This week I have been troubleshooting a Multicast problem, before Tuesday I knew the basics of it but I did not realize how deep that rabbit hole went, I can only imagine how deep it goes in CCIE level since it has not really crossed my path in any of my studies thus far. I thought I should write this blog post to share what I have learned this week about Multicast. I will also write another post on how to configure and troubleshoot it.
What is Multicast?
Multicast is a group communication where data packets are sent to a group of receiver/destination computers at the same time. Multicast is one to many or many to many real time communication protocol, where Unicast is one to one and Broadcast is one to all. Multicast is mainly used in IPTV (Netflix, Hulu, Prime Video).
Multicast Address Ranges
They use Class D Range – 184.108.40.206 – 220.127.116.11
18.104.22.168 22.214.171.124 – Reserved for Local Addresses
126.96.36.199 – 188.8.131.52 – Globally Scoped Addresses
184.108.40.206 – 220.127.116.11 – Source Specific Multicast Addresses
18.104.22.168 – 22.214.171.124 – GLOP Addresses
126.96.36.199 – 188.8.131.52 – Limited Scope Address (Similar to Private IP addresses but for Multicast)
How does Multicast Work?
The Receiver sends a packet to the Router asking to Join the Multicast Group. Only the clients that want to receive Multicast join what is called a Multicast Group. If the router doesn’t know about it it will send requests out its next hop interface.
Protocol Independent Multicast (PIM) Think of this like a multicast routing protocol sits on top of the Internet Group Management Protocol (IGMP) and builds a pipe back to the source from the destination. There are 3 versions of IGMP:
IGMPv1 – has a 60 second timer and continually asks.
IGMPv2 – can send an I want to leave the multicast group message to the router.
IGMPv3 – can include a source in the join packets.
If the multicast sender has multiple paths to the receiver it will send the multicast multi ways meaning the receiver will receive duplicate messages. The receiver then has to send the reply but it uses the unicast routing table to return the traffic and uses that interface.
Reverse Path Forwarding (RPF) is a check that the receiving device does before the sender sends anything so it knows how to get back to get back to the sender without receiving multiple multicast messages.
I am sure there are more than two types of multicast but all I have covered are Sparse and Dense mode.
Sparse Mode is like a Join Protocol, where traffic is not forwarded on a segment unless it is explicitly requested. Sparse mode is typically deployed where the receivers are sparsely populated over the network, so that most of the network bandwidth is conserved.
Dense Mode is like a flood and prune system, where everyone receives the traffic until they inform (through the prune system) that they do not want that particular multicast messages. Dense Mode is typically deployed in topologies where listeners are densely populated, but it can be a very chatty protocol.
You can setup sparse mode or dense mode on a per interface basis. Once they are setup and enabled interfaces can run sparse mode while others run dense mode.
A Networking genius (Denise Fishburne – https://www.networkingwithfish.com/ ) said on a training video I watched yesterday “Friends don’t let friends do dense mode”. Denise also recently started following my blog, which is a huge honor. I hope this post is doing some basic justice to Multicast. I am trying to share information I have gathered in the last couple of days. If anyone has any comments or further insight let me know.