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

Home -> Community -> Usenet -> c.d.o.server -> Re: Async I/O

Re: Async I/O

From: <mark.powell_at_eds.com>
Date: Wed, 24 Mar 1999 15:56:28 GMT
Message-ID: <7db1v7$9gk$1@nnrp1.dejanews.com>


In article <36F720E7.45BD_at_oriolecorp.com>,   sfaroult_at_oriolecorp.com wrote:
> The problem of async I/O is a data integrity problem. The purpose of a
> RDBMS is to ensure that any committed transaction has indeed being
> written to disk - this is the purpose of the log writer. Basically, when
> you say 'commit', then the log writer writes the log buffer to the redo
> log file, and when it is safely written on disk you get a 'committed'
> acknowledgement. If you use async I/O it just means that you send a
> write command, and anticipate a successful return. The question is then
> what will happen if somebody unplugs your computer at this crucial
> moment? It depends on how your buffers are managed and how your hardware
> behaves (i.e. if it has some battery to cope with a power failure). This
> is why async I/O may indeed work well with anything, until ..., but also
> why Oracle support people will only recommend it when it's totally safe.
> Who do you think you are going to blame if you ever have a data
> integrity problem ?
> --
> Regards,
>
> Stéphane Faroult
> Oriole Corporation
> ------------------------------------------------------------------
> http://www.oriolecorp.com, designed by Oracle DBAs for Oracle DBAs
> ------------------------------------------------------------------
>

Note - Async I/O does not assume the I/O worked. It is non-blocking I/O that allows the program (dbwr) to not be suspended while the I/O is in progress. The program is responsible for checking on the status of the I/O. Unlike normal ufs I/O, Async I/O bypasses the Unix buffer pool so when Oracle receives notification that the write has completed it is to disk (or cache device) and not just to the buffer pool. This is more robust and secure than normal I/O. This is not just my opinion but is based on Ahmed Alomari's Oracle and Unix Performance Tuning, ch. 3, Configuring and Tuning I/O, where he explains the various and preferred methods of I/O as Async, Direct, Kernelized Asynchronous, List I/O, READV, and lastly normal Unix file systems.

Mark D. Powell -- The only advice that counts is the advice that  you follow so follow your own advice --

-----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own Received on Wed Mar 24 1999 - 09:56:28 CST

Original text of this message

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