Re: Database vs Directory Service

[Dmitry Kovalev <mitya@seismic.geol.msu.ru>]

Robert Klerer wrote:

> Dmitry,
>
> Let me understand what you are suggesting.  I thought backsql was the code
> to attach slapd to a database.  It assumes that certain tables exist with
> certain attributes.

No, it does not assume any specific RDBMS schema. It is tunable to virtually
any schema you may have, that is the point of back-sql.
Setting up RDBMS as astorage for directory is rather pointless if you don't
already have the data there, or at least there are some arguments for such
point of view:

http://www.openldap.org/faq/data/cache/378.html

> If I wanted an LDAP front end to a already existing
> database of people without changing the tables, are you suggesting I write a
> new backsql to fit those tables.

no,  I am suggesting to tune back-sql to that database, by providing
appropriate metainformation. This includes some amount of SQL authoring (tha
amount depends on complexity of your data), no programming at all.

Actually, if you care too much about performance, it could lead you to writing
your own, application-specific backend, to serve only those queries that your
application makes, but serve them most efficiently.

>  What if I don't have the concept of object
> classes or parents or such what do I do when slapd what to perform those
> operations?

This is rather easy, please read "concept" file in back-sql directory of
OpenLDAP distribution - there is my attempt to explain how it is done in
back-sql.  There is a way to support general hierarchy, no matter how complex,
or you could use a simple trick with views if you need
a simple one.

You can also access this document through CVSWeb:

http://www.openldap.org/devel/cvsweb.cgi/~checkout~/servers/slapd/back-sql/docs/concept?rev=1.3&hideattic=1&sortbydate=0

WBW, Dmitry