[prev in list] [next in list] [prev in thread] [next in thread] 

List:       evms-devel
Subject:    Re: [Evms-devel] evms+bbr better than plain RAID 1 ?
From:       Steve Dobbelstein <steved () us ! ibm ! com>
Date:       2004-04-02 15:36:48
Message-ID: OF6D5D0313.2B313644-ON85256E6A.0052CA49-06256E6A.0055C7CE () us ! ibm ! com
[Download RAW message or body]





Russell Poyner wrote:
> I am in the process of setting up a server for near-line backup of user
data
> on a small collection of work stations.  My original plan was to rsync,
or
> rdiff-backup (still deciding) the data nightly to a RAID 1 built from a
pair
> of 200G seagate ide drives using the md driver.  Then occasionally
archive
> data to dvd.
>
> Then I read this <http://linas.org/linux/raid.html>
> cautioning against the dangers of data loss due to bad block
accumulation.
> With RAID 1, do I really need bad block relocation as provided by evms?
> If so what are the implications for disaster recovery?
> 1. One of the RAID 1 drives fails
> 2. The drive holding my root filesytem fails (a separate drive)

Hi, Russell.

As far as handling a bad block error, RAID1 and Bad Block Relocation (BBR)
will accomplish the same thing for you, just in different fashions.  In
RAID1, the entire drive is brought offline; I/O will continue to go to the
remaining good disk(s) in the array.  With BBR, the bad block is simply
relocated to a reserved space on the disk.

BBR is generally a last line of defense against bad blocks.  Once BBR
starts detecting and relocating bad blocks it is a good idea to work on
replacing the disk.  BBR reserves a finite number of sectors.  Once those
are used up the BBR device will start reporting bad block errors.

Let's assume you have a drive that is beginning to degrade and will start
failing I/O to some blocks.  With RAID1, once an error is detected the
whole drive is brought offline.  Any further I/O errors the drive might
have are never given a chance to surface.  With BBR, subsequent errors will
use up the reserved sectors.  If the drive is going down quickly, BBR may
not give you enough time to replace the drive before all the reserved
sectors are used up.  RAID 1 won't have that problem.

Then there is the cost for the implementation.  RAID1 will require two or
more disks to protect one disk's worth of data.  BBR just carves off some
reserved sectors on the disk.  For the cost of an additional disk, RAID1
will give you full protection against all failures on a single disk.  BBR
will give you protection for bad block errors up to the number of reserved
sectors.  BBR does not give you protection against total disk failure.

I don't see any point in using BBR and RAID1 together.  If you are using
RAID1, you get all the protection that BBR can offer.

I'm not sure what you are asking in question 2.  The drive holding the root
file system will have as much protection as is built into the volume.  If
the volume is made on a RAID device then it will have whatever protection
is offered by the RAID personality.  If it is on a BBR device, it will have
protection for bad block errors up to the number of sectors reserved by
BBR.

Hope this helps.  I would be glad to address any further questions.

Steve D.



-------------------------------------------------------
This SF.Net email is sponsored by: IBM Linux Tutorials
Free Linux tutorial presented by Daniel Robbins, President and CEO of
GenToo technologies. Learn everything from fundamentals to system
administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
_______________________________________________
Evms-devel mailing list
Evms-devel@lists.sourceforge.net
To subscribe/unsubscribe, please visit:
https://lists.sourceforge.net/lists/listinfo/evms-devel
[prev in list] [next in list] [prev in thread] [next in thread]