Date: Sun, 19 Sep 1999 12:28:41 -0700 (PDT) From: Matthew Dillon <dillon@apollo.backplane.com> To: sthaug@nethelp.no Cc: phk@critter.freebsd.dk, mjacob@feral.com, dg@root.com, grog@lemis.com, cvs-committers@FreeBSD.ORG, cvs-all@FreeBSD.ORG Subject: Re: User block device access Message-ID: <199909191928.MAA74072@apollo.backplane.com> References: <199909191900.MAA73792@apollo.backplane.com> <44387.937768408@verdi.nethelp.no>
next in thread | previous in thread | raw e-mail | index | archive | help
:> In fact, a memory-mappable buffered block device with write-through would
:> be much, much more useful to a database then a character device, and I
:> think it's only a two line patch to make mmap() work, and probably a
:> four line patch to implement write-through. It would be virtually
:> unbeatable...
:
:Possibly so. However, the database systems I'm used to would much rather
:control caching themselves rather than rely on the OS. This is a very old
:discussion - you'll find it in OS and DB papers at least 15 years ago.
:
:Finding the appropriate mechanisms that the OS can offer to the DBMS is
:*not* easy.
:
:Steinar Haug, Nethelp consulting, sthaug@nethelp.no
True enough. You have to implement both. A big time database will want
to do its own caching, but that's because they've spent the time and money
to build an infrastructure that can take advantage of it (e.g. in order
to perform a query synchronously when it is known that all the necessary
data is already in-core).
A big chunk of the consumer and freeware database market, however,
would find a blockdev with mmap() much more useful. I certainly would.
A caching web proxy would find the blockdev approach more useful.
-Matt
Matthew Dillon
<dillon@backplane.com>
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199909191928.MAA74072>