|
|
sybperl-l Archive
Up Prev Next
From: "Lee, William (ETS, EAS)" <WYLee at exchange dot ml dot com>
Subject: RE: Question about RAID levels using Sybase 11.9.2....
Date: Jul 7 2000 12:06PM
It all depends how much money you are willing to pay.
RAID 0+1: faster than RAID 5 but you will have to pay double price for your
disk
RAID 5: slower than RAID 0+1 (this might not be true with today's 64MB
cache inside RAID hardware, and also depend on how is your data get
accessed). good for big database server, for example > 100 GB, especially
for data warehouse.
William
> -----Original Message-----
> From: Patrick S. Leung [SMTP:pleung@easygroup.com]
> Sent: Friday, July 07, 2000 6:09 AM
> To: SybPerl Discussion List
> Subject: Re: Question about RAID levels using Sybase 11.9.2....
>
> I worked for an US investment bank before, and there they used mirror for
> their
> sybase. They are on Solaris. I never tried to find out why they weren't
> on
> RAID. It may be the fact that they use raw device and RAID may prevent
> recoveribility? Just my speculation, please correct me if I am wrong.
>
> Patrick
>
> Robert Banniza wrote:
>
> > Guys,
> > I have a question here and I don't mean to start a religuous
> RAID war
> > either. I got into a discussion with my company's data center manager
> the
> > other day and we started talking about RAID. One thing led to another
> and I
> > told him that I was going to build a particular Sybase installation
> using
> > RAID-5 on a Sun E450 (using the built in hardware RAID controller). The
> next
> > thing I know, this guy starts telling me how stupid of an idea this is
> and
> > the only way to build a database installation is using RAID-1. In the
> shops
> > that I have worked in as either a programmer or sysadmin. in the past,
> we
> > have always used RAID-5. I did my research and several sites were
> pointing
> > to using RAID 0+1 (striping+mirroring). Can someone tell me what RAID
> level
> > most Sybase shops use? This particualr Sybase installation will be
> writing
> > from some big files with more a lot of reading (BulkStats project).
> Either
> > way, I would appreciate any information as I want to know the best way
> to do
> > this and the reasoning behind it. I'm looking for ease of data recovery
> and
> > performance whilst configuring my RAID levels. Thanks for any help.
> >
> > Robert
> >
> > P.S. I know this is a Linux/Sybperl list but I have gotten some great
> > information in the past from here.
|
