Users Guide
Weighted ECMP for Static Routes
Dell Networking OS also supports Weighted ECMP for static routes.
You can configure weights corresponding to the paths for a static destination. If all configured paths have weights, traffic
distribution is performed using the Weighted ECMP method with the RTM these passing weights to the FIB.
If all configured paths do not have weights, regular ECMP is used to determine traffic paths. Also, paths that are configured with
a weight value of 0 are explicitly excluded from Weighted ECMP calculations. The RTM does not inform the FIB about these
paths (next-hops).
NOTE: Dell Networking OS also supports a global configuration parameter to enable or disable Weighted ECMP for static
routes on the system.
The following example shows weighted ECMP configuration for Static Routes:
Dell(conf)#ip route 1.1.1.0/24 4.4.4.2 weight 100
Dell(conf)#ip route 1.1.1.0/24 6.6.6.2 weight 200
Dell#show running-config | grep route
ip route 1.1.1.0/24 4.4.4.2 weight 100
ip route 1.1.1.0/24 6.6.6.2 weight 200
Dell(conf)#ip route vrf test 1.1.1.0/24 4.4.4.2 weight 100
Dell(conf)#ip route vrf test 1.1.1.0/24 6.6.6.2 weight 200
Dell(conf)#
Dell(conf)#
Dell#show running-config | grep route
ip route vrf test 1.1.1.0/24 4.4.4.2 weight 100
ip route vrf test 1.1.1.0/24 6.6.6.2 weight 200
ECMP Support in L3 Host and LPM Tables
The L3 host and Longest Prefix Match (LPM) tables provide ECMP next-hop forwarding for destination addresses. You can
program IPv6 /128 and IPv4 /32 route prefixes to be stored in the L3 host table and move IPv6 /128 and IPv4 /32 route prefixes
between the host table and the LPM route table.
By default, IPv4 route prefixes are installed only in the LPM table and IPv6/128 route prefixes are installed only in the L3 host
table. In previous releases, the IPv6 /128 entries in the host table were not supported by ECMP.
NOTE
: When moving destination prefixes from the LPM to the host table, there may be a hash collision because the host
table is a hash table. In this case, a workaround does not exist for programming route entries in the host table.
NOTE: Before moving IPv6/128 route prefixes from the host table to the LPM table, you must enable LPM CAM partitioning
for extended IPv6 prefixes. See Configuring the LPM Table for IPv6 Extended Prefixes for more information.
Use the ipv4 unicast-host-route or ipv6 unicast-host-route commands to program IPv4 /32 or IPv6 /128 route
prefixes to be stored in the L3 host table. A warning message states that the change takes effect only when IPv4 or IPv6 route
prefixes are cleared from the routing table (RTM) using the clear ip route * command. The IPv6 /128 and IPv4 /32 route-
prefix entries that you move to the host table receive ECMP handling.
To verify ECMP support for IPv6 /128 route prefixes stored in the host table, use the show ipv6 cam command. The
command output includes the ECMP field with IPv6 neighbor addresses. 1 indicates ECMP handling of destination routes.
Dell# show ipv6 cam linecard 0 port-set 0
Neighbor Mac-Addr Port Vid EC
--------------------------------------------------
[ 132] 20::1 00:00:20:d5:ec:a0 Fo 0/16 0 1
[ 132] 20::1 00:00:20:d5:ec:a1 Fo 0/24 0 1
To re-enable programming of IPv6 /128 route prefixes in the LPM table, use the no ipv6 unicast-host-route command.
A warning message states that the change takes effect only when IPv4 or IPv6 route prefixes are cleared from the routing table
(RTM) using the clear ip route * command.
Equal Cost Multi-Path (ECMP) 352