Round robin sounds good however it still hasn't solved my real issue.
What I want to achieve is failure-recovery, through Round Robin I can
achieve a bit load balance however when one server is down, the client
who happens to be RR'ed to query that server is going to get an error
response still. It seems the load balancing switch might be a good choice
since it would probably try to detect the status of those servers.
I know windows domain controller works this issue out with DNS pretty
well, it would periodically update its records in the DNS and the client
actually queries the DNS to get the list of healthy servers before really
contacting the server. I don't think this approach is available for
openLDAP server and other platforms however I might be wrong.
More comments?