[rescue] Small RAID array setup
kurt at csh.rit.edu
Thu Jun 26 16:05:47 CDT 2003
On Thu, 26 Jun 2003, C. Magnus Hedemark wrote:
> On Thursday, June 26, 2003, at 01:07 PM, Sheldon T. Hall wrote:
> > I would prefer security, capacity, and speed, in that order, for the
> > RAID
> > setup.
> RAID 1 is going to give you the greatest security, but the worst
> RAID 1 is not an option with more than 2 drives.
Not true. You can use as many drives as you want, but you only
get the capacity of the first drive. It's not common because of
expense, but you can do a 3 or 4-way mirror.
> RAID 0+1
> or RAID 10 (distinct technical difference but no real practical
> difference) have the same 50% available capacity, the same resiliance
> as RAID 1, and better performance.
Only argument I have here is the 50% number, only true if you
are doing a base 2-way mirror.
> You need an even number of drives to do RAID 1, RAID 0+1, or RAID 10.
> The fifth drive would be a paperweight unless called into duty as a
> spare when one drive failed.
Not true with them all being based upon RAID 1. Again, it's just
not commonly done because of expense.
> RAID 5 is less secure (you can only stand to lose one drive in the
> array) but gives you much more capacity. In your case, 80% of the
> space would be usable and the remaining 20% would be used for parity.
> Read performance would be fine but write operations will be slow.
> Especially if you do software RAID instead of hardware RAID. You need
> to determine how much writing you do to figure out if you can live with
> the performance.
> Many "sysadmin mill"-papered sysadmins will just use RAID 5 without
> considering the consequences. If you're running an ftp site where you
> get lots of downloads but few uploads, RAID 5 is likely fine. If you
> are running a file server that does lots of writing, you may find it a
> bit slow.
Very true. Some sysadmins choose it because they don't know... some
choose it because the higher-ups won't support the cost of a RAID-1/0+1/1+0
More information about the rescue