diff options
author | Daniel Thorpe <1077065+dantho281@users.noreply.github.com> | 2021-02-11 02:25:57 +0000 |
---|---|---|
committer | GitHub <noreply@github.com> | 2021-02-11 02:25:57 +0000 |
commit | e88fba68357181bd54fcc7489cbba08780cee6cd (patch) | |
tree | b67e88b1208fa835edf0420a42dd2b624ec2105b /docs/configuration/protocols/mpls.rst | |
parent | dab473bfd04ab2930c043b853ba9995d1ff335e6 (diff) | |
parent | f33b0c78b07c80998d2c0e64d6a20bcb109f6db5 (diff) | |
download | vyos-documentation-e88fba68357181bd54fcc7489cbba08780cee6cd.tar.gz vyos-documentation-e88fba68357181bd54fcc7489cbba08780cee6cd.zip |
Merge pull request #1 from vyos/master
Update fork
Diffstat (limited to 'docs/configuration/protocols/mpls.rst')
-rw-r--r-- | docs/configuration/protocols/mpls.rst | 253 |
1 files changed, 253 insertions, 0 deletions
diff --git a/docs/configuration/protocols/mpls.rst b/docs/configuration/protocols/mpls.rst new file mode 100644 index 00000000..312a0df2 --- /dev/null +++ b/docs/configuration/protocols/mpls.rst @@ -0,0 +1,253 @@ +.. _mpls: + +#### +MPLS +#### + +:abbr:`MPLS (Multi-Protocol Label Switching)` is a packet forwarding paradigm +which differs from regular IP forwarding. Instead of IP addresses being used to +make the decision on finding the exit interface, a router will instead use an +exact match on a 32 bit/4 byte header called the MPLS label. This label is +inserted between the ethernet (layer 2) header and the IP (layer 3) header. +One can statically or dynamically assign label allocations, but we will focus +on dynamic allocation of labels using some sort of label distribution protocol +(such as the aptly named Label Distribution Protocol / LDP, Resource Reservation +Protocol / RSVP, or Segment Routing through OSPF/ISIS). These protocols allow +for the creation of a unidirectional/unicast path called a labeled switched +path (initialized as LSP) throughout the network that operates very much like +a tunnel through the network. An easy way of thinking about how an MPLS LSP +actually forwards traffic throughout a network is to think of a GRE tunnel. +They are not the same in how they operate, but they are the same in how they +handle the tunneled packet. It would be good to think of MPLS as a tunneling +technology that can be used to transport many different types of packets, to +aid in traffic engineering by allowing one to specify paths throughout the +network (using RSVP or SR), and to generally allow for easier intra/inter +network transport of data packets. + +For more information on how MPLS label switching works, please go visit +`Wikipedia (MPLS)`_. + +.. note:: MPLS support in VyOS is not finished yet, and therefore its + functionality is limited. Currently there is no support for MPLS enabled VPN + services such as L3VPNs, L2VPNs, and mVPNs. RSVP support is also not present + as the underlying routing stack (FRR) does not implement it. Currently VyOS + can be configured as a label switched router (MPLS P router), in both + penultimate and ultimate hop popping operations. + +Label Distribution Protocol +=========================== + +The :abbr:`MPLS (Multi-Protocol Label Switching)` architecture does not assume +a single protocol to create MPLS paths. VyOS supports the Label Distribution +Protocol (LDP) as implemented by FRR, based on :rfc:`5036`. + +:abbr:`LDP (Label Distribution Protocol)` is a TCP based MPLS signaling protocol +that distributes labels creating MPLS label switched paths in a dynamic manner. +LDP is not a routing protocol, as it relies on other routing protocols for +forwarding decisions. LDP cannot bootstrap itself, and therefore relies on said +routing protocols for communication with other routers that use LDP. + +In order to allow for LDP on the local router to exchange label advertisements +with other routers, a TCP session will be established between automatically +discovered and statically assigned routers. LDP will try to establish a TCP +session to the **transport address** of other routers. Therefore for LDP to +function properly please make sure the transport address is shown in the +routing table and reachable to traffic at all times. + +It is highly recommended to use the same address for both the LDP router-id and +the discovery transport address, but for VyOS MPLS LDP to work both parameters +must be explicitly set in the configuration. + +Another thing to keep in mind with LDP is that much like BGP, it is a protocol +that runs on top of TCP. It however does not have an ability to do something +like a refresh capability like BGPs route refresh capability. Therefore one +might have to reset the neighbor for a capability change or a configuration +change to work. + +Configuration Options +===================== + +.. cfgcmd:: set protocols mpls ldp interface <interface> + + Use this command to enable LDP, and enable MPLS processing on the interface + you define. + +.. cfgcmd:: set protocols mpls ldp router-id <address> + + Use this command to configure the IP address used as the LDP router-id of the + local device. + +.. cfgcmd:: set protocols mpls ldp discovery transport-ipv4-address <address> +.. cfgcmd:: set protocols mpls ldp discovery transport-ipv6-address <address> + + Use this command to set the IPv4 or IPv6 transport-address used by LDP. + +.. cfgcmd:: set protocols mpls ldp neighbor <address> password <password> + + Use this command to configure authentication for LDP peers. Set the + IP address of the LDP peer and a password that should be shared in + order to become neighbors. + +.. cfgcmd:: set protocols mpls ldp neighbor <address> session-holdtime <seconds> + + Use this command to configure a specific session hold time for LDP peers. + Set the IP address of the LDP peer and a session hold time that should be + configured for it. You may have to reset the neighbor for this to work. + +.. cfgcmd:: set protocols mpls ldp neighbor <address> ttl-security + <disable | hop count> + + Use this command to enable, disable, or specify hop count for TTL security + for LDP peers. By default the value is set to 255 (or max TTL). + +.. cfgcmd:: set protocols mpls ldp discovery hello-ipv4-interval <seconds> +.. cfgcmd:: set protocols mpls ldp discovery hello-ipv4-holdtime <seconds> +.. cfgcmd:: set protocols mpls ldp discovery hello-ipv6-interval <seconds> +.. cfgcmd:: set protocols mpls ldp discovery hello-ipv6-holdtime <seconds> + + Use these commands if you would like to set the discovery hello and hold time + parameters. + +.. cfgcmd:: set protocols mpls ldp discovery session-ipv4-holdtime <seconds> +.. cfgcmd:: set protocols mpls ldp discovery session-ipv6-holdtime <seconds> + + Use this command if you would like to set the TCP session hold time intervals. + +.. cfgcmd:: set protocols mpls ldp import ipv4 import-filter filter-access-list + <access list number> +.. cfgcmd:: set protocols mpls ldp import ipv6 import-filter filter-access-list6 + <access list number> + + Use these commands to control the importing of forwarding equivalence classes + (FECs) for LDP from neighbors. This would be useful for example on only + accepting the labeled routes that are needed and not ones that are not + needed, such as accepting loopback interfaces and rejecting all others. + +.. cfgcmd:: set protocols mpls ldp export ipv4 export-filter filter-access-list + <access list number> +.. cfgcmd:: set protocols mpls ldp export ipv6 export-filter filter-access-list6 + <access list number> + + Use these commands to control the exporting of forwarding equivalence classes + (FECs) for LDP to neighbors. This would be useful for example on only + announcing the labeled routes that are needed and not ones that are not + needed, such as announcing loopback interfaces and no others. + +.. cfgcmd:: set protocols mpls ldp export ipv4 explicit-null +.. cfgcmd:: set protocols mpls ldp export ipv6 explicit-null + + Use this command if you would like for the router to advertise FECs with a + label of 0 for explicit null operations. + +.. cfgcmd:: set protocols mpls ldp allocation ipv4 access-list + <access list number> +.. cfgcmd:: set protocols mpls ldp allocation ipv6 access-list6 + <access list number> + + Use this command if you would like to control the local FEC allocations for + LDP. A good example would be for your local router to not allocate a label for + everything. Just a label for what it's useful. A good example would be just a + loopback label. + +.. cfgcmd:: set protocols mpls ldp parameters cisco-interop-tlv + + Use this command to use a Cisco non-compliant format to send and interpret + the Dual-Stack capability TLV for IPv6 LDP communications. This is related to + :rfc:`7552`. + +.. cfgcmd:: set protocols mpls ldp parameters ordered-control + + Use this command to use ordered label distribution control mode. FRR + by default uses independent label distribution control mode for label + distribution. This is related to :rfc:`5036`. + +.. cfgcmd:: set protocols mpls ldp parameters transport-prefer-ipv4 + + Use this command to prefer IPv4 for TCP peer transport connection for LDP + when both an IPv4 and IPv6 LDP address are configured on the same interface. + +.. cfgcmd:: set protocols mpls ldp targeted-neighbor ipv4 enable +.. cfgcmd:: set protocols mpls ldp targeted-neighbor ipv6 enable + + Use this command to enable targeted LDP sessions to the local router. The + router will then respond to any sessions that are trying to connect to it that + are not a link local type of TCP connection. + +.. cfgcmd:: set protocols mpls ldp targeted-neighbor ipv4 address <address> +.. cfgcmd:: set protocols mpls ldp targeted-neighbor ipv6 address <address> + + Use this command to enable the local router to try and connect with a targeted + LDP session to another router. + +.. cfgcmd:: set protocols mpls ldp targeted-neighbor ipv4 hello-holdtime + <seconds> +.. cfgcmd:: set protocols mpls ldp targeted-neighbor ipv4 hello-interval + <seconds> +.. cfgcmd:: set protocols mpls ldp targeted-neighbor ipv6 hello-holdtime + <seconds> +.. cfgcmd:: set protocols mpls ldp targeted-neighbor ipv6 hello-interval + <seconds> + + Use these commands if you would like to set the discovery hello and hold time + parameters for the targeted LDP neighbors. + + +Sample configuration to setup LDP on VyOS +----------------------------------------- + +.. code-block:: none + + set protocols ospf area 0 network '192.168.255.252/32' <--- Routing for loopback + set protocols ospf area 0 network '192.168.0.5/32' <--- Routing for an interface connecting to the network + set protocols ospf parameters router-id '192.168.255.252' <--- Router ID setting for OSPF + set protocols mpls ldp discovery transport-ipv4-address '192.168.255.252' <--- Transport address for LDP for TCP sessions to connect to + set protocols mpls ldp interface 'eth1' <--- Enable MPLS and LDP for an interface connecting to network + set protocols mpls ldp interface 'lo' <--- Enable MPLS and LDP on loopback for future services connectivity + set protocols mpls ldp router-id '192.168.255.252' <--- Router ID setting for LDP + set interfaces ethernet eth1 address '192.168.0.5/31' <--- Interface IP for connecting to network + set interfaces loopback lo address '192.168.255.252/32' <--- Interface loopback IP for router ID and other uses + + +Operational Mode Commands +========================= + +When LDP is working, you will be able to see label information in the outcome +of ``show ip route``. Besides that information, there are also specific *show* +commands for LDP: + +Show +---- + +.. opcmd:: show mpls ldp binding + + Use this command to see the Label Information Base. + +.. opcmd:: show mpls ldp discovery + + Use this command to see discovery hello information + +.. opcmd:: show mpls ldp interface + + Use this command to see LDP interface information + +.. opcmd:: show mpls ldp neighbor + + Use this command to see LDP neighbor information + +.. opcmd:: show mpls ldp neighbor detail + + Use this command to see detailed LDP neighbor information + +Reset +----- + +.. opcmd:: reset mpls ldp neighbor <IPv4 or IPv6 address> + + Use this command to reset an LDP neighbor/TCP session that is established + + +.. stop_vyoslinter + +.. _`Wikipedia (MPLS)`: https://en.wikipedia.org/wiki/Multiprotocol_Label_Switching + +.. start_vyoslinter
\ No newline at end of file |