[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: how about back mysql


----- Original Message -----
From: "rj" <reetijaiswal01@rediffmail.com>
To: <openldap-software@OpenLDAP.org>
Cc: "Howard Chu" <hyc@highlandsun.com>; <mitya@seismic.geol.msu.ru>
Sent: Monday, December 03, 2001 10:08 AM
Subject: how about back mysql

>  Hi,
>  I am using back ldbm with  BDB 2.4.14 of sleepycat
>  which comes with the default rpm on Red hat Linux 6.2.
>  what i wanted was to setup multiple ldap servers on
>  different machines and all of them contacting one
>  machine called the data server, for the database files
>  of ldap(in this case .dbb files).

why would you prefer this to replication scheme using slurpd?..

>  if not bdb is it possible with mysql.

of course it is, like with any other multi-user RDBMS... in fact, this is
one of main purposes they are built for ;). and it should also work much
more effective than locking the whole db file...

> if yes
>  will i be having any performance problems.

it depends very much on what and how will you be mapping from RDBMS to
LDAP - see docs, faqs and maillist archives...

1) you can take a look on latest updates to back-sql which are available in
development branch in CVS - there are a lot of performance-related changes
2) use Postgres or any other RDBMS supporting stored procedures instead of
mySQL if you want write support

>  since i am using version 2 of ldap using schemacheck off
>  option of slapd.conf i can remove the issue of storing object class in
ldap for each enty. thus eliminating the requirement of > multiple attribute
values.and i have no other data which has multiple values for a single
attribute right at this moment.

won't you need it when searching, to distinguish between objects of
different type?
nevertheless, if you are going to use back-sql, the objectclass is a
cornerstone of it's algorythm...