Should I split a big (8TB) backup disk into more than one partition, or not?

David Fletcher dave at thefletchers.net
Mon Sep 21 09:15:31 UTC 2020


On Sun, 2020-09-20 at 08:45 +1000, Karl Auer wrote:
> On Sat, 2020-09-19 at 07:07 -0400, Charles M wrote:
> > 
> > I suggest everyone read up on "mean time between failures" and
> > then,
> even more importantly, read up on "bit error rates".

I've been reading this thread and would just like to say that some
years ago (just looked at the date of the CAD files at the time which
is 30 Dec 2007) I was working on an aluminium case design for a small
home server which has always run versions of Ubuntu Server. I was
discussing this at my local LUG meetings and decided to run it with
just a single HDD.

Myself, I've never used RAID. Everybody must understand that if you
have an array of n units of a particular HDD with a known MTBF (FTs)
then the MTBF of the array (FTa) is going to be FTa = FTs/n. Plus
you've probably got additional hardware to interface to all the drives
which will further reduce the MTBF.

For applications where guaranteed uptime is essential such as retail
sites then yes, you chuck in as many drives as you need to guarantee
uptime and enable hot replacement of any failed unit.

In my application I thought, I'm paying for the electricity to run the
server 24/7 so power consumption needs to be as low as possible!
Therefore it's a single HDD and fingers crossed.

The server started out running Hardy on an ATX mini ITX board. Can't
remember what make of HDD I had then, but it was trouble free. Sometime
in 2010 I upgraded to a 32 bit 12V industrial type board running Lucid
on a 1TB Samsung drive. Next it was Trusty on a 2TB Toshiba drive. In
May I upgraded to a 64 bit industrial board with Focal on another 2TB
Toshiba.

None of the drives have caused me any problems. Note that the system
first has the incoming mains and broadband cable protected by a Belkin
Gold connector block followed by a UPS so it has the best protection I
can provide against surges and power failures.

Why choose this brand of hard drive? A glance at the failure
statistics https://www.zdnet.com/article/which-company-makes-the-most-r
eliable-hard-drives-which-drives-might-be-worth-avoiding/ makes this
obvious.

My conclusion is that for a surge and power protected home system,
minimal hardware gives you maximum MTBF along with minimal expense and
electricity consumption. If interested you can get my case design and a
couple of other things here https://flightmaker.hopto.org/ I used
LibreCAD for the case design.

Dave




More information about the ubuntu-users mailing list