Hi,
I recently tried to build a setup as follows :
Tree : dc=mycompany,dc=com
Subtree 1 : dc=subtree1,dc=mycompany,dc=com
Subtree 2 : dc=subtree2,dc=mycompany,dc=com
Assuming that my needs are to store subtrees 1 and 2 in separate and
remote LDAP servers, I used the LDAP backend.
It worked pretty fine, and then I tried to check what was the
behaviour if a link went down, or a remote server was unavailable.
Testing, I saw that if I have a 10 seconds timeout for each server, if
they're both down, my server spends 20 seconds waiting for the
timeouts. I assume that's because the queries are done sequentially
and not in parallel.
Is there any way to be able to parallelize the queries and build a
"buffer" for each remote backend in order to reconstruct it later,
allowing a user not only to use each remote server, but also to find
which subtree a user belongs to, without having those kind of problems
?
My test case is pretty simple, but as the final implementation I want
to build involves a lot more remote LDAP servers, and as I can't
modify the architecture, I'm wondering what are the available
solutions.
Thanks,
--
Julien Cassignol