Re: ORACLE AND UNIX QUESTION

From: Carlos Netto <Carlos.Netto_at_ibase.org.br>
Date: Tue, 11 Oct 94 00:32:00 -0300
Message-ID: <2aa.446.846%mpcbbs_at_ibase.org.br>


On 10 Oct 94 11:31pm, wrote to ALL:

> Newsgroup: comp.databases.oracle
> Organization: SAM Business Systems Ltd
 

> In article <outputCxFnIs.8C7_at_netcom.com>
> output_at_netcom.com "Output Services" writes:
 

>> In article <781720147snz_at_sambusys.demon.co.uk> psb_at_sambusys.demon.co.uk:
>> >In article <2aa.553.846%mpcbbs_at_ibase.org.br>
>> >           Carlos.Netto_at_ibase.org.br "Carlos Netto" writes:
>> >> I use raw-devices on a production machine. I don't know anything about
>> >> performance improvement. I choose to use raw-devices because it's possible
>> >> to loose a data-file when using fsck. But be sure, it's harder to
>> >> administrate
>> >
>> >That's like saying that I leave the lawnmower outside so that we don't trip
>> >over it in the lounge.  
>> 
>> I implemented raw FS at our site for the same reason.  After a crash,
>> fsck will come up with orphaned inodes that it will then "clear."
>> This looks awfully spooky to me.  Maybe I'm paranoid, and this is
>> perfectly ok, but I don't think so.  Altho, I'm still waiting for
>> someone to run newfs against an "unmounted" oracle partition and
>> waste the tablespace. :-)
>> 
>> Marty
 

> The point I was making was that there might well be good reasons for
> using raw partitions rather than filesystems but that is NOT a good
> one B_E_C_A_U_S_E (and perhaps I should have made this explicit) the
> tablespaces can go in their own filesystems away from all other OS or
> user files. Then there will be no inode creation activity or file
> enlargement on that tablespace and fsck will never barf on _those_
> tablespaces.

While we was writing about this discution, one customer lost a entire data-file (filesystem) without explication. I don't know if this file is in a own filesystem; I'll check it.

I understood your point of view, but I'm also paranoid. What about Unix's bugs?! What about an i-node block corruption?!

BTW, what's your point of view about performance issues on choosing filesystems or raw-devices? The only information I have is from Oracle's manual: 20% of improvement on raw-devices. I read other mails saying that we can use ASYNC I/O on filesystems but only SYNC I/O on raw-devices; is it true? I think that's not true on SCO Unix, right?

Thanks,

   ----->>>>>>>>>>>> Don't reply. Reply will not work! <<<<<<<<<<<<<<----


|  Carlos Augusto Leite Netto   | Voice  : +55.192.51.1153                  |
|  Software Design Informatica  | Fax    : +55.192.54.6518                  | 
|  Oracle VAR                   | e-mail : carlos.netto%mpcbbs_at_ibase.org.br |
|  Software Development & DBA   | FidoNet: 4:801/31 (Carlos Netto)          |
|---------------------------------------------------------------------------|
|  Softex 2000 member           | Brazil, Sao Paulo state, Campinas city    | 
=============================================================================
  • Evaluation copy of Silver Xpress. Day # 109 --- via Silver Xpress V4.01P [NR]

User Name: Carlos Netto - ( %mpcbbs_at_ibase.org.br ) This message was processed by RAuucp from Merlin Systems Inc.
Received on Tue Oct 11 1994 - 04:32:00 CET

Original text of this message