MDT scale for IP Fabric depends on total VxLAN scale. Each VxLAN tunnel consumes two types of hardware resource: Tunnel Termination and Encapsulation.
IP Fabric with Optimized Replication also contains unicast VxLAN tunnels. The hardware resources are shared between unicast and Multicast tunnels, with FCFS policy.
The following table shows the total hardware resources.
| Max limit | |||
|---|---|---|---|
| SLX 9150/Extreme 8520 SLX 9250/Extreme 8720 | Tunnel Termination | 256 | FCFS (Unicast + Multicast Tunnels) |
| Encapsulation | 16k | 15.5k LIF + 512 FCFS Unicast/Multicast | |
| SLX 9540/SLX 9640 | Tunnel Termination | 512 | FCFS (Unicast + Multicast VxLAN) |
| Encapsulation | 96k | FCFS (LIF + L3 + VxLAN) |
Total unicast VxLAN tunnels in a Fabric depends on the number leaf nodes. Each leaf node has unicast VxLAN tunnels. Each unicast VxLAN tunnel consumes 1 Termination and 2 Encap entries. Depending on size of the fabric, MDT can scale based on the remaining hardware resources.
The following example shows how an MDT scales when there are no unicast tunnels and all resources are available for multicast tunnels.
1 MDT = 1 Multicast VxLAN Tunnel = (1 Termination) + (1 Encap per Spine)
SLX 9250: