On Thu, Sep 11, 2003 at 11:24:59AM +0800, Fritz Mesedilla wrote:
> How does raid work?
>
> I saw this http://www.acnc.com/04_01_05_flash.html

Doesn't that URL explain the various RAID levels to you well enough?
It's a pretty comprehensive summary of the various RAID levels. :)

> And it said "Difficult to rebuild in the event of a disk failure (as
> compared to RAID level 1)"

You're probably comparing RAID 1 with RAID 5. Obviously, it's much
easier to rebuild RAID 1, which does plain mirroring. When one drive
dies, the other drive has a complete copy and you can operate fully with
it. When you replace the disk, the copies are mirrored onto it and
you're back with your redundancy.

With RAID 5 you have the parity bits spread out across your drives.
Let's take a minimum scenario, where you have three drives. When one
drive dies, you run in degraded mode where both active drives use the
parity information to "simulate" the existence of the third drive. If
another drive dies, you're dead. Things also slow down, since your
drives work overtime to cover up for your missing drive.

In case it isn't obvious yet, RAID 5 needs more drives than RAID 1 (RAID
1 can be implemented with a minimum of two drives, RAID 5 needs at least
three), but "wastes" less space.

> How does one recover from a raid 5 disk failure?  Can I just remove
> the defective drive while the server is running and "insert" a new
> one? (hot-swap)
> 
> Will it automatically recover itself?

When you replace the damaged disk, you instruct your controller to
rebuild the array. It will use the parity information on the surviving
drives to rebuild the data on the third. When the rebuild is complete,
you will run in normal mode and the speed penalty of running in degraded
mode will disappear. In typical operation you will not lose any data.

> From what I knew before... if I have 5 scsi drives, 4 are used while 1
> is used in case of disk failure.

This really depends on you. Assuming you're talking about RAID 5 (note,
we're talking about RAID levels, not number of drives), the parity
information is spread out across all your drives. You can have five
active drives in RAID 5, AFAIK, since it doesn't require an even number
of drives to operate. Alternatively, you can have a four-drive RAID 5
array, with one drive allocated as a hot-swap drive. A hot-swap is
activated to replace a faulty drive as soon as that faulty drive goes
offline. This reduces the time that your array operates in degraded
mode, since it rebuilds the array using the hot-swap as soon as the
defective drive goes down.

What do you plan to use your RAID array for? Depending on your
forecasted loads, storage needs and budget, you may want to look at RAID
10 instead of RAID 5. RAID 10 is faster than RAID 5 because it doesn't
have the overhead of parity calculation and storage. It benefits from
the striping of the mirrored pairs, which further improves both read and
write speeds. RAID 10 is also more fault tolerant than RAID 5. On a good
day, a four-drive RAID 10 array can withstand two downed drives,
assuming of course that they don't belong to the same mirrored pair.

 --> Jijo

-- 
Federico Sevilla III  : http://jijo.free.net.ph      : When we speak of free
Network Administrator : The Leather Collection, Inc. : software we refer to
GnuPG Key ID          : 0x93B746BE                   : freedom, not price.
--
Philippine Linux Users' Group (PLUG) Mailing List
[EMAIL PROTECTED] (#PLUG @ irc.free.net.ph)
Official Website: http://plug.linux.org.ph
Searchable Archives: http://marc.free.net.ph
.
To leave, go to http://lists.q-linux.com/mailman/listinfo/plug
.
Are you a Linux newbie? To join the newbie list, go to
http://lists.q-linux.com/mailman/listinfo/ph-linux-newbie

Reply via email to