Re: Question regarding MDB

Re: Question regarding MDB_NOLOCK

To: Hallvard Breien Furuseth <h.b.furuseth@usit.uio.no>

Subject: Re: Question regarding MDB_NOLOCK

From: David Barbour <dmbarbour@gmail.com>

Date: Fri, 30 Jan 2015 16:49:17 -0600

Cc: Howard Chu <hyc@symas.com>, OpenLDAP Devel <openldap-devel@openldap.org>

Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=Es6g51Wj20mlz0kxFwibRerQ2Zc2V5ls9IoXzoPSaPE=; b=xsfB85bDtAgWEt0TCc08e7UM0zV7fspmfSioIzxnM71kR92xbqw76HSKillC4CfaYA xaOwoDi0s2OPoWNsIiUBmytBJu2TZxlWZnvsmccW/VCVCfEiX7KeOUtFwTsmlIajlgRn njryO71sdl0EVuX18WUmsxuGLlWTDFMtqOyURVN+LmsARpy5PCNat2Y5C65HxIOX7n0K aLyS3U/gzgLhVV4K+0KDu2NiqY5WjRzqnGvUnJPLFyRVpHq3584XRTPq7kHASk/ciAfd fqgw2m6FL56rAxBuXvd3qBMr7Hv6B2k6Jyw3Sq2LMRI5Qvv/CAMK7AvB7wyiEpI2ZzBw cJeQ==

In-reply-to: <CAAOQMSt2M0XDW=BrNdcSiEgGV4xGDJWZxSh_5QJW6HS7jNYEBg@mail.gmail.com>

References: <CAAOQMSvoHLLH==PgjiQ_RQvfWQ265MCFC3SU2x+FQEcm8fWTQg@mail.gmail.com> <547C654E.8080300@usit.uio.no> <CAAOQMSt2M0XDW=BrNdcSiEgGV4xGDJWZxSh_5QJW6HS7jNYEBg@mail.gmail.com>

On Fri, Jan 30, 2015 at 3:57 PM, David Barbour <dmbarbour@gmail.com> wrote:

For my current use case, I believe that I can still achieve a sufficient level of parallelism even if limited to double-buffering (whereas two snapshots would give me triple-buffering). I'm not going to press for any changes at this time.

After having examined this further, I've changed my mind.

With triple buffering, I can guarantee that the writer *almost* never waits on a short-running reader, and that the readers never wait on the writer. With double buffering, the probability of the writer waiting on even short-running readers, assuming they are frequent, is nearly 100%. Triple buffering is thus a huge advantage for users of MDB_NOLOCK.

The update to support this is almost trivial: tweak `mdb_find_oldest` such that both meta-page snapshots are considered to have active readers. I'm willing to develop and submit a patch, but only if this change also sounds good to the main LMDB developers.

Regards,

Dave

Follow-Ups:

Re: Question regarding MDB_NOLOCK
- From: Howard Chu <hyc@symas.com>

References:

Re: Question regarding MDB_NOLOCK
- From: David Barbour <dmbarbour@gmail.com>