entries,
is used for CAM entry matching. See MPLS ECMP on page 9.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 9
• No Propagate TTL—Disabling TTL propagation changes how ingress and egress LSR nodes read
and process the TTL value in a label. A label must have a value in the TTL field. By default, an ingress
LSR reads the TTL field in the IP header of the incoming packet, decrements it by 1, and copies what
is left into the TTL field of the MPLS header. Core LSRs forward the packet if the TTL value in the
uppermost label is not 0. With the no mpls ip propagate-ttl command, the behavior changes such that
the IP header TTL does not reflect the hops taken across the MPLS core; and the routers in the MPLS
core network do not appear in the traceroute information. See Disable TTL Propagation on page 104.
• LSR MIB—The LSR MIB, based on RFC 3813, MPLS-LSR-STD-MIB, provides information on the
status, character, and performance of MPLS-capable interfaces on the LSR, incoming MPLS segments
(labels) at an LSR and their associated parameters, and outgoing segments at an LSR and their
associated parameters. See Label Switch Router MIB on page 105.
• TE MIB—The Traffic Engineering (TE) MIB, defined in 3812, provides SNMP OIDs equivalent to
the fields in show mpls traffic-eng tunnels. Use the TE MIB to view display status at the head-end,
information on traffic flows on the tunnels, TE parameters, and MPLS TE tunnel routes, including the
configured route, the Interior Gateway Protocol (IGP) calculated route, and the actual route traversed.
The mplsTunnelTable, mplsTunnelResourceTable, mplsTunnelHopTable, mplsTunnelARHopTable,
and mplsTunnelCHopTable tables are supported. See Traffic Engineering MIB on page 110.
10 New Features
Contents
New Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Chapter 1
Multiprotocol Label Switching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Chapter 2
MPLS CAM Configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Chapter 3
Constrained Shortest Path First . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Explicit Routes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 11
Exlude an Address from an Explicit Route . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Session Attributes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Resource Class Attribute . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Resource Class Affinity Ingress Check . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Disable Resource Affinity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Setup and Holding Priority . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Equal Cost Path Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Administrative Weight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Verbatim Path Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Chapter 4
MPLS High Availability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Chapter 5
Label Distribution Protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Creating Labels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
When do LSRs Create Labels? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Label Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Label Advertisement Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Label Distribution Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Label Retention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
LDP Peering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
LDP Protocol Data Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Peering Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Notification Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Label Distribution Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
12 Contents
Reserved Labels—Penultimate Hop Popping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Implementation Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Configure Label Distribution Protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Related Configuration Tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Enable LDP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Change the LSR Router ID . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Change the Holdtimes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Change the Label Distribution Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Direct the Flow of Label Advertisement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Filter Incoming Label Bindings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
MD5 Authentication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Manage and Monitor your LDP Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Log Neighbor Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Clear Sessions and Counters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
LDP show and debug Commands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Chapter 6
MPLS Quality of Service . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Chapter 7
RSVP-TE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Implementation Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Configure RSVP-TE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Related Configuration Tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Enable RSVP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Specify the Reservable Amount of Interface Bandwidth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Enable Bandwidth Logging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
Graceful Shutdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Enable Record Route . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Enable RSVP Hello Signaling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Specify the Signaling Refresh Interval . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Enable Signaling Refresh Reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Bandwidth Holding for Path Messages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Enable TTL Checking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Chapter 8
MPLS Traffic Engineering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Create a Tunnel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Allocate Bandwidth to a Tunnel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 13
OSPF—Traffic Engineering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
Enable MPLS-TE in an OSPF Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Configure Bandwidth Usage Advertisement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
IS-IS—Traffic Engineering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Route Traffic though Tunnels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Map Traffic to Tunnels using Static Routes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Enable IGP Autoroute . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Enable IGP Forwarding Adjacency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Chapter 9
MPLS System Management and Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Implementation Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
MPLS ping and traceroute . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
traffic-eng ping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
traffic-eng traceroute . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
ldp ping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
ldp traceroute . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
FTOS-supported-Error Codes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Disable TTL Propagation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Chapter 10
MPLS MIBs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Chapter 11
FTOS-supported RFCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
14 Contents
Chapter 1 Multiprotocol Label Switching
Multiprotocol Label Switching (MPLS) is a set of protocols that together combine Layer 3 routing with
Layer 2 switching according to the architecture defined in RFC 3031. Multiprotocol Label Switching
(MPLS) directs and carries data from one network node to the next, and makes it easy to create virtual
links between distant nodes.
MPLS is independent of Layer 2 and :Layer 3 technologies, so it allows integration of networks with
different Layer 2 and Layer 3 protocols
In conventional IP routing, each router in the path from source to destination determines the next hop by
matching the destination IP address to the longest matching address-prefix in the routing table. Any packet
that matches the same entry in the routing table takes the same next-hop, and in this way, the lookup can be
considered a type of forwarding classification.
In an MPLS network, classification may be based on any information in the Layer 3 header, for example,
IP Precedence, or information not contained in the packet, for example, ingress port; classification is not
limited to the results of the routing table lookup. Classification is only performed once, as the packet enters
the network, by the ingress MPLS router, which marks (labels) each packet with its class; downstream
routers then determine the next-hop based on the MPLS label.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 15
Unlike IP, the MPLS packet classification/label can be based on:
Thus, the Forwarding Equivalence Class (FEC) can represent any of the following:
MPLS works within the core of an IP network. The core routers have MPLS enabled on interfaces. Labels
are added to IP packets when entering MPLS network, and removed from IP packets when leaving an
MPLS network.
Provider IP Network
Customer IP Customer IP
Network Running MPLS Network
Forwarding packets based on labels rather than headers results in several important advantages:
• When a packet is assigned a Forwarding Equivalence Class (FEC) when it enters the network,
information that is not taken from the network layer header can be used for FEC assignment. For
example, classification of packets based on the source of the packets.
• Packets can be assigned a priority label, making Frame Relay and ATM-like quality-of-service
guarantees possible.
• The considerations that determine how a packet is assigned to an FEC can become very complex,
without impacting routers that merely forward labeled packets.
• Packet payloads are not examined by the forwarding routers which allows for different levels of traffic
encryption and the transport of multiple protocols.
• A packet can be forced to follow an explicit route rather than the route chosen by normal dynamic
algorithm as the packet travels through the network.
A packet may need to cross several different nested MPLS domains. A nested domain is an MPLS domain
included in another MPLS domain. In this case, the MPLS headers may be stacked, so that there is more
than one 32-bit header in the packet header.
• The Label field is 20 bits and carries the actual label value. Label numbering follows these rules:
• 0 - IPv4 explicit NULL label
• 1 - Router alert label
• 2 - IPv6 explicit NULL label
• 3 - Implicit NULL label
• 4-15 - Reserved for future use
• 15 - 1, 048,575 are assigned by the LSR
• The Exp (Experimental) field is 3 bits, and is used to identify traffic classes or congestion. This field is
used for QoS.
• The S (Stacking) portion is 1 bit, and is used when the MPLS headers are stacked to support multiple
header within the packet.
• 1 - indicates that this is the last MPLS header in the packet
• 0 - identifies the header as a stacked MPLS header
• The TTL (Time To Live) portion is 8 bits and shows the remaining number of hops left. This is the
same as the standard IP datagram TTL identifier.
TTL handling
In IP networks, the TTL field is used to prevent packets from travelling indefinitely in the network, and so
preventing loops due to mis-configuration or temporary convergence issues.
MPLS uses the same mechanism as IP, but not on all encapsulations:
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 17
• The TTL is present in the label header for PPP and LAN headers as part of the shim header.
• When a packet moves through through one or more LSPs, it will have the same TTL as if it was
forwarded without MPLS.
On E-Series ExaScale, tunnel packets are fragmented based on physical port MTU. You must ensure that
the link MTU is sufficiently greater than the IP MTU to accommodate Ethernet and MPLS headers.
In the event of fragmentation, tunnel interface counters do not reflect the actual number of packets sent
over the tunnel. Instead, the number of packets shown in the "Output statistics" field reflect only the
number of first fragments.
MPLS components
MPLS works within the core networks. The key components to that ability are described here and
illustrated in Figure 3.
MPLS labels
A label defines the path through an MPLS network to reach a destination based on one or more criteria.A
set of parameters that determine the next-hop of a packet is called a Forwarding Equivalence Class (FEC).
A label is a (in most cases) unique, arbitrary value that represents an FEC and is inserted into a packet. A
label-FEC pair is called a binding; bindings are local to adjacent MPLS routers (LSR) and are exchanged
using a label distribution protocol; Label Distribution Protocol (LDP) is available on FTOS. Labels may be
added or removed in transit from ingress MPLS Label Switch Router (LSR) to egress MPLS LSR, so
packets may have multiple labels.
Label stacking
A labeled packet can carry one or more labels. These labels are organized on a last-in-first-out (LIFO)
basis.
If a packet consists of more than 1 label, the label at the bottom of the stack is referred to as level 1, the
next label is level 2, and so on. When a packet arrives at an LSR, the processing is based on the top-most
label and the underlying labels are not considered.
Label merging
A label can be merged or aggregated. For example, an LSR can have multiple incoming labels for a
particular FEC. When forwarding packets from that FEC the LSR can use a single outgoing label
An LSR that perform this type of merging is identified as a merge capable LSR. The merge-capable LSR is
is one that can receive two packets with two different labels, two packets with same label but from two
different interfaces, and send both packets out the same outgoing interface with same label.
When label merge is used, the number of incoming labels that an LSR needs is never greater than the
number of adjacencies.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 19
Without label merge, the number of incoming labels that an LSR needs is as large as the number of
upstream routers, which forward traffic from this FEC to this LSR.
Note: A label is in most cases globally unique. However, a system may bind the same label to more than
one FEC when FECs are used in different applications, or contexts. The context in which the label is used
is called a label space, and is represented by a bit value concatenated with the LDP Identifier for label
distribution. There are two types of label spaces, per-interface and per-platform.
• A per-interface label space exists when an LSR binds a label to more than one FEC and distributes
each binding to a different system connected to it by a point-to-point link. Because of the point-to-point
connection, the LSR can discern which FEC a label represents based on the ingress interface.
• A per-platform label space exists when all of the binding created by the system are unique.
• Incoming Label Map (ILM)—used for labeled packets, maps an incoming label to a Next Hop Label
Forwarding Entry (NHLFE).
• FEC-to-NHLFE Map (FTN)—used for unlabeled packets, maps an FEC to an NHLFE.
The NHLFE contains, among other information, the next hop and an action that the LSR must perform on
the the label stack. The actions can be:
• replace the top label (with a new label that represents an FEC that has the same egress LSR as the
replaced label)
• remove (pop) the top label
• replace the top label and insert (push) an additional label
When a labeled packet arrives at the LSP ingress LSR, it performs the following actions:
When an unlabeled packet arrives at the LSP ingress LSR, it performs the following actions:
Next-hop Label
FEC-to-NHLFE Forwarding Entry
Unlabeled
Map (FTN): label (NHLFE): next-hop,
unlabeled Packets action (pop,replace,
or replace and push)
Incoming Label
Labeled Map (ILM): maps
labels to NHLFE
The FEC LSP can use hop-by-hop routing, or explicit routing. Explicit routing can be loose or strict.
• A hop-by-hop routed LSP is created by allowing each LSR to select the next hop.
• An explicitly routed LSP is when the ingress or egress LSR chooses some or all of the LSRs in the
LSP.
• If some of the LSRs are chosen the route is loosely explicit.
• If all of the LSRs are chosen the route is strictly explicit.
FTOS supports MPLS with the Default CAM-profile and Microdcode only.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 21
High Availability
Force10 Networks supports multiple MPLS High Availability (HA) features:
Refer to Chapter 4, MPLS High Availability for complete information regarding MPLS HA.
Refer to Chapter 2, Label Distribution Protocol for complete information regarding LDP.
RSVP uses the IGP to determine paths. Extended RSVP uses label extensions to support establishing and
maintaining LSPs.
Quality of Service
MPLS Quality of Service (QoS) does not define a new QoS architecture. It leverages the IP DiffServ QoS
architecture and applies it to the MPLS network. MPLS uses the EXP field in the header rather than the
DCSP field used by IP.
FTOS supports DSCP in IP networks and EXP in MPLS networks. When DSCP is configured , EXP is
also enabled and takes its value from the DSCP settings. If DCSP is not enabled, EXP is not enabled.
The EXP field is 3 bits in the MPLS header while the DSCP field is 6 bits in the IP header. This difference
is managed in two ways, and both can use either LDP or RSVP to signal and distribute labels.
There are three tunneling models: Uniform, Pipe, and Short Pipe. FTOS supports the short-pipe and
uniform models.
In the pipe and short pipe models, any traffic conditioning that is applied when traffic goes through the
tunnel has no effect on the EXP bits coding in the inner header. In other words, when traffic exits an LSP
(when a label is popped) or when traffic enters an LSP, the inner header's EXP bits coding is not changed.
The pipe and short-pipe models differ in the header that the tunnel egress uses when it determines the PHB
of an incoming packet. With the short-pipe model, the tunnel egress uses an inner header that is used for
forwarding. With the pipe model, the outermost label is always used.
The uniform model is the default FTOS behavior; the IP packet's DSCP value is propagated to the MPLS
experimental bit field.
Optionally, the set-dscp command in an input policy map can be used to mark the EXP bit field. Setting
the DSCP value also sets the EXP bit value; no separate command exists to set only the EXP bit. Also,
Marking is supported only for IP Packet DSCP to EXP and not for EXP to EXP.
Traffic Engineering
MPLS Traffic Engineering (TE) provides the means to direct and manage the traffic between LSRs. MPLS
tunnels can be routed around network congestion or failures. Resource affinities can be defined to include
or exclude specific links during path selection. Explicit paths can be defined for LSPs.
Refer to Chapter 8, MPLS Traffic Engineering for complete information regarding TE.
ECMP
FTOS supports 16 ECMPs with MPLS. The FTOS implementation of MPLS ECMP requires that the
ECMP number be a power of 2. For example, if 5 ECMPs are available, 8 ECMPs are written into
hardware - 12345123. Thus, the first 3 paths are written into hardware two times and are therefore more
likely to be used, potentially resulting in an uneven distribution of traffic.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 23
We program the mask in the CAM for lookup. The mask is based on the number of ECMP entries. If there
are 2 ECMP entries, then the mask is 1 bit. If there are 16 ECMP entries, then the mask is 4 bits. So if the
mask is 4 bits, in the last 4 bits of the destination IP address (host portion) are used for matching to hit the
appropriate CAM entry. For example if the mask is 1 bit, with 2 CAM entries programmed as follows:
If an IP packet destined for 10.0.0.1 (with an in-label that matches the in-label of the above ILM entries),
this packet hits CAM entry 2 since the last bit of the address is 1. Similarly, an IP packet destined to IP
10.0.0.4 hits CAM entry 1 since the last bit is 0.
• LAG hashing on ingress and transit routers—second label from the top + 5 tuple on a label stack
packet
• ECMP hashing for LDP on transit routers—label + IP Destination Address (last few bits)
• LDP with RSVP on ingress router—5-tuple based
You can modify the number of ILM entries in the MPLS ILM table using the command mpls ilm.
Number of Next-Hop and First-Hop entries are fixed and based on the line card CAM type. When the
CAM reaches 90% usage, FTOS displays a Syslog message.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 25
Display the MPLS CAM region using the command show cam-usage from EXEC Privilege mode.
Force10#show cam-usage
Linecard|Portpipe| CAM Partition | Total CAM | Used CAM |Available CAM
========|========|=================|=============|=============|==============
8 | 0 | IN-L2 ACL | 5120 | 0 | 5120
| | IN-L2PT | 266 | 0 | 266
| | IN-L2-SysFlow | 102 | 6 | 96
| | IN-L2-FRRP | 102 | 0 | 102
| | IN-L2-Qos | 500 | 0 | 500
| | IN-L2 FIB | 16384 | 1129 | 15255
| | IN-L3 ACL | 16384 | 1 | 16383
| | IN-L3 FIB | 524285 | 9 | 524276
| | IN-L3-SysFlow | 4096 | 35 | 4061
| | IN-L3-McastFib | 9216 | 0 | 9216
| | IN-L3-Qos | 10240 | 0 | 10240
| | IN-L3-PBR | 1024 | 0 | 1024
| | IN-MPLS-ILM | 61440 | 0 | 61440
| | IN-V6 ACL | 6144 | 0 | 6144
| | IN-V6 FIB | 12287 | 0 | 12287
| | IN-V6-SysFlow | 2048 | 0 | 2048
| | IN-V6-McastFib | 3072 | 4 | 3068
| | OUT-L2 ACL | 1729 | 0 | 1729
| | OUT-L2PT | 256 | 0 | 256
| | OUT-L2 SysFlow | 63 | 2 | 61
| | OUT-L3 ACL | 4096 | 0 | 4096
| | OUT-V6 ACL | 1024 | 1 | 1023
...
The Constrained Shortest Path First (CSPF) algorithm is an advanced form of SPF. The path
determination process in CSPF is not designed to find the best path to all routers, but rather only to a
particular router (the tunnel destination). CSPF is used in computing paths that are subject to multiple
constraints and not just the distance metric. When computing paths for Label Switched Paths (LSPs),
CSPF considers not only the topology of the network, but also the attributes of the LSP and the links, and it
attempts to minimize congestion by intelligently balancing the network load.
CSPF maintains the Traffic Engineering Database (TED). The TED is populated by the traffic engineering
related route/link updates from OSPF-TE and ISIS-TE. This consolidated TED is queried by RSVP-TE
when establishing LSPs. CSPF resolves the Quality of Service routing queries, finds the best route that
meets the specified constraints, such as a specified minimum bandwidth, priority, and number of hops.
The following features are configurable constraints which CSPF may use for path selection:
Explicit Routes
You can define some or all of the nodes in the LSP before the path message is sent. In this case, the LSP is
called an explicit route. Explict routes are used to optimize network resource utilization and reroute around
failures and congestion. The nodes in the explict path are specified in the EXPLICT_ROUTE object
(ERO), which is included in Path messages.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 27
• A strict node must be directly connected to the previous hop in the explicit path.
• A loose node may or may not be directly connected, and the path between the loose node the previous
node may be computed by the IGP.
1 Give the LSP a name. This command moves ip explicit-path name name CONFIGURATION
you to the EXPLICIT PATH context.
2 Specify some or all of the nodes in the path, [seq number] next-address EXPLICIT PATH
one by one, in downstream order using ip-address [strict | loose]
sequence numbers. Make the node a strict or Default: Strict
loose hop.
Session Attributes
Attributes are topology state parameters that are used to constrain path selection to specific resources.
Session attributes are contained in the SESSION_ATTRIBUTES object.
Object Header Exclude Include Include Setup Holding Flags Name Session Name
Any Any All Priority Priority Length
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 29
Step Task Command Syntax Command Mode
Force10(conf)#inter tun 1
Force10(conf-if-tu-1)#tunnel destination 100.1.1.4
Force10(conf-if-tu-1)#tunnel mode mpls
Force10(conf-if-tu-1)#tunnel mpls traffic-eng path-option 10 dynamic
Force10(conf-if-tu-1)#tunnel mpls traffic-eng admin-group include premium
Force10(conf-if-tu-1)#tunnel mpls traffic-eng admin-group exclude leased
Force10(conf-if-tu-1)#no shut
Force10(conf-if-tu-1)#show conf
!
interface Tunnel 1
tunnel destination 100.1.1.4
tunnel mode mpls
tunnel mpls traffic-eng path-option 10 dynamic
tunnel mpls traffic-eng admin-group exclude leased
tunnel mpls traffic-eng admin-group include premium
no shutdown
mp-e600i-1(conf-if-tu-1)#do show mpls traffic-eng tunnels tunnel 1
Config Parameters:
Configured path options:
path option 10, type dynamic
Tunnel level config:
Bandwidth: 0 kbps (Global) Priority: 7 7
Include admin-group: premium
Exclude admin-group: leased
Retry Timer: 30 seconds
Fast reroute: disabled
[output omitted]
General Parameters::
Bandwidth Hold time: 15 secs
GigabitEthernet 0/12 :
Link connected to multiaccess network
Link ID : 192.168.20.2
Interface Address : 192.168.20.1
Admin Metric te: 1 igp: 1
Maximum bandwidth : 125000000
Maximum reservable bandwidth : 93750000
Number of Priority : 8
Priority 0 : 93750000 Priority 1 : 93750000
Priority 2 : 68750000 Priority 3 : 68750000
Priority 4 : 68750000 Priority 5 : 68750000
Priority 6 : 68750000 Priority 7 : 68750000
Affinity Bit : 0x21
[output omitted]
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 31
Step Task Command Syntax Command Mode
Send RSVP messages without the mpls traffic-eng affinity send-non-ra INTERFACE
affinity attribute. TUNNEL
Configure tunnel traffic-engineering tunnel mpls traffic-eng setup-priority spriority EXPLICIT PATH
setup and hold priorities. These are hold-priority hpriority
used when determining whether a Default (spriority and hpriority): 7. 0 is the
particular tunnel can receive or hold on highest priority.
to a bandwidth reservation.
Chose the criterion for selecting one of mpls traffic-eng path-selection CONFIGURATION
multiple equal cost paths to a particular hop-count
node.
Administrative Weight
Adminstrative Weight, also called TE metric, is an administratively assigned metric for an interface
participating in IGP traffic engineering. It is advertised by IGPs and overrides the IGP cost for CSPF
computation.
FTOS always advertises the TE metric; it does not support the IGP metric. So, if no TE metric is
configured, then the advertised TE metric value is equal to the IGP cost. For example, if the IGP cost is 10
and no TE metric is configured, then the advertised TE metric value is 10. If the IGP cost is 10 and a TE
metric of 5 (or any other value) is configured, the advertised TE metric value is 5.
When there are multiple equal cost paths to reach the next-hop, by default, path with highest available
bandwidth is selected. If a tie still exists, a link is selected at random. Optionally, you can configure the
system to select the path with the lowest hop count. If there is a tie break in this case, a link is selected at
random.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 33
Verbatim Path Support
You can configure multiple paths for an LSP with the tunnel mpls path-option command. The paths are
configured with sequence numbers so that the most preferred path is selected. If the preferred path is an
explicit path, but CSPF cannot find a path to the tunnel destination using the explicit path list, the first hop
in the explicit path list may be resolved using a routing table lookup so that RSVP signaling can be
performed.
Enable Verbatim Path Support for a path tunnel mpls traffic-eng path-option EXPLICIT PATH
option. [protect] path-num explicit name
path-name verbatim [lockdown]
[bandwidth bandwidth]
Nodes include RESTART_CAP in hello messages to advertise restart capability. The object contains two
values:
• Restart Time—the amount of time the node requires to bring up RSVP communication after a failure.
• Recovery Time—the period after communication is re-established during which the node and
neighbor resynchronize.
Neighbors that receive a hello with RESTART_CAP store these parameters. If communication with a
remote node is lost, the local node waits an amount of time equal to Restart Time, as if it is still receiving
periodic refreshes. During this time, the local node preserves all RSVP forwarding state for LSPs that
include the link between the local node and the neighbor, but does not send refresh messages. Neighbors
continue to send hello messages to the recovering router.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 35
When hellos are re-established, the local node examines the hello instance from the neighbor. If the
instance value is the same as before communication was lost, the failure was limited to the control channel.
If the instance is different, the local node can assume that the neighbor reset. In either case, the Recovery
Time is used to refresh Path state in the failed node.
In the case of a nodal fault, the neighbor of the recovering node refreshes all Path state shared with the
neighbor using Path messages. The Path messages carry the RECOVERY_LABEL object, instead of a
LABEL object; it is the same as a LABEL, but indicates to the recovering node that it has existing state for
the label. RECOVERY_LABEL contains the same label value as the most recently received Resv message.
When the recovering node receives a Path message during the recovery period, the node: 1) looks for
matching RSVP state entries and refreshes them, or 2) looks for matching forwarding state and installs
RSVP state for them. If no matching state is found, the Path message is treated as setup for a new LSP.
When sending the corresponding outgoing Path message downstream, the message includes the
SUGGESTED_LABEL object, rather than LABEL, indicating that the recovering node used the label
before the failure.
For a control channel failure, no recovery procedure is executed, and though the upstream router sends
RECOVERY_LABEL objects, they are not processed.
The graceful restart extensions create one new message and one new object:
• RecoveryPath message—this message has the same structure as Path messages, but carries upstream
objects previously sent by the recovering neighbor.
• CAPABILITY object—Carried in hello messages, this object indicates to neighbors that the node is
graceful-restart capable.
After RSVP communication is reestablished, the downstream neighbor sends one RecoveryPath message
for every LSP for which it sent a Resv message. The objects in the message are copied from the last
received Path message for the LSP.
When the recovering node receives a RecoveryPath message, it uses the information in the message to
locate an existing forwarding state entry, or if no forwarding state is found, setup a new LSP based on the
contents of both messages. The downstream data interface and corresponding label and all other
information in the RecoveryPath message are reused when setting up the LSP.
Enable RSVP-TE Graceful Restart. mpls rsvp signaling hello graceful-restart CONFIGURATION
When enabled, the system includes enable
the RESTART_CAP object in hellos. Default: Disabled
(OPTIONAL) Set the Restart Time mpls rsvp signaling hello graceful-restart CONFIGURATION
and Recovery Time. When the {restart-time interval | recovery-time interval}
system is not recovering from a Default Restart Time: 60000 milliseconds
failure, it sets the Recovery Time to 0
in its hellos. It sends hellos with a Default Recovery Time: 120000 milliseconds
Recovery Time to a value greater
than 0 only when the node is in
recovery mode.
Display your graceful restart configuration using the command show mpls rsvp hello graceful-restart:
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 37
RSVP-TE Graceful Restart Helper
When graceful restart is enabled, the system is restart-capable and a graceful-restart helper.
Tunnel Reoptimization
Loosely routed LSPs are those that are defined entirely by the IGP, or explicitly routed paths with one or
more loose hops. There are two types of nodes in an explicit route:
• A strict node is directly connected to the previous hop in the explicit path.
• A loose node might or might not be directly connected, but the path between the node and the previous
strict node is computed by the IGP.
Whenever a path or a portion of a path is computed by the IGP, there is an opportunity to discover better
routes—better meaning lower cost—that are created due to network topology changes.
Reoptimize all tunnels or a specific mpls traffic-eng reoptimize [all | tunnel EXEC Privilege
tunnel on demand. number]
An explicit LSP that has a backup LSP is a protected LSP. You can protect an LSP from link failure, node
failure, or bandwidth exhaustion. More than one backup tunnel may be configured for an interface.
Enable RSVP-TE Fast Reroute for a tunnel. tunnel mpls traffic-eng fast-reroute INTERFACE
Optionally set the bandwidth protection desired [bw-protect || node-protect] TUNNEL
and/or node protection desired flag.
FTOS Behavior: MPLS TTL, IP TTL, MPLS EXP, and IP DSCP may change when FRR is triggered. If
FRR is triggered on the topology in Figure 9 for example, and IP traffic with IP TTL = 100 and IP DSCP
= 011000 is sent to R1:
On egress of R1: IP to MPLS TTL copy
• MPLS label TTL = 99
• MPLS EXP = 3 (011)
On egress of R2:
• MPLS label TTL = 98
• MPLS EXP = 3 (011)
• FRR Push Label TTL = 254 (FRR push TTL is 254 by default)
• FRR Push Label EXP = 0 (FRR push EXP is 0 by default)
On egress of R5: PHP of backup tunnel
• FRR push label TTL is copied to MPLS Label TTL: MPLS label TTL = 253
• FRR push label EXP is copied to MPLS Label EXP: MPLS EXP = 0 (000)
On egress of R3: PHP of primary tunnel
• MPLS Label TTL is copied to IP TTL: IP TTL = 252
• MPLS Label EXP is copied to IP DSCP: IP DSCP = 000000
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 39
Protect an LSP from Link Failure
In the Figure 9, the R2 is the Point of Local Repair (PLR), and R3 is the Merge Point.
Gig 0/18
Gig 0/60
R5
100.1.1.1 100.1.1.4
Gig 0/5
Tunnel 2
100.1.1.5
5 Verify that the backup path status show mpls traffic-eng fast-reroute database
is Ready.
Force10#show mpls traffic-eng fast-reroute database
Tunnel head end item frr information:
Protected tunnel In-label Out intf/label FRR intf/label Status
---------------- -------- -------------- -------------- ------
6 Verify that the backup tunnel type show mpls rsvp fast-reroute
is Nhop, which means that the
LSP is protected from a link
failure.
Force10#show mpls rsvp fast-reroute
Primary Protect BW Backup
Tunnel I/F BPS:Type Tunnel:Label State Level Type
------ ------- -------- ------------- ------ ----- -----
Force10_T1 Gi 0/18 0K:G Tu 2:16 Ready any-unl Nhop
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 41
Below, FRR is triggered by shutting down Gig 0/18, and the backup path is activated. Then the original
link is restored by reenabling the interface.
Force10(conf)#int gi 0/18
Force10(conf-if-gi-0/18)#shut
Sep 23 07:20:41: %RPM0-P:CP %IFMGR-5-ASTATE_DN: Changed interface Admin state to down: Gi
0/18
Force10(conf-if-gi-0/18)#Sep 23 07:20:41: %RPM0-P:CP %IFMGR-5-OSTATE_DN: Changed interface
state to down: Gi 0/18
Force10(conf-if-gi-0/18)#do show mpls traffic-eng fast-reroute database
Tunnel head end item frr information:
Protected tunnel In-label Out intf/label FRR intf/label Status
---------------- -------- -------------- -------------- ------
Gig 0/66
Gig 0/60
Gig 0/31
Tunnel 2
Verify that the backup path status show mpls traffic-eng fast-reroute database
is Ready.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 43
Step Task Command Syntax
Verify that the backup tunnel type show mpls rsvp fast-reroute
is N-Nhop, which means that the
LSP is protected from a failure of
the Transit LSR (node failure
protection is enabled).
Force10#show mpls rsvp fast-reroute
Primary Protect BW Backup
Tunnel I/F BPS:Type Tunnel:Label State Level Type
------ ------- -------- ------------- ------ ----- -----
Force10_T1 Gi 0/66 0K:G Tu 2:3 Ready any-unl N-Nhop
Below, FRR is triggered by shutting down Gig 0/66, and the backup path is activated. Then the original
link is restored by reenabling the interface.
Force10(conf)#int gi 0/66
Force10(conf-if-gi-0/66)#shut
00:32:51: %RPM0-P:CP %IFMGR-5-ASTATE_DN: Changed interface Admin state to down: Gi 0/66
Force10(conf-if-gi-0/66)#00:32:51: %RPM0-P:CP %IFMGR-5-OSTATE_DN: Changed interface state
to down: Gi 0/66
Force10(conf-if-gi-0/66)#do show mpls traffic-eng fast-reroute database
Tunnel head end item frr information:
Protected tunnel In-label Out intf/label FRR intf/label Status
---------------- -------- -------------- -------------- ------
Fast reroute provides bandwidth and node protection by default, though there are no commands.
Bandwidth protection enhances fast rerouteby providing a maximum protectable capacity during a failure.
Unprotected LSPs may use remaining capacity. The maximum utilization may be greater than 100%.
• If there is no bandwidth configuration on either the primary or the backup, then the primary receives
protection since, this is essentially an unlimited bandwidth situation.
• If the primary tunnel has 100m but the backup tunnel has no bandwidth configuration, the primary
tunnel still receives protection since the default is unlimited bandwidth on backup tunnels.
• If there is no bandwidth configuration on the primary tunnel and this a bandwidth configuration on the
backup tunnel, then the primary tunnel does not receive protection.
• For example, if primary tunnel 1 has 100m, primary tunnel 2 has 100m, but the backup tunnel has only
100m bandwith, then only one of the primary tunnels receives protection. If the backup tunnel was
increased to 200m, then both tunnels receive protection.
2 Configure the backup tunnel and enable FRR and bandwidth protection.
Force10#show run int tu 1
!
interface Tunnel 1
ip unnumbered Loopback 0
tunnel destination 100.1.1.3
tunnel mode mpls
tunnel mpls traffic-eng path-option 1 explicit name p-1
tunnel mpls traffic-eng fast-reroute bw-protect
no shutdown
3 Verify your configuration using show mpls show mpls rsvp EXEC Privilege
rsvp fast-reroute bw-protect. The value of fast-reroute bw-protect
BW-P will be "ON" only if fast-reroute with
bw-protect is configured.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 45
Step Task Command Syntax Command Mode
FTOS Behavior: When FRR is triggered and the backup tunnel becomes active, the BFD
neighborship for the backup tunnel at the merge point shows the neighbor in down state although it is
up. In the following example, Transit R2 is the PLR and Transit R3 is the merge point.
R2 R3
Transit-PLR Transit-MP
R1 R4
Head-end Tail-end
2/12 2/16
2.1.1.1 Tunnel 1 2.1.1.2 3.1.1.1
5.1.1.1 2/13
5.1.1.2
Tunnel 2
When the Tunnel 1 link between R2 and R3 fails, the backup tunnel is activated. On the merge point,
R3, the BFD state for the session with 100.1.1.2 is shows as Down, although the link is in fact up.
Force10#show bfd neighbors
[output omitted]
LocalAddr RemoteAddr Interface State Rx-int Tx-int Mult Clients
* 2.1.1.2 2.1.1.1 Gi 2/12 Up 100 100 3 M
* 3.1.1.1 3.1.1.2 Gi 2/16 Up 100 100 3 M
* 5.1.1.2 5.1.1.1 Gi 2/13 Up 100 100 3 M
Force10#Jul 24 19:51:11: %RPM0-P:CP %BFDMGR-1-BFD_STATE_CHANGE: Changed session
state to Down for
neighbor 2.1.1.1 on interface Gi 2/12 (diag: TMR_EXP)
Force10#show bfd neighbors
[output omitted]
LocalAddr RemoteAddr Interface State Rx-int Tx-int Mult Clients
* 2.1.1.2 2.1.1.1 Gi 2/12 Down 1000 1000 3 M
* 3.1.1.1 3.1.1.2 Gi 2/16 Up 100 100 3 M
* 5.1.1.2 5.1.1.1 Gi 2/13 Up 100 100 3 M
* 0.0.0.0 100.1.1.2 Gi 2/13 Down 1000 1000 3 M
Two important parameters are calculated using the values contained in the control packet.
• Transmit interval — Transmit interval is the agreed-upon rate at which a system sends control
packets. Each system has its own transmit interval, which is the greater of the last received remote
Desired TX Interval and the local Required Min RX Interval.
• Detection time — Detection time is the amount of time that a system does not receive a control
packet, after which the system determines that the session has failed. Each system has its own
detection time.
BFD must be enabled on both sides of a link in order to establish a session. The two participating systems
can assume either of two roles:
• Active—The active system initiates the BFD session. Both systems can be active for the same session.
• Passive—The passive system does not initiate a session. It only responds to a request for session
initialization from the active system.
A session can have four states: Adminstratively Down, Down, Init, and Up.
• Administratively Down—The local system will not participate in a particular session.
• Down—The remote system is not sending any control packets or at least not within the detection time
for a particular session.
• Init—The local system is communicating.
• Up—The both systems are exchanging control packets.
1 Verify that you have RSVP neighbors. show mpls rsvp neighbors EXEC Privilege
Force10(conf-if-range-tu-1-3)#do show mpls rsvp neighbor
Neighbor Interface Time since Msg rcvd Msg sent
10.1.1.2 Gi 3/0 00:00:05 00:00:20
11.1.1.2 Gi 3/1 00:00:00 00:00:10
12.1.1.2 Gi 3/2 00:00:18 00:00:16
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 47
Step Task Command Syntax Command Mode
3 Establish sessions with all RSVP mpls rsvp bfd all-neighbors CONFIGURATION/
neighbors, or all RSVP neighbors out of an INTERFACE
interface.
4 Display established sessions using the show bfd neighbors EXEC Privilege
command .
Force10(conf)#do show bfd neighbors
BFD sessions are configured with default intervals and a default role. The parameters that can be
configured are: Desired TX Interval, Required Min RX Interval, Detection Multiplier, and system role. If
you change a parameter globally, the change affects all RSVP neighbors sessions. If you change a
parameter at interface level, the change affects all RSVP sessions on that interface, and it supercedes the
global parameter value.
Change parameters for all RSVP mpls rsvp bfd all-neighbors interval seconds CONFIGURATION/
sessions on an interface. min_rx min_rx multiplier value role {active | INTERFACE
passive}
If BFD is disabled globally, all sessions are torn down, and sessions on the remote system are placed in a
Down state. If BFD is disabled on an interface, sessions on the interface are torn down, and sessions on the
remote system are placed in a Down state. Disabling BFD does not trigger a change in BFD clients; a final
Admin Down packet is sent before the session is terminated.
Disable BFD sessions with all RSVP mpls rsvp bfd all-neighbors disable CONFIGURATION
neighbors.
Disable BFD sessions with all RSVP mpls rsvp bfd all-neighbors disable INTERFACE
neighbors out of an interface when BFD is
enabled globally
Path Protection
Path Protection is essentially the establishment of an additional LSP in parallel with an existing LSP, where
the additional LSP is used only in case of failure. This LSP is sometimes called the backup, secondary, or
standby LSP. The backup LSP is not used to carry traffic except during a failure condition.
The backup LSP is built along paths that are as diverse as possible from the LSP they are protecting. This
ensures that a failure along the path of the primary LSP does not also affect the backup LSP.
Path protection is a simple concept. Each primary LSP is backed up by a standby LSP. Both the primary
and backup LSPs are configured at the head-end LSR. Both are signalled ahead of time on the control
plane. The primary and backup LSPs might have the same constraints. If the primary LSP has a bandwidth
reservation of 100 Mbps, the backup LSP may also reserve 100 Mbps. This way, the end-to-end
characteristics essentially remain the same, regardless of whether the LSP used to carry the traffic is the
primary LSP or secondary LSP.
Path protection has better convergence than IGP convergence in an IP network or MPLS TE LSP reroute
because it makes use of a presignalled LSP that is ready to go in case the primary LSP fails. With path
protection, the relationship between the backup LSP and the number of primary LSPs it is protecting is 1:1.
In other words, for every LSP you want to protect, you have to signal another LSP.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 49
Figure 11 Path Protection
R2
Transit R3
R1
Head-end Gig 0/60 1.1.2.2 1.1.3.2 Tail-end
Tunnel 1 Primary Path
1 On the Head-end LSR [R1], create a tunnel show mpls traffic-eng tunnels EXEC Privilege
numbered tunnel 1, the destination of which brief
is the Tail-end LSR [R3] router-ID. Borrow an
IP address from any configured loopback
interface or physical interface. This will be the
primary path.
Force10#show mpls traffic-eng tunnels brief
TUNNEL NAME DESTINATION UP IF DOWN IF STATE/PROT
Force10_T1 100.1.1.3 - Gi 0/60 up/up
Force10#show running-config interface tunnel 1
!
interface Tunnel 1
ip unnumbered Loopback 0
tunnel destination 100.1.1.3
tunnel mode mpls
tunnel mpls traffic-eng path-option 1 explicit name p-1
no shutdown
2 Create a new explicit path, and configure the ip explicit-path name name CONFIGURATION
next addresses so that a link different from next-address ip-address {strict |
the primary tunnel is chosen from the loose}
head-end through the transit router to the
tail-end.
Force10(conf)#ip explicit-path name standbylsp
Force10(conf-ip-exp-path)#next-address 1.1.11.2 strict
Force10(conf-ip-exp-path)#next-address 1.1. 4.2 strict
3 Add the explicit path to the primary tunnel as tunnel mpls traffic-eng CONFIGURATION
protection. path-option protect path-num
{dynamic | explicit name
path-name} [lockdown]
[bandwidth bandwidth]
Force10(conf)#int tunnel 1
Force10(conf-if-tu-1)#$fic-eng path-option protect 1 explicit name standbylsp
Force10(conf-if-tu-1)#show config
!
interface Tunnel 1
ip unnumbered Loopback 0
tunnel destination 100.1.1.3
tunnel mode mpls
tunnel mpls traffic-eng path-option 1 explicit name p-1
tunnel mpls traffic-eng path-option protect 1 explicit name standbylsp
no shutdown
4 Verify that the primary path is protected, and show mpls traffic-eng tunnels EXEC Privilege
that the secondary path is up. protection
In Label: -
Out Label: GigabitEthernet 0/61, 20
RSVP Signalling Information:
Src 100.1.1.1, Dst 100.1.1.3, Tunnel Id 1, Instance 7
RSVP Path Info:
Local IP: 100.1.1.1
Explicit Route: 1.1.11.1 1.1.11.2 1.1.4.1 1.1.4.2
Record Route: None
Tspec: ave rate 0 kbits, burst 0 bytes peak rate 0 kbits
RSVP Resv Info:
Record Route: None
Fspec: ave rate 0 kbits, burst 0 bytes, peak rate 0 kbits
Configure the retry interval for tunnel mpls traffic-eng retries INTERFACE
establishing LSPs. TUNNEL
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 51
MPLS on LAGs
FTOS includes a default CAM profile and microcode that treats MPLS packets as follows:
• When MPLS IP packets are received, hashing is based on the label + IP 5 tuple (source IP address,
destination IP address, IP protocol, and source and destination UDP or TCP port number).
• If the packet is part of a VPN, hashing is based on the inner and outer labels.
Creating Labels
In conventional IP routing, each router determines the next hop by matching the destination IP address to
the longest matching address-prefix in the routing table. Packets that match the same routing table entry
normally take the same path, and in this way, the lookup can be considered a type of packet classification.
In an MPLS network, a group of packets that take the same path from ingress to egress is called a
Forwarding Equivelance Class (FEC). Packets are classified (the next hop is determined) by matching the
destination IP against one or both of two criteria.
• an address prefix—the network address portion of the destination IP of the packet, and/or
• a host address—the complete destination IP address of the packet
Host address FEC elements are explicitly configured, while address prefixes are found in the routing table,
which is primarily populated by a routing protocol. So, generally, each address prefix in the routing table is
an FEC.
If an FEC has more than one element, the LSR matches according to the following rules:
1. if a match against a host address element occurs, then the packet is classified to the corresponding
FEC, else
2. the packet is classified to the FEC that has the longest matching address prefix.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 53
When do LSRs Create Labels?
A label is a (in most cases) unique, arbitrary value that represents an FEC and is inserted into a packet. A
label-FEC pair is called a binding; bindings are local to adjacent MPLS routers (LSR) and are exchanged
using LDP.
Note: A label is in most cases globally unique. However, a system may bind the same label to more than
one FEC when FECs are used in different applications, or contexts. The context in which the label is used
is called a label space, and is represented by a16- bit value concatenated with the LSR Identifier for label
distribution. There are two types of label spaces, per-interface and per-platform.
• A per-interface label space exists when an LSR binds a label to more than one FEC and distributes
each binding to a different system connected to it by a point-to-point link. Because of the point-to-point
connection, the LSR can discern which FEC a label represents based on the ingress interface.
• A per-platform label space (global label space) exists when all of the binding created by the sytem
are unique.
MPLS is designed so that only one router performs the routing table lookup for a given address prefix. It
then creates a label so that the remaining routers switch based on the label. So which LSR performs the
lookup and creates the label for a given prefix? Each LSR looks into its routing table and decides for which
address prefixes (FECs) it will create labels based on its Label Control Mode.
MPLS architecture defines two label control modes. Although both modes may be used in the same
network, ordered control is achieved only when all LSRs in the network use ordered control.
• Independent Label Distribution Control—an LSR, when it recognizes an FEC, binds a label to it,
and distributes it.
• Ordered Label Distribution Control—an LSR binds a label to an FEC only if it is the egress LSR or
if it has received a binding for an FEC from its next hop. This is the FTOS default.
• The FEC refers one of the LSRs directly connected interfaces (R3, Figure 12).
Label Distribution
Labels are distributed in the direction of egress to ingress LSR (upstream) for a given FEC. The
distribution procedure is based on the combination of Label Advertisement Mode and Label Control
Mode.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 55
• PulledConditional—Downstream-on-Demand + Ordered Control. An LSR responds to a request for a
label mapping only if it is the egress LSR for the FEC, or if it has received a label mapping from the
next-hop.
Label Retention
Self-recognized and learned labels are stored in a Label Information Base (LIB). Each entry contains an
address prefix and a collection of (LDP Identifier, label) pairs, one from each advertising peer. The label
that is used for forwarding is the one with an LDP Identifier that belongs to the next hop.
Each LSR independently decides whether to keep or discard the label mappings it receives.
LDP Peering
In order to exchange label mappings, two LSRs must peer. LSRs may peer even if they are directly are not
directly connected. A peership is a TCP session between the two LSRs.
Initialization Message
session operational
LDP PDUs begin with a header that contains three fields, as shown in Figure 14.
Peering Messages
Peering messages are used to establish LDP adjacencies and sessions.
Hello Messages
Hello messages are used to discover LDP neighbors. LSRs may be in active mode or passive mode. Active
mode LSRs send hello messages out of all interfaces; passive LSRs only respond to hellos.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 57
Figure 14 Hello Message
Protocol Header Src IP Addr Dest IP Addr Options Padding UDP Packet
(17) Checksum (224.0.0.2)
Version PDU Length LDP Identifier U Message Type Msg Msg ID U F Type Length Hold Time T R Reserved Optional Parameters
(1) (0) (0x0100) Length (0) (0) (0x0400)
Initialization messages are used to negotiate or advertise session parameters such as label advertisement
mode and loop detection.
Initialization Message
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 59
Notification Messages
Notification messages communicate the outcome of processing an LDP message or the state of the LDP
session, including:
Notification Message
Version PDU Length LDP Identifier U Message Type Msg Msg ID StatusTLV Optional Parameters
(1) (0) (0x0001) Length
E F Status Data
Address Message
Each LSR sends to its peer active interface addresses. This list is used to determine whether the next hop
for an address prefix is an LDP peer. When an LSR receives an label mapping from a downstream peer, it
determines if the source LSR is the next hop for the FEC. If it is, the LSR installs and uses the label for
forwarding. If it is not, the LSR either installs (but does not use) or releases the label based on its label
retetion mode.
Address Withdraw messages are used to signal to a peer that a previously advertised address is no longer
valid.
Version PDU Length LDP Identifier U Message Type Msg Msg ID Address List TLV Optional Parameters
(1) (0) (0x0300) Length
None defined
Code:
1: IPv4
2: IPv6
LSRs use label mapping messages to advertise label mappings to upstream peers. An LSR advertises a
label mapping to a peer when:
• it recognizes a new FEC in its forwarding table, and its distribution mode is Unsolicited Downstream,
• it receives a request for a label mapping, or
• it receives a mapping from a downstream peer that has not yet been distributed upstream.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 61
Figure 18 Label Mapping Message
Label Exp S TTL
Value
Version PDU Length LDP Identifier U Message Type Msg Msg ID FEC TLV Label TLV Optional Parameters
(1) (0) (0x0400) Length
The receiving LSR looks into its routing table to determine the label mapping. It responds to the request
message with the label mapping or with a notification message indicating why it cannot satisfy the request.
Version PDU Length LDP Identifier U Message Type Msg Msg ID FEC TLV Optional Parameters
(1) (0) (0x0401) Length
A Label Release Message signals to a peer that the LSR is no longer using a label mapping that either it
requested or was advertised to it. A label is released when:
• the advertisting LSR is no longer the next hop for the advertised FEC, and the releasing LSR is in
conservative label retention mode
• an LSR receives a label mapping from an LSR that is not the next hop for the FEC
• an LSR receives a withdraw message
A Label Withdraw Message signals to a peer that a label mapping that the LSR no longer recognizes an
FEC that it previously advertised.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 63
Reserved Labels—Penultimate Hop Popping
FTOS performs penultimate hop popping (PHP) by default. FTOS binds an “Implicit Null” label to FECs
for which it is the egress LSR, and distributes the label upstream. For those FECs, the upstream neighbor
removes the current top label, rather than swapping it. This relieves the egress LSR of having to perform
two lookups, one for the top label which it will remove after discovering it is the egress, and another for the
new top label. The “Implict Null” label has a value of 3, which is a reserved value. Though the label is
assigned and distributed, forwarded packets are never actually labeled.
Implementation Information
• The FTOS implementation of Label Distribution Protocol is based on RFC 3036.
• Per-interface label space is not available; only per-platform label space is available.
• Downstream-on-Demand advertisement mode is not available; only Unsolicited Downstream is
available because it converges faster.
• Conservative Label Retention is not available; only Liberal Label Retention is available.
Change the LSR router ID. mpls ldp router-id interface CONFIGURATION
The router ID is contained in the optional hello parameter IPv4 Transport Address. When you enter this
command, the router ID is not immediately changed; FTOS uses the new ID when it has to compute one,
most likely when LDP is globally disabled and re-enabled. At this time, the specified interface is used if:
If the conditions are not satisfied upon entering the configuration, but are later satisfied, FTOS uses the
new router ID at that time. If the interface currently chosen for router ID is deleted, or if the IP address is
removed, FTOS computes a new router ID. LDP reinitializes when the router ID is re-computed.
• The Hello Adjacency Holdtime a timeout value for an adjacency, and by default is 15 seconds; hellos
are sent every 5 seconds.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 65
• The Session Holdtime is the timeout value for the LDP Session. The default is 40 seconds for holdtime
and keepalives are sent every 13 seconds.
Change the timeout value for an LDP mpls ldp discovery holdtime seconds CONFIGURATION
adjacency. Default: 15 seconds
Change the timeout value for an LDP mpls ldp holdtime seconds CONFIGURATION
session.
• Independent Label Distribution Control—an LSR, when it recognizes an FEC, binds a label to it,
and distributes it.
• Ordered Label Distribution Control—an LSR binds a label to an FEC only if it is the egress LSR or
if it has received a binding for an FEC from its next hop.
• allocated for a label for the IP address of an interface and advertise it to neighbors with /32 mask, or
• select the prefixes for which labels are advertised, and
• optionally, select the peers to which specified prefixes are advertised.
Allocate a label for an interface IP mpls ldp advertise-labels [interface interface | CONFIGURATION
address, or select the prefixes for for prefix-list [to peer-list]]
which labels are advertised.
Choose the labels that an LSR will mpls ldp neighbor ip-address labels accept CONFIGURATION
accept from a neighbor. prefix-list
MD5 Authentication
LDP uses the TCP/IP protocol suite; LDP listens on well-known TCP port 646 for incoming messages.
MD5 Authentication is a method that each LSR may use to authenticates its peer LSRs. Each LSR stores
the MD5 password, and only after password verification during the TCP handshake is the TCP session
established.
Enable neighbor authentication [no] mpls ldp neighbor ip-address password [ 7 CONFIGURATION
using an MD5 password. encrypted-pass | clear-pass]
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 67
Task Command Syntax Command Mode
Whenever the MD5 password is updated (added/deleted/modified), the affected LDP sessions are
reinitialized, and FTOS displays Message 1.
Jun 23 15:22:07: %RPM0-P:RP1 %LDP-5-NBRCHG: Connection with LDP neighbor 101.1.1.10:0 closed.
(Password Changed)
Jun 21 15:39:11: %RPM0-P:RP1 %KERN-6-INT: LDP MD5 password mismatch from 101.1.1.10:11082 to
101.1.1.2:646
If the MD5 password is not configured on one LSR, FTOS displays Message 3.
Clear LDP sessions for a specified clear mpls ldp neighbor {ip-address | *} EXEC Privilege
neighbor or all neighbors.
Clear LDP statistics for a specified clear mpls ldp statistics neighbor EXEC Privilege
neighbor or all neighbors. This {ip-address | *}
command clears the statistics
displayed in show mpls ldp
neighbor.
Display LDP label errrors and debug mpls ldp {binding | error | events | EXEC Privilege
events. routing}
Display LDP protocol messages in debug mpls ldp dump [in | out | [address | EXEC Privilege
hexadecimal format. hello | init | keepalive | label | notification]
[in | out]]
Display LDP protocol messages in debug mpls ldp messages [in | out | EXEC Privilege
clear text. [address | hello | init | keepalive | label |
notification]] [in | out]
Display the LDP Label Information show mpls ldp bindings [ip-address/mask] EXEC Privilege
Base (LIB). [{local-label | remote-label} label] [neighbor
ip-address] [local]
Display sources for locally show mpls ldp discovery [detail] [interface] EXEC Privilege
generated discovery hello PDUs.
Display MPLS interfaces. show mpls ldp interfaces {interface | brief | EXEC Privilege
advertise-label}
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 69
Task Command Syntax Command Mode
Display LDP neighbor information. show mpls ldp neighbor [ip-address [detail] | EXEC Privilege
brief]
At ingress, MPLS-EXP is set based on the packet DSCP value. The incoming DSCP value can be matched
and mapped to the MPLS-EXP bit using class maps or QoS policies. Marking is based only on the DSCP
value, and marking is done on ingress only.
Note: It is possible to mark packets in the MPLS-EXP bit using both class maps and policy maps. In this
case, the class-map configuration takes precedence. If DSCP marking is also configured then the three
most significant bytes of the DSCP value, and the MPLS-EXP value, wherever configured (in a class map
or policy map), must match. It is possible that DSCP marking is configured with one value, and in another
context, MPLS-EXP marking is configured with a different value. In this case, the MPLS-EXP value used for
both DSCP and MPLS-EXP.
Force10(conf)#class-map match-any BB
Force10(conf-class-map)#match ip dscp 4 ?
multicast Multicast entry
set-ip-dscp Mark input traffic
set-mpls-exp Mark input traffic from ip to mpls
<cr>
MPLS-EXP marking is independent of DSCP marking, so marking MPLS-EXP alone does not mark a
DSCP value.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 71
MPLS-EXP and DSCP can be marked at the same point in time, but you must configure them with the
same value.
By the usual Force10 QoS architecture, once you create a class map, you must map it to a policy map,
which must then be mapped to a service-policy on an interface.
Force10(conf)#policy-map-input BB
Force10(conf-policy-map-in)#service-queue 0 class-map BB
Force10(conf-policy-map-in)#int gig1/6
Force10(conf-if-gi-1/6)#service-policy input BB
Force10(conf)#qos-policy-input BB
Force10(conf-qos-policy-in)#set ?
ip-dscp Specify dscp values
mac-dot1p Specify dot1p values
mpls-exp Specify mpls exp values
Force10(conf-qos-policy-in)#set mpls-exp 3
Force10(conf-qos-policy-in)#show conf
!
qos-policy-input BB
set mpls-exp 3
By the usual Force10 QoS architecture, this QoS policy must be mapped to a policy map, which must then
be mapped to a service-policy on an interface
The following is a mapping from MPLS-EXP to DSCP when mpls ip propagate-exp is configured:
Head-end Tail-end
FTOS Behavior: When two labels remain, and no mpls ip propagate-ttl is configured, mpls ip
propagate-exp (default) does not work. If you want to propagate MPLS-EXP in this case, you must
enable TTL Propagation (mpls ip propagate-ttl).
Advertising Explicit-Null
The Implicit-Null label is advertised by the Egress LSR, and it instructs the penultimate hop to remove the
top label before forwarding the packet, a process called Penultimate-Hop Popping (PHP). Explicit-Null is
also available on FTOS. At the penultimate hop, the incoming label is swapped for the Explicit-Null label
(Label 0 for IPv4, and Label 2 for IPv6). The Explicit-Null label alerts the Egress LSR that it is the egress
so that it can immediately remove the top label, a process called Ultimate-Hop Popping. Ultimate-hop
popping ensures that any packets traversing an MPLS network include a label. This is important for MPLS
Class of Service because packets are classified using the MPLS-EXP bits contained in the label. If PHP is
used, the MPLS-EXP bits are removed before the packet can be switched across the entire LSP. When the
top label is Explicit-Null, and only one label remains queuing is based on the packet DSCP value.
Note: After you configure mpls traffic-eng advertise label explicit-null you restart RSVP on the
interface by executing shutdown followed by no shutdown. Sessions through the interface are torn down
but Path Tear messages are not sent.
Force10(conf-if-gi-1/6)#mpls traffic-eng advertise label explicit-null
% Warning: Requires shut/noshut of GigabitEthernet 1/6
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 73
Task Command Syntax Command Mode
On the penultimate LSR, configure the mpls traffic-eng advertise explicit-null INTERFACE
system to swap the incoming top label
with the Explicit-Null label.
FTOS Behavior: With explicit-null, the penultimate hop queues packets based on the MPLS-EXP
value. However, on egress of the ultimate hop, Force10 systems queue based on the DSCP value.
Transit
Head-end Tail-end
1. RSVP-TE uses the Downstream-on-Demand label advertisement mode. The ingress LSR creates an
RSVP Path message with a LABEL_REQUEST object, and sends the message to the tunnel
destination.
2. The tunnel destination allocates a label and responds to the label request with an RSVP Resv message
containing a LABEL object, which label object immediately follows the relevant filter spec.
3. Each node along the LSP that receives the Resv message updates the Incoming Label Map (ILM) with
the label.
• The SESSION object contains the tunnel destination address, the tunnel ID, and the extended tunnel
ID, which is the ingress node’s address.
• The SENDER_TEMPLATE and FILTER_SPEC objects are identical. They contain the ingress node’s
address and the LSP ID.
Implementation Information
• RSVP-TE is not available on VLAN interfaces.
Configure RSVP-TE
1. Enable RSVP globally. See page 76.
2. Enable RSVP on an interface by making a portion of the interface bandwidth reservable. See page 76.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 75
Related Configuration Tasks
• Enable Bandwidth Logging on page 76
• Graceful Shutdown on page 77
• Enable Record Route on page 77
• Enable RSVP Hello Signaling on page 77
• Specify the Signaling Refresh Interval on page 79
• Enable Signaling Refresh Reduction on page 79
• Bandwidth Holding for Path Messages on page 81
• Enable TTL Checking on page 81
Enable RSVP
RSVP is disabled by default.
76 RSVP-TE
Graceful Shutdown
RSVP Graceful Shutdown provides a user-initiated approach to shutting down RSVP. By default, graceful
shutdown is disabled since a large number of RSVP sessions may take a long time to shut down, and a
session timeout on neighbor routers may be preferred.
Recording starts when the tunnel ingress sends a Path message with this object, which contains the
sender’s IP address, and the tunnel ingress optionally sets the Label_Recording flag. Each node that
receives either a Path or Resv message with an RRO, prior to any other RRO processing, inspects each
address in the list. If the router’s own address appears in the list, a loop exists. If a loop is found in a Path
message, a PathErr message is sent to the originator. If a loop is found in a Resv message, the message is
dropped. Assuming no loop, each intermediate node stores the RRO as path state, adds a label, if required,
and then adds the address of the message’s outgoing interface. When the destination node receives the Path
message, it sends back a Resv message also with an RRO, and the process is completed in reverse, towards
the sender. Each node from source to destination is then aware of the entire route, which is useful for
network managment.
Enable route recording. Enabling recording after a tunnel traffic-eng record-route INTERFACE TUNNEL
tunnel is established does not re-establish it.
Recording begins when the next Path message is
sent.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 77
Each node uses only one hello exchange per neighbor. The local node inserts an arbitrary instance value in
the request to identify the hello exchange that which must be reflected by the remote node in its ACK. The
local node assumes that the remote node has failed if it does not respond to a consecutive number of
requests, or the local node assumes that the remote node has reset if the the instance value in the ACK does
not match the value in the request. When communication with the remote node is lost, the remote node
resets, or the local node resets, the local node changes its instance value.
Enable RSVP hello signaling. mpls rsvp signalling hello enable INTERFACE
Optionally, choose a hello interval, mpls rsvp signalling hello {interval milliseconds | INTERFACE
and the number of hello packets misses number}
that must be unacknowledged to
consider the neighbor down.
78 RSVP-TE
You can display all of the active and inactive hello sessions in which the system is particpating, clear
message counters, and clear session statistics.
Display the hello sessions in which show mpls rsvp hello instance {summary | detail} EXEC Privilege
the system is participating.
Force10#show mpls rsvp hello instance summary
Active Instances:
Neighbor I/F State Num Lost LSPs Interval
192.168.53.4 Gi 0/21 Active 1 1 1000
Inactive instances:
- None -
Clear hello counters for all clear mpls rsvp hello counters {packets | EXEC Privilege
instances. statistics}
Configure the signaling refresh mpls rsvp signalling refresh {interval seconds | CONFIGURATION
values R and L. misses number}
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 79
• Refresh messages increase proportionally with the number of sessions. Processing resources are
required for generating, transmitting, and receiving refresh messages every R seconds.
• RSVP messages are transmitted using unreliable IP. Latency is introduced when non-refresh messages
are lost in transmission because RSVP recovers from lost messages via the next periodic refresh. When
R is large, there is a delay in changing reservation state, but when R is small, processing overhead
increases.
• RSVP recovers from loss of tear down messages through the local state lifetime timer (L), which clears
unrefreshed state, but this means that resources are allocated 4R longer than necessary.
The refresh reduction extensions address refresh volume and reliability with mechanisms other than
adjusting the refesh interval. RFC 2691 - RSVP Refresh Overhead Reduction Extensions introduces:
• Bundle Message—When configured, RSVP can delay messages so that many messages can be
combined into a single PDU. Bundle messages are exchanged between neighbors and can contain any
message (ACK, Srefresh, Resv, Path, PathTear) which has a destination address of a next hop.
Bundling conserves bandwidth and reduces processing overhead.
• MESSAGE_ID—Each RSVP Path and Resv message carries this object which contains an arbitrary
value that, combined with the sender’s IP address, uniquely identifes the message. The receiver uses
the message ID to determine if the message represents new state or a state refresh, and in the case of
new state, stores the message ID as part of the path or reservation state for future reference. The sender
may also set the ACK_Desired flag in this object to request an ACK from the next hop to ensure
reliability. Messages that are not acknowleged are retransmitted.
• MESSAGE_ID ACK—Each message in which the ACK_Desired flag is set must be acknowledged
using this object, which contains the ID of the message being acknowledged. One object is generated
per message ID, and they may be included in any RSVP message that is addressed to the node that
generated the original MESSAGE_ID object.
• Srefresh Message—Srefresh messages contain a list of previously advertised messages
(MESSAGE_ID_LIST) for which the state is unchanged. Since the receiver has stored the
MESSAGE_ID that contained the original state, it can refresh the corresponding entries. Srefresh
messages reduce processing overhead for the sender and receiver because complete Path and Resv
messages are not required to refresh state.
Enable RSVP signalling refresh mpls rsvp signalling refresh reduction enable CONFIGURATION
reduction. Default: Disabled
Do not set the ACK_Desired flag mpls rsvp signalling refresh reduction no-ack CONFIGURATION
for RSVP messages. Received Default: ACK enabled
messages with this flag set are still
acknowleged.
80 RSVP-TE
Bandwidth Holding for Path Messages
When an LSR receives a Path message indicating a bandwidth requirement, it earmarks the bandwidth on
the interface until a Reservation message is received. You can configure the amount of time the LSR holds
the bandwidth before releasing it for other reservations in case a reservation is unacceptably delayed or
never arrives.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 81
82 RSVP-TE
Chapter 8 MPLS Traffic Engineering
A traffic trunk is a path along which a set of flows of the same FEC is switched. One of many possible
LSPs can be created for an FEC depending upon traffic engineering parameters, and so trunks can be
treated as movable objects that can be routed away from network failures, congestion, and bottlenecks.
Create a Tunnel
Step Task Command Syntax Command Mode
4 Enable the tunnel to operate in MPLS-TE tunnel mode mpls INTERFACE TUNNEL
mode.
FTOS Behavior: In a triangular configuration, prefixes before the tail-end are shown as reachable via
the tunnel. In the following configuration, Tunnel 2 is established through Te 5/1 and ends at R3.
Though the 21.1 network is before the tail-end, its is shown as reachable via Tunnel 2.
R1 R3
1.1.1.1 3.3.3.3
Force10(conf)#do sh run ospf
! 5/5 (15.1.1.1) 6/5
router ospf 10
network 0.0.0.0/0 area 0 5/6 (16.1.1.1) 6/6
mpls traffic-eng router-id Loopback 0
mpls traffic-eng area 0
Force10(conf)#do sh ip route ospf
Destination Gateway Dist/Metric Last Change Te 5/2
----------- ------- ----------- -----------
O 2.2.2.2/32 via 11.1.1.2, Te 5/1 110/2 00:09:55 11.1 21.1
O 3.3.3.3/32 via 0.0.0.0, Tu 2 110/2 00:07:06
O 21.1.1.0/24 via 11.1.1.2, Te 5/1 110/2 00:07:06 Tunnel 2 Tunnel 2
via 0.0.0.0, Tu 2
Force10(conf)#do sh mpl traffic-eng tunnels tunnel 2 | grep Exp
Explicit Route: 11.1.1.1 11.1.1.2 21.1.1.1 21.1.1.2
R2
2.2.2.2
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 83
Allocate Bandwidth to a Tunnel
Step Task Command Syntax Command Mode
OSPF—Traffic Engineering
OSPF—Traffic Engineering (OSPF-TE) Extensions is a set of administrator-defined link attributes, not
advertised in OSPF itself, that administrators can use to manipulate the standard OSPF topology. These
link attributes, primarily bandwidth-reservation constraints, are advertised in Type 10 Opaque LSAs, and
used by TE-capable routers to create an extended link state database (TE database), separate from the
OSPF link state database. Routers can then use the TE database to compute constraint-based (CSPF)
optimal routes from source to destination.
OSPF-TE extensions, defined in RFC 3630, capture the reservation state of primarily point-to-point links.
It defines a new Type 10, Opaque LSA, with an intra-area flooding scope, called the Traffic Engineering
LSA (TE LSA). Since Type 10 LSAs are intra-area, FTOS maintains a separate TE database for each area.
The TE LSA is Type 1 and starts with a standard LSA header. The payload is in Type, Length, Value (TLV)
format. There are two types of TLVs, both of which are mandatory in each LSA:
• Type 1: Router Address—always reachable IP address on the advertising router, typically a loopback
address.
• Type2: Link—describes a single link, and the descriptors are in TLV format. Each sub-TLV appears in
an LSA only once.
Table 6 Link TLV Sub-types
1 Enable MPLS-TE in an OSPF area or mpls traffic-eng area number ROUTER OSPF
on an interface. INTERFACE
2 Select a router ID for the area. mpls traffic-eng router-id interface ROUTER OSPF
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 85
Display the TE advertisements that OSPF is sending on behalf of a particular interface or for all interfaces
using the command show ip ospf mpls traffic-eng interface. The advertised information includes the
bandwidth that is currently available with OSPF after threshold filtering is done by link management:
GigabitEthernet 5/7 :
Link connected to multiaccess network
Link ID : 17.1.1.2
Interface Address : 17.1.1.1
Admin Metric te: 1 igp: 1
Maximum bandwidth : 125000000
Maximum reservable bandwidth : 93750000
Number of Priority : 8
Priority 0 : 93750000 Priority 1 : 93750000
Priority 2 : 93750000 Priority 3 : 93750000
Priority 4 : 93750000 Priority 5 : 93750000
Priority 6 : 93750000 Priority 7 : 93000000
Affinity Bit : 0x0
Set bandwidth usage mpls traffic-eng flooding thresholds {up | down} {value} INTERFACE
threshold values, that [[value] …]
when crossed, trigger
an IGP TE
advertisement.
LSAs are flooded immediately upon a topology change and when a bandwidth change crosses a threshold
value in the up or down direction. You can configure the periodic-flooding interval.
• The IS Reachability TLV is replaced by the Extended IS Reachability TLV, which contains the
sub-TLVs in Table 7.
• The IS IP Reachability TLV is replaced by the Extended IS Reachability TLV, which increases the
metric range.
Table 7 Extended IS Reachabiltiy Sub-TLVs
2 Enable TE on IS-IS Level 1 or Level 2. mpls traffic-eng {level-1 | level-2} ROUTER ISIS
3 Configure a traffic engineering router ID. mpls traffic-eng router-id ROUTER ISIS
interface
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 87
Route Traffic though Tunnels
There are three methods you can use to route traffic through tunnels rather than normal IGP routes:
Force10(conf-router_ospf-100)#show config
!
router ospf 100
network 100.0.0.0/8 area 100
network 1.0.0.0/8 area 100
mpls traffic-eng router-id Loopback 1
mpls traffic-eng area 100
interface Tunnel 1
tunnel destination 100.10.10.2
tunnel mode mpls
no shutdown
show mpls forwarding table displays the incoming and outgoing labels of the tunnel, the outgoing
interface, and the next hop in the tunnel path.
show mpls forwarding entries linecard tunnel lists line card entries installed for a tunnel. The output
shows the labels per tunnel, outgoing interfaces, and packet and byte counters. When a packet is
forwarded through a tunnel the counters increment. In the output below, ten 1500-byte packets were sent
over the tunnel 1.
The cost of using a tunnel to reach prefixes beyond it is the tunnel cost plus the IGP link cost from the
tunnel destination to the ultimate destination. You can assign one of three types of metrics to a tunnel:
• metric— the specified value is used to reach the tunnel destination. Beyond the tunnel destination, the
IGP metric is used.
• absolute metric—the specified value is used to reach the tunnel destination and any destination
topologically behind the tunnel.
• relative metric— the specified value is added or subtracted from the IGP metric for the tunnel
destination. Beyond the tunnel destination, the IGP metric is used.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 89
Figure 21 IGP Autoroute Metrics
destination topologically
Head-end Transit Tail-end behind the tunnel destination
LSP LSP
Absolute
3 Assign a metric to the tunnel mpls traffic-eng autoroute metric {value | INTERFACE TUNNEL
tunnel. absolute absolute-metric | relative relative-metric}
Force10(conf)#int tun 1
Force10(conf-if-tu-1)#tunnel mpls traffic-eng autoroute announce
Force10(conf-if-tu-1)#show config
!
interface Tunnel 1
ip unnumbered Loopback 0
tunnel destination 3.3.3.3
tunnel mode mpls
tunnel mpls traffic-eng path-option 1 explicit name Tunnel1
tunnel mpls traffic-eng autoroute announce
no shutdown
Destinations reachable through Router C now have Tunnel 1 as their egress interface.
Note: Do not configure Autoroute and Forwarding Adjacency together; this yeilds undesirable results.
Enabling Forwarding Adjacency increases the size of the routing table in all systems in the area.
• Verify that the tunnel’s borrowed IP address is advertised into the IGP.
• Verify that the tunnel destination is the TE router-ID of the tail-end.
• Verify that the IP address used as TE router-ID is advertised into the IGP.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 91
• For OSPF, the tail-end must be in the same area.
2 Enable Forwarding Adjacency on each router. tunnel mpls traffic-eng INTERFACE TUNNEL
forwarding-adjacency
Prior to enabling Forwarding Adjacenty the head-end route table contains the following routes:
Force10#show ip route
Router (Area 0)
LS age: 138
Options: (No TOS-capability, No DC, E)
LS type: Router
Link State ID: 100.1.1.1
Advertising Router: 100.1.1.1
LS Seq Number: 0x80000010
Checksum: 0xea91
Length: 60
Number of Links: 3
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 93
The the tailend router in the reverse direction:
Router (Area 0)
LS age: 29
Options: (No TOS-capability, No DC, E)
LS type: Router
Link State ID: 100.1.1.3
Advertising Router: 100.1.1.3
LS Seq Number: 0x8000000a
Checksum: 0x2d39
Length: 60
Number of Links: 3
Router (Area 0)
LS age: 5
Options: (No TOS-capability, No DC, E)
LS type: Router
Link State ID: 100.1.1.1
Advertising Router: 100.1.1.1
LS Seq Number: 0x80000017
Checksum: 0xd98d
Length: 60
Number of Links: 3
Implementation Information
• Neither MPLS nor LDP ping will be sucessful if the frame is framented at any transit router.
MPLS echo packets are sent to the tunnel destination using the LSP label stack. MPLS echo packets use
127.X.X.X/8 as a destination address to prevent the packet from being IP routed to the destination if the
LSP is broken. MPLS echo replies are IP packets and can be IP routed or MPLS switched.
traffic-eng ping
Figure 22 MPLS ping and Extended MPLS ping
Tail-end
Head-end
1.4.1.1
1.3.4.2
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 95
Tunnel must be operationally up and have a borrowed IP address, as shown below.
Ping a tunnel destination. ping mpls traffic-eng tunnel number EXEC Privilege
Force10#ping mpls traffic-eng tunnel 1
Force10#ping
Protocol [ip] : mpls
Type: Traffic engineering or ldp
Type [Traffic-eng]: Traffic-eng
Tunnel Id: 1
Repeat Count [5]: 1
Datagram size [100]: 200
Timeout in secs [3]:
Verbose [n]:
Extended commands [n]: y
Source IP address: 100.1.1.1
Reply with Router-alert [n]:
Exp bits: 4
Time to live [255]:
Padding data pattern [0xABCD]:
Destination start address [127.0.0.1]: 127.0.0.1
Destination end address [127.0.0.1]: 127.0.0.10
Destination address increment [0.0.0.1]:
Sweep range of sizes [n]:
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 97
traffic-eng traceroute
Figure 23 MPLS traceroute and Extended MPLS traceroute
Tail-end
Head-end
1.3.4.1
Transit 1.4.1.2
1.4.1.1
1.3.4.2
Trace the route through a traceroute mpls traffic-eng tunnel number EXEC Privilege
tunnel.
Force10#traceroute mpls traffic-eng tunnel 1
Force10#traceroute
Protocol [ip] : mpls
Type: Traffic engineering or ldp
Type [Traffic-eng]: Traffic-eng
Tunnel Id: 1
Timeout in secs [3]:
Extended commands [n]: y
Source IP address: 100.1.1.1
Reply with Router-alert [n]:
Exp bits: 4
Time to live [255]:
ldp ping
Figure 24 LDP Ping and Extended LDP ping
Tail-end
R-ID: Lo 100.1.1.3./32
Head-end
192.168.20.1
Transit 192.168.30.3
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 99
Task Command Syntax Command Mode
Force10#ping
Protocol [ip] : mpls
Type: Traffic engineering or ldp
Type [Traffic-eng]: ldp
Target FEC IPv4 address: 100.1.1.3
Target FEC mask length: 32
Repeat Count [5]: 1
Datagram size [100]: 200
Timeout in secs [3]:
Verbose [n]:
Extended commands [n]: y
Source IP address: 100.1.1.1
Reply with Router-alert [n]:
Exp bits: 4
Time to live [255]:
Padding data pattern [0xABCD]:
Destination start address [127.0.0.1]: 127.0.0.1
Destination end address [127.0.0.1]: 127.0.0.10
Destination address increment [0.0.0.1]:
Sweep range of sizes [n]:
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 101
ldp traceroute
Figure 25 LDP traceroute and Extended LDP traceroute
Tail-end
R-ID: Lo 100.1.1.3./32
Head-end
192.168.20.1
Transit 192.168.30.3
Force10#traceroute
Protocol [ip] : mpls
Type: Traffic engineering or ldp
Type [Traffic-eng]: ldp
Target FEC IPv4 address: 100.1.1.3
Target FEC mask length: 32
Timeout in secs [3]:
Extended commands [n]: y
Source IP address: 100.1.1.1
Reply with Router-alert [n]:
Exp bits: 4
Time to live [255]:
FTOS-supported-Error Codes
Table 8 is a list of the error codes—defined in RFC 4379, Detecting Multi-Protocol Label Switched
(MPLS) Data Plane Failures—which FTOS supports.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 103
Disable TTL Propagation
Disabling TTL propagation changes how ingress and egress LSR nodes read and process the TTL value in
a label. A label must have a value in the TTL field. By default, an ingress LSR reads the TTL field in the IP
header of the incoming packet, decrements it by 1, and copies what is left into the TTL field of the MPLS
header. Core LSRs forward the packet only if the TTL value in the uppermost label is not 0. With the no
mpls ip propagate-ttl command, the behavior changes such that the IP header TTL does not reflect the
hops taken across the MPLS core. Routers in the MPLS core network do not appear in traceroute
information.
FTOS Behavior: mpls ip propagate-exp (default) does not work for MPLS label-stacked packets
(more than one label), when no mpls ip propagate-ttl is configured. If you want to propagate
MPLS-EXP in this case, you must enable TTL.
• mplsinterfaceTable
• mplsInterfacePerfTable
• mplsInSegmentTable
• mplsInSegmentPerfTable
• mplsOutSegmentTable
• mplsOutSegmentPerfTable
• mplsXCTable
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 105
Table 9 mplsLSRMIB Tables and OIDs
The maximum value of an MPLS label that this LSR is
willing to receive on this interface.
mplsInterfaceLabelMaxIn .1.3.6.1.2.1.10.166.2.1.1.1.3 Return Value: 1048575
mplsInterfaceTable—.1.3.6.1.2.1.10.166.2.1.1
The minimum value of an MPLS label that this LSR is
willing to send on this interface.
mplsInterfaceLabelMinOut .1.3.6.1.2.1.10.166.2.1.1.1.4 Return Value: 16
The maximum value of an MPLS label that this LSR is
willing to send on this interface.
mplsInterfaceLabelMaxOut .1.3.6.1.2.1.10.166.2.1.1.1.5 Return Value: 1048575
Indicates the total amount of usable bandwidth on this
interface and is specified in kilobits per second (Kbps).
This variable is not applicable when applied to the
interface with index 0. When this value cannot be
measured, this value should contain the nominal
bandwidth.
mplsInterfaceTotalBandwidth .1.3.6.1.2.1.10.166.2.1.1.1.6 Return Value: 0
Indicates the total amount of available bandwidth
available on this interface and is specified in kilobits per
second (Kbps). This value is calculated as the difference
between the amount of bandwidth currently in use and
that specified in mplsInterfaceTotalBandwidth. This
variable is not applicable when applied to the interface
with index 0. When this value cannot be measured, this
value should contain the nominal bandwidth.
mplsInterfaceAvailableBandwidth .1.3.6.1.2.1.10.166.2.1.1.1.7 Return Value: 0
If the value of the mplsInterfaceIndex for this entry is
zero, then this entry corresponds to the per-platform label
space for all interfaces configured to use that label
space. In this case the perPlatform(0) bit MUST be set;
the per Interface(1) bit is meaningless and MUST be
ignored.
mplsInterfaceLabelParticipationType .1.3.6.1.2.1.10.166.2.1.1.1.8 Return Value: 00
mplsInterfacePerfTable—.1.3.6.1.2.1.10.166.2.1.2
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 107
Table 9 mplsLSRMIB Tables and OIDs
A variable that indicates the storage type for this object.
The agent MUST ensure that this value remains
consistent with the associated mplsXCEntry. Conceptual
rows having the value
'permanent' need not allow write-access to any columnar
objects in the row.
mplsInSegmentStorageType .1.3.6.1.2.1.10.166.2.1.4.1.11 Return Value: 2
mplsInSegmentPerfTable—.1.3.6.1.2.1.10.166.2.1.5
A unique index for this row. While a value of a string
containing the single octet 0x00 is not valid as an index
for entries in this table, it can be supplied as a valid value
to index the mplsXCTable to represent entries for which
mplsOutSegmentIndex .1.3.6.1.2.1.10.166.2.1.7.1.1 no out-segment exists or has been configured.
The interface index of the outgoing interface. Cannot be
modified if mplsOutSegmentRowStatus is active(1).
mplsOutSegmentRowStatus cannot be set to active(1)
until this object is set to a value corresponding to a valid
ifEntry.
mplsOutSegmentInterface .1.3.6.1.2.1.10.166.2.1.7.1.2 Return Value: Interface index of the outgoing interface.
Indicates whether or not a top label should be pushed
onto the outgoing packet's label stack. The value of this
variable MUST be set to true(1) if the outgoing interface
does not support pop-and-go (and no label stack
remains). For example, on an ATM interface, or if the
segment represents a tunnel origination. Note that it is
considered an error in the case that
mplsOutSegmentPushTopLabel
is set to false, but the cross-connect entry which refers to
this out-segment has a non-zero mplsLabelStackIndex.
The LSR MUST ensure that this situation does not
happen. This object cannot be modified if
mplsOutSegmentRowStatus is active(1).
Return Value: 1 if an FRR swap label value is present. 0
mplsOutSegmentPushTopLabel .1.3.6.1.2.1.10.166.2.1.7.1.3 if the FRR swap label is Implicit-Null.
If mplsOutSegmentPushTopLabel is true then this
represents the label that should be pushed onto the top
of the outgoing packet's label stack. Otherwise this value
SHOULD be set to 0 by the management station and
MUST be ignored by the agent. This object cannot be
modified if mplsOutSegmentRowStatus is active(1).
Return Value: FRR swap label value. 3 if the FRR swap
mplsOutSegmentTopLabel .1.3.6.1.2.1.10.166.2.1.7.1.4 label is Implicit-Null.
If the label for this segment cannot be represented fully
within the mplsOutSegmentLabel object, this object
MUST point to the first accessible column of a
conceptual row in an external table containing the label.
In this case, the mplsOutSegmentTopLabel object
SHOULD be set to 0 and ignored. This object MUST be
set to zeroDotZero otherwise.
mplsOutSegmentTopLabelPtr .1.3.6.1.2.1.10.166.2.1.7.1.5 Return Value: 0.0
mplsOutSegmentTable—.1.3.6.1.2.1.10.166.2.1.7
Indicates the next-hop Internet address type. Only values
unknown(0), IPv4(1) or IPv6(2) are supported. A value of
unknown(0) is allowed only when the outgoing interface
is of type point-to-point. If any other unsupported values
are attempted in a set operation, the agent MUST return
an inconsistent-value error.
mplsOutSegmentNextHopAddrType .1.3.6.1.2.1.10.166.2.1.7.1.6 Return Value: Returns a value of 1, which maps to IPV4.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 109
Table 9 mplsLSRMIB Tables and OIDs
Index of out-segment for LSPs not terminating on this
LSR if not set to the string containing the single octet
0x00. If the segment identified by this entry is
terminating, then this object MUST be set to the string
containing a single octet 0x00 to indicate that no
mplsXCOutSegmentIndex .1.3.6.1.2.1.10.166.2.1.10.1.3 corresponding mplsOutSegmentEntry shall exist.
Identifies the label switched path that this cross-connect
entry belongs to. This object cannot be modified if
mplsXCRowStatus is active(1) except for this object.
Return Value: The instance IDs of transit router tunnels,
which show the instance IDs of the tunnels originating
mplsXCLspId .1.3.6.1.2.1.10.166.2.1.10.1.4 from transit router as well.
mplsXCTable—.1.3.6.1.2.1.10.166.2.1.10
Primary index into mplsLabelStackTable identifying a
stack of labels to be pushed beneath the top label. Note
that the top label identified by the out-segment ensures
that all the components of a multipoint-to-point
connection have the same outgoing label. A value of the
string containing the single octet 0x00 indicates that no
labels are to be stacked beneath the top label. This
object cannot be modified if mplsXCRowStatus is
active(1).
mplsXCLabelStackIndex .1.3.6.1.2.1.10.166.2.1.10.1.5 Return Value: 0
The entity that created and is responsible for managing
this cross-connect.
Return Value: Returns a value of 6, which means
mplsXCOwner .1.3.6.1.2.1.10.166.2.1.10.1.6 RSVP-TE.
For creating, modifying, and deleting this row. When a
row in this table has a row in the active(1) state, no
objects in this row except this object and the
mplsXCStorageType can be modified.
mplsXCRowStatus .1.3.6.1.2.1.10.166.2.1.10.1.7 Return Value: 1
A variable that indicates the storage type for this object.
The agent MUST ensure that the associated in and out
segments also have the same StorageType value and
are restored consistently upon system restart. This value
SHOULD be set to permanent(4) if created as a result of
a static LSP configuration. Conceptual rows having the
value 'permanent' need not allow write-access to any
columnar objects in the row.
mplsXCStorageType .1.3.6.1.2.1.10.166.2.1.10.1.8 Return Value: 2
The desired operational status of this segment.
Return Value: “Up” if there is a tunnel/segment present,
mplsXCAdminStatus .1.3.6.1.2.1.10.166.2.1.10.1.9 and no value returned if no segment is present.
The actual operational status of this cross-connect.
Return Value: “Up” if there is a tunnel/segment present,
mplsXCOperStatus .1.3.6.1.2.1.10.166.2.1.10.1.10 and no value returned if no segment is present.
• mplsTunnelTable
mplsTunnelTable
• Entries in this table with an instance of 0 and a source address of 0 represent tunnel head
configurations. All other entries in this table represent instances of LSPs, both signaled and standby.
• If a tunnel instance is signaled, its operating status (operStatus) is set to "up" (1), and its instance
corresponds to an active LSP.
• Tunnel configurations exist only on the tunnel head where the tunnel interface is defined. LSPs
traverse the network and involve tunnel heads, tunnel midpoints, and tunnel tails.
• Pointers in the tunnel table refer to corresponding entries in other MIB tables. By using these pointers,
you can find an entry in the mplsTunnelTable and follow a pointer to other tables for additional
information. The pointers are the following: mplsTunnelResourcePointer, mplsTunnelHopTableIndex,
mplsTunnelARHopTableIndex, and mplsTunnelCHopTableIndex.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 111
Table 10 mplsTunnelTable MIB Objects
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 113
Table 10 mplsTunnelTable MIB Objects
mplsTunnelLastPathChange .1.3.6.1.2.1.10.166.3.2.2.1.31 The time since the last change to the actual
path for this tunnel instance. A path change
inside a tunnel should change the value.
The value is reflected in the output of show
mpls traffic-eng tunnels tunnel.
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 115
116 MPLS MIBs
Chapter 11 FTOS-supported RFCs
The FTOS implementation of MPLS is in accordance with the following RFCs:
E-Series
RFC Title ExaScale
2702 Requirements for Traffic Engineering Over MPLS 8.3.1.0
3031 Multiprotocol Label Switching Architecture 8.3.1.0
3032 MPLS Label Stack Encoding 8.3.1.0
3209 RSVP-TE: Extensions to RSVP for LSP Tunnels 8.3.1.0
3630 Traffic Engineering (TE) Extensions to OSPF Version 2 8.3.1.0
3784 Intermediate System to Intermediate System (IS-IS) Extensions for Traffic Engineering (TE) 8.3.1.0
3812 Multiprotocol Label Switching (MPLS) Traffic Engineering (TE) Management Information 8.3.1.0
Base (MIB)
3813 Multiprotocol Label Switching (MPLS) Label Switching Router (LSR) Management 8.3.1.0
Information Base (MIB)
4090 Fast Reroute Extensions to RSVP-TE for LSP Tunnels 8.3.1.0
4379 Detecting Multi-Protocol Label Switched Data Plane Failures (MPLS TE/LDP Ping & 8.3.1.0
traceroute)
5036 LDP Specification 8.3.1.0
5063 Extensions to GMPLS Resource Reservation Protocol (RSVP) Graceful Restart 8.3.1.0
MPLS Configuration Guide for FTOS version 8.3.1.0 Publication Date: December 21,2009 117
118 FTOS-supported RFCs