I don't know if it is normal, and cannot found allot of experience on this part of my problem from other Mikrotik Users.
We have 4 CCR on this moment:
- Router1: CCR1072
- Router2/3/4: CCR2116
Router 3 and 4 are used for PPPoE for customers, default route to router1 through OSPF and failover to router2.
Router1 and 2 are used for BGP with our Transit and Peering, and have a IBGP session together.
Also have router1 and 2 a internal VLAN (for public IP's, so no nat) in datacenter with VRRP interface, used for servers etc.
They have also connection tracking disabled.
Router1 has 4 transit providers and peers on 2 Internet Exchanges
Router2 has 1 transit provider (same as 1 of the 4 providers of router1), router 2 is mainly used for back-up function.
When we enable a transit or peering with route server on a IXP we get higher latency during updates and some packet loss.
Also the routes impacted by the new updatet (which get a new shorter) route have about 10/20 seconds no traffic.
I can understand that traffic is disrupted when a transit/peer is disconnected abruptly, but is it normal that it happens on connecting to a transit/peer?
1. Is there anything we can do to prevent this or to optimise this?
2. is it a good way/design to use router1/2, who are used for BGP, as gateway with VRRP for servers or is het recommend to add some kind of core routers for it with IBGP and use router1/2 only like edge routers.
I hope my question and explanation is clear, if not please let me know!