Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Mailing Lists -> Oracle-L -> RE: UNIX file system read/write locks

RE: UNIX file system read/write locks

From: Kevin Closson <kevinc_at_polyserve.com>
Date: Mon, 11 Sep 2006 07:47:26 -0700
Message-ID: <5D2570CAFC98974F9B6A759D1C74BAD0E5AFB3@ex2.ms.polyserve.com>

                 I haven't heard about CIO for Linux so I guess they don't have it.

...PolyServe direct I/O is free of write-ordering locks and mtime/atime updates
as is ODM and Quick I/O on Veritas SFRAC. I/O on these are fully concurrent, we
just down say CIO, because DIO that is not CIO is not really DIO...the CIO guys
lagged in the critical deliverable so when they finally produced it they had to
make it sound like it was something novel. That is, anythnig called DIO that is
not also inherently CIO is just a partial implementation. After all, we aren't
talking about splitting atoms. Oracle 6.0.26 on Sequent Dynix/ptx supported
async/direct I/O on ufs and it was fully concurrent...in 1991.

--
http://www.freelists.org/webpage/oracle-l
Received on Mon Sep 11 2006 - 09:47:26 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US