I use this personal notes for my own study purposes, this method using bullet-points is what makes me comfortable about the topics I study, but the same doesn’t mean for you because everybody has his own method of studying.
More notes about MPLS-VPN are to be pasted here…
Three control plane protocols define the MPLS functionality:
1) TDP – Cisco proprietary, rarelly used nowadays.
2) LDP – Standardized by IETF. The most used protocol nowadays.
3) RSVP – mainly used for Traffic Engineering in MPLS.
LDP neighbor discovery:
– UDP port 646 to 126.96.36.199
– TCP port 646 to remote LDP Router-ID
One of the most famous MPLS application is MPLS-VPN.
– Provider Edge (PE)
– Customer Edge (CE)
– LSR (Label Switch Router)
– Core of MPLS should have reachability to all networks including loopback’s of PE’s within the Core using an IGP protocol (OSPF, IS-IS, EIGRP, or static).
– MPLS Router-ID is recommended to be a loopback, especially on PE devices.
– This loopback ip address will be also used as a Tunnel end point on PE-PE in MP-BGP.
– Multiprotocol BGP is used as control plane protocol for different CE routes (VRF’s) on all PE’s.
– Route Distinguisher (RD) – represents a unique route (network) within MP-BGP. A unique route has the following format: ASN:ASN:xxx.xxx.xxx.xxx/32.
– Route Target (RT) – defines the VPN membership.
– Import and Export policies define which Routes will be installed on local and/or remote PE’s VRF routing table.
– Distribution from Mp-Bgp to Ospf and vice versa has some extra Extended Communities (OSPF Route type, OSPF Domain ID, OSPF Router ID). If OSPF domain ID doesnt match with local OSPF process number it shows the redistributed routes as External (E1, E2). But if the OSPF domain ID and the local OSPF process number match than the routes show as Inter or Intra area routes (in case we use Sham-Link).
– In Mp-BGP OSPF domain ID (hexadecimal) is inherited from the OSPF process number.
– OSPF will always choose routes from Area 0 over all other areas even if it is has better cost, route types (internal, external) etc.
NOTE: VRF’s allow PE’s to store routes learned from differenet CE’s even if the prefixes overlap. RD allows the PE to advertise these routes (VPNv4) as unique prefixes even the IPv4 prefixes overlap with each other. Finally the RT tells the PE which routes should be added to which VRF, which provides a great control and ability to allow sites to be reachable from multiple VPN’s.