VxLAN Layer 2 gateway considerations and
		limitations
 
   
  
		
			
				- For the maximum number of tunnels supported
						are:
                        - 250 for SLX 9150, SLX 9250, Extreme 8520, and Extreme 8720
 
                        - 1024 for SLX 9540/SLX 9640
 
					
				 
				- It supports maximum of 64 ECMP paths. 
 
				- It do not support Layer 2 snooping. 
 
				- It also do not support VRRP source IP addresses and
					Multi-Chassis Trunks (MCTs). 
 
				-  The QoS DSCP field is not configurable for
					VxLAN tunnel and its default value is 0. The QoS DSCP for VxLAN at the ingress
					VTEP is derived from the User Packet. In the Egress VTEP, DSCP for the
					decapsulated packet is either taken from the Outer header DSCP or inner packet
					of DSCP value. 
 
					- The QoS TTL behaviour follows the Pipe model
					both at Ingress and Egress VTEPs. The QoS TTL field for VxLAN tunnel is not
					configurable. The default value is 255, which gets applied to the Outer header
					for the VxLAN encapsulated packets. 
 
					- MTU : MTU field is not configurable. The MTU
					is based on the IP interface MTU. If a packet is bigger than the IP interface
					MTU minus the VxLAN header, then the packet is dropped.
 
				- VxLAN tunnels have the standard UDP header
					encapsulated with the standard defined value of 4789. This value cannot be
					changed or modified. The SLX-OS expects VxLAN tunnel packets to be received with
					this value. 
 
				- VxLAN tunnels are supported in the default
					profile. 
 
				-  VxLAN tunnels are not supported when the
					counter profile 1 or 4 is configured. These profiles do not allocate hardware
					resources for TX statistics, which is needed for VxLAN tunnels. (SLX 9540 and
					SLX 9640 only).
 
				- Static VxLAN Tunnels are not supported on the
                SLX 9150, SLX 9250, Extreme 8520, and Extreme 8720 devices. 
 
				- The tunnel TX bytes statistics do not
					account for the outer VLAN header size. 
 
				-  When BUM packets received on a tunnel
					are flooded, split horizon drops the packets to the same tunnel. However, the TX
					Statistics counter on that tunnel increments. 
 
			
		 
		VNI Mapping
			
			VLAN-VNI mapping is shared by all VxLAN tunnels including the ICL. There are three methods of
				configuring VNI mapping.
			
				- Auto-Mapping
						- This is the default configuration and is
							recommended for use with the cluster. The first 32K are reserved for the
							VNI range. A one-to-one mapping between the VLAN/BD and the VNI is
							configured.
 
					
 
				- Manual Mapping for a specific VLAN/BD (hybrid mode)
						- With map vni auto configured, a specific VLAN/BD can
							be manually mapped. The manual mapping VNI range starts at 32768. Other
							VLANs/BDs will continue to use auto mapping.
 
					
 
				- Manual mapping of all VLANs/BDs (disable auto mapping)
						- When auto mapping is disabled, manual VNI mapping is required for all
							VLANs/BDs created on the system (even the VLANs/BDs that are not
							configured under EVPN). In this mode, the complete VNI range (1-2^24) is
							available for manual mapping.

Note   
								
									- When the VNI mapping is changed, traffic on the ICL is
										impacted. 
 
									- The VNI Mapping on the two nodes must match.