Fabric SPB (EN)

Multicast with IS-IS

Why choose the SPBM Fabric for your multicast deployment ?

Introduction

Multicast is widely used in enterprise networks. Whether you are a bank and need to deliver financial data from a unique source to all your employees, or a healthcare institution where multicast is used by vital monitoring devices to centralize their data onto monitoring screens, multicast plays a critical role. If you are working in physical security you know how multicast is critical, as cameras rely exclusively on it for their streams. And if none of these examples were applicable to your business, imagine OSPF not working (it does use multicast) or imagine imaging your brand new computers one by one instead of doing them in parallel.

Multicast is everywhere, sometimes where we don’t expect it, and it is a crucial component of your network, even if your core business does not rely on it. Let’s see together how the SPBM Fabric can help you with your multicast deployment.

What is multicast and how does it work ?

“[…] multicast is a type of group communication where data transmission is addressed to a group of destination computers simultaneously.” Wikipedia

The sender, the source of the multicast stream, sends its data on a specific range of IP addresses from 224.0.0.0 to 239.255.255.255 (the infamous class D). The receiver will subscribe to this multicast stream using the IGMP protocol which will be interpreted by the network switches to establish multicast group membership and deliver the traffic accordingly.

While it works well within the local subnet, it requires an extra protocol to make it work when routing is required. That’s usually where PIM, and all of its variants, are used despite their complexity:

  • Choose a RP mode : auto, anycast, BSR, single static
  • Choose a PIM mode: sparse (SM), dense (DM), sparse-dense, bidir or even source-specific Multicast (SSM)
  • Add a mandatory IP configuration on each router.
  • And multiply this by the number of interfaces you want PIM on.

Also add to the equation that since PIM is protocol-independent (that’s what PI stands for), it relies on an underlay routing protocol to converge which is sub-optimal.

Or you could use SPBM, enable multicast in two lines, not configure anything on interfaces, and take advantage of the IS-IS routing protocol for its rapid convergence.

Why is the SPBM Fabric the right choice for multicast ?

Faster

In the Fabric, only one control plane protocol (IS-IS) exists, which allows convergence times in the event of a network failure to be typically sub-second. PIM experiences longer convergence times, in part, because unicast IP routing protocols must converge before PIM can converge. PIM also maintains the network state for every multicast group and uses a mechanism based on each hop to update the network about state changes, which affects scalability.

The best way to be convinced that multicast with the SPBM Fabric converges rapidly is to watch this video, demonstrating convergence time for a multicast stream in a virtualized environment.

Simpler

IP Multicast over Fabric Connect greatly simplifies multicast deployment, with no need for any multicast routing protocols such as PIM. A switch within the Fabric can forward a multicast stream anywhere in an SPBM network where IS-IS advertises the stream to the rest of the fabric.

The advantage of this solution over traditional approaches is the simplicity in provisioning and deploying IP multicast bridging and routing. Below are the only two lines required to enable multicast routing within the Fabric. It will enable multicast on the GRT, so all VLANs routed within the GRT will be able to send and receive multicast stream. It does not mean that the multicast stream will be broadcasted to all these VLANs, it means that with these commands, the VLANs in the GRT are now multicast-ready.

router isis
spbm 1 multicast enable

What would be the configuration if the VLANs were within a VRF (L3VSN) ?

router vrf my-vrf
mvpn enable
interface vlan 500
ip spb-multicast enable

And eventually, what would be the configuration if you only needed igmp snooping for a specific VLAN ?

interface vlan 500
ip igmp snooping

All these scenarios show how easy it is to configure multicast within the SPBM Fabric. Note that these configurations are only required on the switches where the multicast stream is sent or received. There is nothing to configured on switches in-between as the multicast stream will be associated with an I-SID. And as we have seen in a previous article, I-SID are propagated within the Fabric with LSP used in IS-IS.

So if your sender is on one side of the Fabric, and the receiver is on the other side, and your stream crosses a hundred switches, you will only need to configure multicast on the two switches where the sender and the receiver are.

You may have also noticed that no snoop-querier configuration is required. By default, the IGMP Querier uses the source address 0.0.0.0 for its IGMP queries. When a interconnection with an external multicast network is required and the layer 2 edge switches do not support a 0.0.0.0 querier (as for PIM), you can use a fictitious IP address as the querier address, and use the same address on all BEBs in the network.

Conclusion

Multicast can be complex to configure, but is part of almost every enterprise networks nowadays. Leveraging a technology like SPBM Fabric help you focus on your core business rather than the infrastructure required to make it work. Keep your network simple, free your time from complex configuration hassle, and take advantage of rapid convergence time, all with the SPBM Fabric from Extreme Networks.

Contact us to find out how to implement these features.

Follow us on LinkedIn to see our next videos.