From: Tampa Bay (ggelaude@TAMPABAY.RR.COM)
Date: Wed Feb 11 2004 - 22:54:53 EST
Hi Alan,
A lot depends on your database and or application. I have a 60+ GB D3
(Pick) database with raw lvm partitions Raid5 that requires rebuilds on a
regular basis. The RAID5 is on the new 15000 rpm 36GB SSA drives which are
extreamly fast. D3 powers an automated process that does approx 18,000
seperate orders (as many as 5 to 10 persecond during the busy times) per
day, with 200+ verry active support users.
I have a 200+ GB NFS Humingbird Miestro share (not sure of the exact size as
it is not my domain) that was configured very poorly way before my time.
Large 100 to 200 MB files to thousands of 1 or 2 K files. Very fragmented
and not under my control. If I had my way it would be on a Samba share and
defragemnted regularly. That would solve a large number of residual
problems.
IMHOP If your data has a lot of update activity it doesn't mater that you
have the load spead across multiple drives. Plain Raid5 or stripped,
fragmented data causes longer searches and can cause premature equipment
failure through what is called disk thrashing. A fragmented drive is like
sending a letter from California to New York via Detroit back to Denver via
Minieapolis and then on to Cleavland and with a return back to Nevada via
Texas and finally on to New York.
The newer equipment has far greater reliability than in the past, but do
your users and yourself a favor, defrag. Incidenatlly how many hot swap
spares are configured into your Raid5? I have 2 per loop for my mission
criticle application.
Regards,
Jerry
"Alan Vidmar" <Alan.Vidmar@Colorado.edu> wrote in message
news:402A5CD4.2776.1775EDB@localhost...
> As a side note to this discussion: how do people feel about defragging
RAID 5
> arrays?
>
> Not needed? Unnecessary? What do you think?
> Alan
>
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 22:17:35 EDT