Hardware Forwarding Limitations
All switches can use hardware forwarding when the route mask is 64 bits or less. If the route mask is greater that 64 bits, limitations apply based on the hardware platform.
-
Fewer slices are available for ACLs.
Use show access-list usage acl-slice to check if any slice is unused. If no slice is available, consider disabling a feature that is consuming ACL slices if that feature is not required. Features that are enabled by default such as IGMP Snooping or MLD Snooping can be disabled to free up ACL resources if not required.
- Table-full messages appear when there is no more space in the GT64B table.
- If an eligible route cannot be added to the GT64B table (because the table is full), there is no guarantee that traffic for that route will be properly routed.
- If enabled, route compression for IPv6 can make room for additional routes by reducing the number of entries in the GT64B table.
- When an IPv6 address with a mask greater that 64 bits is configured on a VLAN or tunnel, that address is automatically added to the GT64B table.
All switches support hardware forwarding routes with masks greater than 64 bits. For number of routes supported, see the limits for the particular switch in the ExtremeXOS Release Notes.
Starting with ExtremeXOS 22.5, the ipv6-mask-length option provides greater hardware route scale and IP route sharing (ECMP) support for IPv6 “long-mask routes”, meaning IPv6 subnets with mask lengths 65–128 bits. This provides additional scale and resilience for IPv6 host routes whose mask length is 128 bits. Increasing scale and providing ECMP for IPv6 mask 65–128 routes decreases IPv4 route scale. The default IPv6 mask length is 64.
The configure forwarding internal-tables [ l2-and-l3 | more [l2 | l3-and-ipmc | routes {ipv6-mask-length [64 | 128]}]] command provides the ability to support additional L2 and L3 hosts, routes, or multicast table entries.
- If no space is available in the hardware table, there is no guarantee that traffic for that route will be properly routed.
- If enabled, route compression for IPv6 can make room for additional routes by reducing the number of entries in the hardware table.
- When an IPv6 address with a mask greater that 64 bits is configured on a VLAN or tunnel, that address is automatically added to the hardware table.