I plan that as soon as ROS7 is stable use MLAG stacked CRS switch against my core routers.
How to think? VRRP? Has two CCR routers with each eBGP full table with iBGP in between.
How can I build the best redundancy with a stack of CRS MLAG underneath?

??
Thanks.

Hi, I am doing tests with 2 CRS in MLAG version RouterOS v7.1 stable
I have not encountered any problems so far
You have to bond the interfaces of your CCRs to their respective Switches
The peer port must have an untagged VLAN with PVID eg 200, on both sides
All VLANs must also be tagged on the peer port and bonding ports in the switches
Associate the port of the VRRP with the bonding ports of the respective CCRs and you will have redundancy
Hello

Hi, I am doing tests with 2 CRS in MLAG version RouterOS v7.1 stable
I have not encountered any problems so far
You have to bond the interfaces of your CCRs to their respective Switches
The peer port must have an untagged VLAN with PVID eg 200, on both sides
All VLANs must also be tagged on the peer port and bonding ports in the switches
Associate the port of the VRRP with the bonding ports of the respective CCRs and you will have redundancy
Hello

Sounds great.. not sure how to do the connection between CCR and CRS.. with bonding and VRRP.
Any one who knows how the eBGP announce of network work with VRRP?.. I cant announce from both?

Hi every CCR can announce your networks to peers. The VRRP you have to use for the redundancy of what you have downstream of the bgp peers, so usually the IP VRRP is the gateway of your network and you will associate it to the bonding interface previously created so as to also have redundancy on the switches

One observation would be to be careful if you plan to implement stateful firewall rules on the CCRs. I can’t think of a simple way to avoid some traffic asymmetry if your BGP border routers and client LAN routers are the same. connectlife is right that there isn’t really anything that VRRP should impact regarding BGP.

My other observation is about requirements for layer-2 redundancy. Do you have dual-homed downstream switches or dual-homed servers connected to the CRS layer? In my experience, MLAG designs are only worth the complexity and hassle when one of those two are a requirement.