Is there any reason why we wouldn't be able to have something like this:
Rserver redirect RS1
Webhost-redirection http://server1.site.com/%p 301
inservice
Rserver redirect RS2
Webhost-redirection http://server2.site.com/%p 301
inservice
Serverfarm redirect SF1
Predictor roundrobin
Rserver RS1
Inservice
Rserver RS2
Inservice
Class-map match-all REDIRECTVIP_CLASS
Match virtual-address 10.1.1.1 tcp eq 80
Policy-map type loadbalance first-match REDIRECTSF_POLICY
Class class-default
Serverfarm SF1
Policy-map multi-match INPUT_POLICY
Class REDIRECTVIP_CLASS
Loadbalance vip inservice
Loadbalance vip icmp-reply active
Loadbalance policy REDIRECTSF_POLICY
Interface vlan100
Description outside client facing
Service-policy input INPUT_POLICY
To permanently redirect users in a round-robin fashion to two servers? The URL that they originally go to would just be a first-hand-off; it would need to change in the redirect URL. So they may go to “http://server.site.com/login.cgi†and server.site.com has a DNS entry of 10.1.1.1. They would get round-robin redirected to server1.site.com or server2.site.com, which have different IP addresses, with the same URL path “/login.cgi.†It would be a permanent redirection (301).
The purpose would be to do simple load balancing at a site where there are no ACE modules, but where there is a connection with a large RTT to another site with an ACE. We wouldn't want to run everything through the ACE as normal due to the RTT.
How would you setup probes to make sure the redirect servers are available in this scenario?
Would there be a way to use sticky with this configuration, so that the redirects were not permanent, and the clients stick to the redirect server they initially are sent to?
Thanks,
