Re: Mysql as a backend for LDAP

[Rick van Rein <rick@vanrein.org>]

Hi Mike,

> > I think because RDBMS have established a sensation of speedy yet robust
> > storage, unlike LDAP which is relatively new and unknown to most.
>
> LDAP has been around since 1993 and in popular usage since 1995.

Most computer-related people know in detail about RDBMS.
Hardly anyone (in my vicinity, a university, a DB group) knows about LDAP.

LDAP is growing in popularity, and I suspect RDBMS are more stable in
popularity.  This causes a large number of newcomers, and many of them
will want to know the difference between LDAP servers and RDBMS.

> > It would be helpful if the qualitative remarks about this issue in the
> > OpenLDAP FAQ were strenthened with numerical data, that is, by benchmarks.
> 
> This is comparing apples to oranges. LDAP and RDBMS are not the
> same thing. Why would you compare apples to oranges as a point in
> an argument?

To avoid getting these questions over and over again.

Furhtermore, call it apples and oranges if you like, we _are_ talking about
two different backends (Berkeley DB and SQL), and these are able to store
the same data, in a completely different way.  A comparison between
performance of the backends helps to get a "feel" of OpenLDAP performance
on top of Berkeley DB.

If you can point people with this question to numbers that clearly state
"backend A is 4.7 times faster than backend B" then the discussion need not
be purely philosophical.  In the end, many of use just want performance,
not a system that clearly separates issues in the artistically correct way.
Also, these figures may be instructive to managers that fail to see the use
of porting the RDBMS data over to an LDAP server.


I stand by my point that a comparison between SQL and BDB backends is a
fruitful exercise.


Cheers,
Rick van Rein.