EMAIL SUPPORT

dclessons@dclessons.com

LOCATION

NZ

OTV Control plane over Multicast & Unicast Infrastructure

OTV Control plane over Multicast & Unicast Infrastructure

Posted on Jan 24, 2020 (0)

OTV Control plane over Multicast & Unicast Infrastructure

In OTV all edge devices must be adjacent to each other and form adjacency between them and then control protocol run between them in order to exchange the MAC address reachability information.

The above adjacency and MAC address reachability information can be achieved by two ways:

  • If the IP transport is multicast enabled, a specific multicast group is used to exchange the control protocol message between OTV edge devices.
  • If IP transport is unicast mode, then one or more OTV edge devices can be configured as an Adjacency server and all other OTV edge devices register themselves and form neighbour list to get the adjacency.

Let’s understand how control Plane is built in both cases.

Control Plane adjacency in Multicast enabled Infrastructure:

When transport is multicast enabled the control plane adjacency can be achieved by two method:

  • OTV Neighbour discovery
  • OTV MAC address advertisement

OTV Neighbour discovery:

In this, all OTV edge devices should be configured to join with a specific ASM (Any Source Multicast) group and by this they can receive and send any multicast packet for this group.

  • Each OTV edge device will send the IGMP report to join the specific ASM group which will be used to carry the control plane information. Here we are assuming this ASM group is group “G”. Now these OTV edge device will join this group as host and this works without configuring PIM and this Group is configured and associated to overlay interface.
  • Now any OTV edge device , here in this case Edge device of site-1 generates the OTV hello , now this hello message is sent to all edge device who have joined to the specific ASM multicast group. This is required to form the control plane adjacencies. To achieve this the OTV hello is encapsulated to OTV frame and in external IP header source IP would be IP address of Join interface and destination IP is ASM multicast Group IP address which is associated to overlay interface.
  • Now these Multicast frame traverse to multicast enabled transport and reach to each OTV edge device who has joined the Multicast group “G”.
  • Now the once the Hello is received to OTV edge device these edge device will decapsulate the hello packet and this result in creation of OTV control protocol adjacency between all edge device.  

OTV control plane is transparently enabled in background after creating OTV Overlay interface and does not require any explicit configuration.

MAC address Advertisement:

Once OTV edge is discovered each other now let’s see how MAC address reachability information is shared between them to form control plane adjacencies.

  • Let’s suppose on Site-1 three host with MAC A, MAC B, and MAC C on VLAN 200 are learned and its information is populated in MAC table.
  • OTV update packet is  created containing all three MAC address and which is further OTV encapsulated and is sent to Layer 3 transport , here destination IP would be specific ASP group and “ Group G” used for control protocol exchange.
  • Now these OTV update is sent across all OTV edge device who are part of this ASM group and upon receiving this OTV update the update is decapsulated
  • Once the OTV update is decapsulated, it is imported in MAC address table of edge devices

Control Plane adjacency in Unicast enabled Infrastructure (Adjacency-Server Mode):

In multicast mode, a single Hello or update packet reaches to all OTV edge device. unicast enabled infrastructure works same way as Multicast mode but in this mode, each OTV end device creates the multiple copies of control plane packet and unicast them to each OTV edge device. Due to this head-end replication behaviour, multicast enabled transport is very much used in OTV.

OTV Neighbor Discovery

In this Mode, an OTV edge device is configured with Adjacency server and all other OTV edge device needs to register to the Adjacency server, this is done by configuring Adjacency server IP address on each OTV edge device and all other OTV edge device send the Hello packet to Adjacency server.

Once Hello message is received by Adjacency server by all other OTV edge device, Adjacency server makes the list of neighbour that want to be part of same overlay which is named as unicast-replication list.

This Neighbour list is periodically unicast to all OTV neighbour listed in Neighbour list of Adjacency server due to which each OTV edge device are aware of all other OTV edge device.

In below figure OTV edge device at Site-1 is an Adjacency server and all other OTV edge device register to this Adjacency server which in turn send the entire neighbour list to each OTV edge device by means of OTV hello.

These Hello packer are sent to each OTV edge device according to neighbour list by performing the head-end replication, creating one copy of Hello message to each remote OTV edge device with source IP is IP address of the Join interface and destination IP would be IP address of each OTV edge device join interface.

MAC address Advertisement:

Once OTV edge devices are discovered MAC address advertisement is done in order to achieve the control plane adjacencies.

Let’s suppose the OTV edge device on Site-1 learns the MAC A, MACB, MAC C on VLAN 200.

These MAC address will be populated in MAC address table of OTV edge device, as soon as it is done, OTV edge device generates the copy of OTV update packet containing these MAC address for each OTV edge sites according to neighbour list and send to join interface.

Join interface will encapsulate this packet with IP packet with source IP “ IP address of its Join interface and destination IP is IP address of join interface of each OTV edge device and unicast it to Layer 3 Transport.

When Each Edge device receives the OTV update, decapsulate it and import it on the MAC table.




Comment

    You are will be the first.

LEAVE A COMMENT

Please login here to comment.