I want to build something like the image.
The MLAG stack isnt any problem. But I want to know if I would use several VLANs should they all be under the VRRP interface?
Each vlan should have its own VRRP interface, on each router, as the default gateway for the VLAN. Make sure it’s a /32 for the VRRP address on all routers/vlans.
If you use new connection tracking sync with VRRP, you’d want the VRRP to be part of the same group for sync to work well. Other than group, the rest of VRRP default should be fine.
Go fully layer 3 and use iBGP between CRS<>CRS and eBGP for CRS to CCR. Enable ECMP using route filters on each device, set route learnt distance to 1 for both paths on each device to get even load balancing.
Remove MLAG/VRRP completely in this way.
Is it possible to run iBGP on CRS326 with full speed on ports L3?
Why do I run eBGP internal from CRS to CCR?
CCR is running full ipv4 table from transit on both side.
I was thinking using OSPF internal on CRS but don’t know if they can handle the routing speed or if they only manage L2 switching.
Where do I set distance 1? And why?
Thanks.
+1 to Amm0 solution as CRS will not have enough processing power to handle routing protocols. With 1 change of connection between CRS, why MLAG and not just LAG.
hi @ingdaka
why MLAG and not just LAG.
it’s the common method in utilizing vrrp switch stacking. as the @op first design proposal. it is a common setup in server farms.
from the server dual nic,
one leg goes to sw1,
the other goes to sw2.
in case sw1 offline, the server still online via sw2 in vrrp backup.
but, unfortunately vrrp is a little bit difficult to mixed with dynamic routing.