A
Alex Smith
I wrote an LDAP framework in Java and now I have it talking to AD. The
framework does not have a notion of a fail-over in a sense that it
only connects to a single LDAP server with no provision to continue
the conversation with another directory. It does have a notion of a
connection pool so that a particular connection will be tested before
or after it's used. If the LDAP server or network doesn't respond for
a while but then comes back up, the application using the framework
will recover gracefully. If my LDAP server is AD and I've got a bunch
of them merrily replicating in a domain, what's the best way to
achieve fail-over without adding this logic on the client side?
I thought I could get away with a round-robin DNS scheme but after
reading Q247811 I am not so sure. My second thought is to use a
Layer-4 switch (Alteon, Cisco, others). Will either of these work or
are there better approaches?
Alex Smith
ACME Consulting
framework does not have a notion of a fail-over in a sense that it
only connects to a single LDAP server with no provision to continue
the conversation with another directory. It does have a notion of a
connection pool so that a particular connection will be tested before
or after it's used. If the LDAP server or network doesn't respond for
a while but then comes back up, the application using the framework
will recover gracefully. If my LDAP server is AD and I've got a bunch
of them merrily replicating in a domain, what's the best way to
achieve fail-over without adding this logic on the client side?
I thought I could get away with a round-robin DNS scheme but after
reading Q247811 I am not so sure. My second thought is to use a
Layer-4 switch (Alteon, Cisco, others). Will either of these work or
are there better approaches?
Alex Smith
ACME Consulting