>Hmm... so what's the purpose of chaining directory
>server? Wouldn't use a read-only replica better
>in this case?
>
>
>
Chaining is complementary to replication : chaining allows you to
leave the data where it is and take the query to the data,
which might be more efficient than moving the data.
It's also possible (by possible, I mean you might need to write code)
to play various proxy tricks when chaining :
multiplex operations onto a cluster of storage servers; re-write
the operations as they pass through; aggregate results from
multiple source servers; and so on.
Mind you, proxying and chaining are not commonly used in my experience,
compared to replication.