I have following topology attached, could not inserted.
1.Cat 3750 stacks are transparent just extending OSPF vlan from site A to site B.
2.All Core switches are Cat 6509 running OSPF in area 0 in single vlan.
3.Core switches at Site A are running multigroup HSRP in load balancing mode for almost 20 vlans.
4.Core switches at Site B are running multigroup HSRP in load balancing mode for almost 20 vlans.
5.In OSPF i have advertised the physical ip address of vlan interfaces ( not the VIP ) , i guess this is correct !! If anybody has suggestions,tips they are always welcome
6.Now trick is ( and i guess this is common design concern ) i get multiple OSPF routes (equal cost)to each vlan in Site A from site B and vice versa.
7.Everytime i do traceroute from Site B to Site A VIP, it keeps alternating first hop as Core-1 or Core-2
8.So OSPF is load balancing.
9.I am not sure whether this is resulting in asymmetric routing, as i did not have hosts/applications to test it .
10. I tried tweaking the OSPF cost for each vlan on standby router ( increased OSPF cost to 10 )
11. This resulted in consistent(first hop) traceroutes to each VIP from other site.
12.So i guess tweaking ospf cost has stopped load balancing and ultimately any chances of asymmetric routing !!!
I need confirmation on above points, any enhancements, tips, corrections.
what you see is not a problem at all, it can be seen as asymmetric routing but if no firewall is involved it is not an issue.
it can be more a problem to have tweaked the metrics of IP subnets that the fact that return traffic can come back on the other core switch.
HSRP just provides a VIP it is a first hop redundancy protocol useful for end user PCs and so on.
return traffic can come back on the other core with no problems.
I would keep normal OSPF metrics. Load balancing can be a benefit
Hope to help