Re: FileSystem to raw device. Performance will Improve?

From: Nalini Vallampati <nalini.vallampati_at_daytonoh.attgis.com>
Date: 1995/10/16
Message-ID: <DGJy10.7qL_at_ranger.daytonoh.attgis.com>#1/1


>In article <45r6bh$68k_at_kelly.teleport.com> Larry Fishman writes:

<snip>

>Joel Garry (joelga_at_rossinc.com) wrote:
>: In article <00001a1b+0000356d_at_msn.com> toronix_at_msn.com (Robert Rowntree)
 writes:
>: >I would like to hear about experiences on possible performance
>: >improvements of setting up the database as a raw device vs.
>: >filesystem. Whats the advantages and disadvantages. Are most projects
>: >going filesystem(block devices) or raw devices.
 

>: While I've been watching this newsgroup, I've seen claims of performance
>: increases of -20% to +300%. Personally, I think a modest performance
>: increase is to be expected, but that is overcome by the possibility that
>: someone will come along and overwrite your data with a filesystem.
 

>: There is an interesting perspective on hot backups in the November, '95
>: Unix Review magazine.

You can call Oracle and ask for a paper called "Making the Decision to use UNIX Raw Devices" by Cary V.Millsap

(No.2 in the Oracle NPST Monograph Series)

This gives a good insight into whether or not UNIX raw devices are appropriate for your site. This paper talks about the confusion about the performance gains that you have when you use raw devices.

Nalini

Nalini Vallampati
UNIX Admin/Oracle DBA
AT&T Capital - GIS
Miamisburg, OH 45342 Received on Mon Oct 16 1995 - 00:00:00 CET

Original text of this message