Sunday, May 18, 2014

mLDP Signalling: In Band and Out of Band


In my previous post, I discussed about the overall architecture of mLDP. This post is more focused on mLDP Signalling. mLDP signalling provides two functions
1. FEC Discovery for a MP LSP
2. Assigning multicast flow to a MP LSP

mLDP can use two signalling methods; In-band Signalling and Out-of-Band Signalling. FEC uniquely defines the MP LSP within the network by using combination of VPN-ID and it’s opaque value. The signalling maps the streams that will run over that MP LSP.
• In-Band Signalling
Opaque Value is used to map an MP LSP
Opaque value is derived from the multicast flow

• Out-Of-Band Signalling
 Uses an overlay protocol to build the tree
 Opaque value are assigned by statically configured
 MP LSP creation is on-demand

In-Band Signalling Operation
It is called in-band signalling because the Egress PE uses the multicast stream information to create the Opaque Value and the Ingress PE uses this Opaque value to learn what multicast flow to send on the MP LSP.
1. The Egress PE receives a (S, G) IGMP join from a receiver. It creates a label mapping message containing the FEC TLV with the Opaque Value and a label TLV based on this information. The root address in the FEC Element will be derived from the BGP next-hop or (S). All ingress PE routers will create exactly the same FEC Element.


Debugs of inband signalling:-
mpls_ldp[1042]: DBG-mLDP[4142372544], Peer(9.9.0.7:0)/Root(9.9.0.3): ldp_mldp_sig_peer_msg_rx_nfn: 'Label-Mapping' msg (type 0x400), msg_id 167, lbl 23, MP2MP-Dn FEC { root 9.9.0.3, opaque-len 14 }, no MP-Status, status_code 0x0
2. The Egress PE then builds an MP LSP towards the Ingress LSP (root of the tree) using the label mapping message with the downstream label( lbl 23, MP2MP-Dn FEC { root 9.9.0.3, opaque-len 14 }). At each hop along the way, the P routers will use the same FEC Element, but the downstream label will change. When the ingress PE receives the label mapping message, it parses the Opaque Value to extract the multicast stream information and creates the appropriate (S, G) state and mapping information.
3. When the Ingress PE receives the (S, G} stream it will then forward it onto the correct MP LSP using it’s label information.


Out-of-Band (Overlay) Signalling Operation
Overlay protocol signals the mapping of the IP multicast flow to a MP LSP.
1. The Egress PE creates the FEC through a static procedure and builds the MP LSP hop-by-hop to the root based on FEC information.
2. Using overlay signalling, the Egress PE signals the Ingress PE to forward the IP multicast stream over the MP LSP with the unique FEC value.
3. The Ingress PE then forwards the (S, G) traffic onto the tree using the mapping information.

Click Here To Read Rest Of The Post...

Saturday, April 19, 2014

MVPN Over MLDP


In legacy MVPN scenarios there will be a GRE encapsulated MDT tunnel is created among the PEs and the entire customer multicast traffic of a particular VPN will be encapsulated with the MDT group address and tunneled it. In MLDP scenario there won't be any PIM configured in the core and the multicast packets are MPLS encapsulated. This makes the core PIM free.

In MLDP cases there will be a VPN-ID used instead of MDT default multicast group address. The default MDT will be created using MP2MP LSP to support low bandwidth and control traffics between the VRFs. The data MDT will be created using P2MP LSP to support high bandwidth traffic from a particular source. There will be PIM neighbor relationships between the VRFs which are seen in LSP-VIF. At the edge towards the customer (CE) PIM multicast routing will be enabled for that VRF.

The Default MDT will be created using MP2MP LSP. Static root has to be defined for MP2MP LSP. Normally two roots will be configured for redundancy. MDT will use LSP-VIF to transmit VPN multicast packets, i.e., PIM will run over MP2MP LSP.

In below depicted figure there will be a MP2MP LSP created with the root at PE1. The opaque value for MVPN is in the format of (VPN-ID). Both MP2MP upstream and downstream LSP will be formed and the labels will be exchanged for the data transfer.

!To create LSP in IOS
ip multicast vrf A mpls source Loopback0
ip vrf A
rd 65001:1
vpn id 65001:1 --- For creating Opaque value
route-target export 65001:1
route-target import 65001:1
mdt default mpls mldp 1.1.1.1 - Root node address

!To create LSP in IOS-XR
multicast-routing
vrf ABC
mdt default mldp ipv4 9.9.0.3
!
mpls ldp
router-id x.x.y.y
mldp
!
vrf ABC
vpn id 65000:1
address-family ipv4 unicast
!
LDP capabilities are extended to support multicast over LSP. LDP capabilities are advertised through LDP message TLVs during LDP initialization. mLDP defines two new FEC elements:-
1. P2MP : TLV 0x0508
2. MP2MP : TLV 0x0509
In mLDP, egress LSR initiates the tree creating by looking into root address. The root address is derived from BGP. Each LSR in the path resolves the next-hop address of root and send the label info. mLDP Control Plane:-

mLDP Data Plane:-


Click Here To Read Rest Of The Post...

Friday, February 7, 2014

Control Plane Flow and Data Plane Flow - BGP RFC 3107


BGP 3107 Architecture for Large ISP-Telecoms covers how BGP RFC 3107 help service providers to share the VPN routes across the boundaries without sharing the PE routes. ABR(Area Border Router) plays a vital role, as it has to modified the BGP next hop attribute to self.



Once the iBGP next-hop is changed, there after we can see how the control and data flow will look like.


Control Flow



Next step is to see how the data flow will look like


Click Here To Read Rest Of The Post...