My company has purchased three ACE modules, two for the main data centre and one for a disaster recovery (DR) site.
There is a fibre link to the DR site which is configured with trunking for some vlans, but the bandwidth will not be sufficient to use the DR ACE as a failover for one of the ACEs in the main data centre.
The two ACEs in the main data centre are setup in fault-tolerant mode for the relevant context. Routed mode is currently implemented for the load balancing.
The DR plan is for servers to be brought up at the DR site with the same IP addresses, and for users to be able to connect to the same VIPs.
There is no point in the DR ACE maintaining state information with the main data centre seeing as the servers will anyway have to be recovered at the DR site. I am also quite happy to manually redo any configuration changes made on the main data centre ACEs on the DR ACE.
What is important is that the VIPs do come up automatically in a DR. I can manually add static routes and do a "no shutdown" on the context interfaces, by I would really like everything to happen automatically.
I know that the IOS SLB failover solution can do this because the failover works with HSRP and more than two 6500s can participate in HSRP failover.
Can anyone please tell me how I can achieve automatic failover of the VIPs in the event of a DR using the ACE modules? I can easily implement an HSRP design, or something that uses dynamic routing, or I can change the ACEs to one-armed mode if that will solve the problem.
There is a feature supported in cisco loadbalancers including ACE called RHI (Route Health injection). With RHI you can configure ACE to communicate VIP address as host route (/32) to the routed network based on the VIPs availability..
Suppose your application www.xyz.com resolves
to 12.x.y.z. then you can configure ACE in bot h data centers with same Virtual IP 12.x.y.z.
You can configure each ACE to inject 12.x.y.z/32 into the routing table.
Now you can redistribute the 12.x.y.z at DR datacenter with a high metric value. This way traffic for 12.x.y.z will only be routed to main Data center (due to lower route metric) and if the VIP is unavailable at the primary data center then ACE in primary DC will seize to advertise the host route 12.x.y.z/32.
This will ensure that users will hit the route with high metric and traffic will be forwarded to ACE in DR Data center.
Syed Iftekhar Ahmed
Thanks for the reply.
I did have a look at RHI. This doesn't solve the problem about having to do a "no shutdown" on the relevant "interface vlan" statements on the DR ACE though? Do you think that RHI would would work a one-armed mode?