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

RE: back-bdb performance

> -----Original Message-----
> From: storm@carnelian.propagation.net
> [mailto:storm@carnelian.propagation.net]On Behalf Of Markus Storm

> Do you also have any figures on concurrent write access ?
> back-bdb uses BDB's Transactional Database Store while back-ldbm
> uses BDB's Concurrent Database Store which has a single-writer
> bottleneck.
> So I hope back-bdb will allow for faster writing which IMHO would
> be the main advantage over back-ldbm.
> regards,
> Markus
> PS: bdb-4.0.14 is out.

I will have to take a look at that. I found a couple bugs in 3.3.11.
Thanks for the notice.

Concurrent write access is a problem because of the possibility of
deadlock. Using the Concurrent Database Store may seem to cause a
bottleneck, but it may be faster overall because it avoids deadlock.
When I split my 10000 entry ldif file into two separate files and
spawn two ldapadd commands on the same server, back-bdb time goes to
1 or 2 minutes (up from 30-some seconds) depending on the deadlock
detector. Back-ldbm is still around 46 seconds.
  -- Howard