Network Engineering Stack Exchange is a question and answer site for network engineers. It only takes a minute to sign up.
Sign up to join this community
When it comes to OSPF and its use of multicast, am I correct in believing that IGMP forwarding and an IGMP querier is still required within the local segment?
My understanding would be without this, the frames would be treated as a broadcast and sent to everyone. Therefore removing the benefit of using multicast.
Based on this is there any mechanism for it to fall back to unicast, to prevent the above. Apologies if my understanding here is way off : )
In terms of what happens to OSPF hello/discovery multicast packets, you're right. The switches send them to everyone on the subnet/VLAN/broadcast domain, as IGMP snooping isn't happening for 224.0.0.0/24, just as Ron explained.
However - and I believe that's the part you didn't consider - a host can still discard these multicast earlier than if they were broadcast. If the host isn't interested in (any/this) multicast group(s), it can drop the ethernet frame (at L2) or the IPv4/IPv6 packet (at L3), without even looking at any inner payload. A broadcast packet however would have to be passed 'further up the stack' to see if it is of interest to any of the running applications/services/deamons.
To add to Ron's list of OSPF's network types/operating modes:
It's good practice to run edge networks connected to OSPF routers with passive interfaces, especially where the OSPF network type is "broadcast" by nature. For example, this is the case in the proverbial "25 client VLANs" of a large campus, each with a few dozen of hosts each at least:
passive-interface
The benefit of multicast is that hosts not subscribing to the multicast group ignore frames they receive that are sent to the multicast group. With broadcast, they cannot ignore the frames.
The OSPF multicast group is a link-local multicast that gets sent to every switch interface, even with IGMP snooping enabled on the switch. OSPF only uses multicast on broadcast networks, and it is necessary in order to use DR and BDR, which are only on broadcast networks. For point-to-point links, OSPF uses unicast, and for non-broadcast networks you use the neighbor statement and it uses unicast.
OSPF multicast uses the 224.0.0.5 and 224.0.0.6 addresses, which are in the link-local address range. Switches with IGMP snooping are supposed to forward all link-local multicast to every interface, regardless of IGMP snooping. This behavior is defined in RFC 4541, Considerations for Internet Group Management Protocol (IGMP) and Multicast Listener Discovery (MLD) Snooping Switches:
Packets with a destination IP (DIP) address in the 224.0.0.X range which are not IGMP must be forwarded on all ports.
This recommendation is based on the fact that many host systems do not send Join IP multicast addresses in this range before sending or listening to IP multicast packets. Furthermore, since the 224.0.0.X address range is defined as link-local (not to be routed), it seems unnecessary to keep the state for each address in this range. Additionally, some routers operate in the 224.0.0.X address range without issuing IGMP Joins, and these applications would break if the switch were to prune them due to not having seen a Join Group message from the router.
When it comes to OSPF and its use of multicast, am I correct in believing that IGMP forwarding and an IGMP querier is still required within the local segment?
OSPF's use of multicast is to a link-local scope group, with IP TTL set to 1 to ensure it says link-local. [ 224.0.0.5 and 224.0.0.6 (RFC2328 pp.185-186)]
OSPF aside, for the multicast pruning functions of switches to work, yes, something has to cause group membership reports. All IGMP joins (unsolicited membership report) will eventually expire (RFC 2236 IGMP v2, p.18). If the switch doesn't see periodic reports (ala IGMP snooping of membership reports), that multicast group may be subject to pruning, or flood to all ports. (depending on how your switch is configured)
