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

Re: BDB database maintenance

Howard Chu wrote:
-----Original Message-----
From: owner-openldap-software@OpenLDAP.org
[mailto:owner-openldap-software@OpenLDAP.org]On Behalf Of Jeremy Ardley

I quote

Data Administration May Not Be Needed

Eliminating SQL often means eliminating or reducing database
costs, too. Berkeley DB installations seldom need a DBA. In
many Berkeley DB
deployments, the end user may not even realize that a
database component is
part of the product - how many cell phone users think about a
data table
lookup when searching for a number in their personal address book?

Berkeley DB operates without end-user backups, installation,
or tuning.

My question to the developers of back-bdb is why does the
back-bdb use of
Berkeley DB require end-user backups and tuning?  This includes the
requirement to mananage the log files and to perform database
recovery after
abnormal shutdown.

Is the Berkeley article just plain wrong?  Or is there
something special about back-bdb ?

There are countless things "special" about back-bdb. Please read the FAQ:

How many cell-phones do you know of with end-user selectable memory sizes,
disk capacity, and/or operating system version?

In the embedded application market, of course you can preconfigure
everything; because you only have one platform configuration to deal with.
Think about what you're asking for, do you really believe "one size fits

back-ldbm using Berkeley DB does not generate log files, but you still need
to tune the dbcachesize, and back-ldbm doesn't offer high concurrency or
transaction-protected updates. After an abnormal shutdown, you generally need
to delete the database and start over from scratch, because when it gets
corrupted there is no telling what's left inside.

  -- Howard Chu
  Chief Architect, Symas Corp.       Director, Highland Sun
  http://www.symas.com               http://highlandsun.com/hyc
  Symas: Premier OpenSource Development and Support

Actually my experience is just the opposite. Load testing a v2.0.27 server with back-ldbm caused slapd to "run away" (90% CPU). A kill -9 followed by a restart and all is OK.
The same experiment with v2.1.16 with back-bdb causes db corruption.
Platform - Solaris 8.

Dave Lewney
Principal Systems Programmer, Computing Service
University of Sussex, Brighton BN1 9QJ. Tel: 01273 678354 Fax: 01273 271956