wrote:

Tim S wrote:
Vortex wrote:

Well the time has come.....as it does every few years.

My HDD is virtually full. I have insufficient memory. My PC is too
noisy.
My son tells me my graphics card is inadequate. No more room for
enhancements and upgrades. It's time for a new one.

I've just ordered one of these http://tinyurl.com/kjp8p with (amongst
other things) a 500G HDD which I envisage assembling this weekend.

The most important decision I have to make in the coming days is
whether to format the 500G as one huge partition or to have multiple
partitions eg: 100G for OS and applications and 400G for My
Documents/Pictures/Videos etc.

What would you do?


David

Hi

As odd as it may seem, there is a "right" answer for this. There are many
variations of "right", but basically:

a) OS onto one partition (eg windows C:\ , Linux / /var /usr etc)

b) User data onto a different partition

a may be further subdivided (OS/applications/swap) and b may be
subdivided (personal user data aka home directory versus "shared" data
like films, mp3s, photos which the whole family access).


Do not ever install the OS onto a whole-disk-sized partition and then
lose your data in amongst the crap. If you do and have to re-install the
OS, you will weep.

Due to the fairly sucky way that DOS partitioning works, I would also
make all partitions "primary partitions" upto the limit of 4, unless you
know that the installer is using the newer MS partitioning scheme.

Something else to consider - leave a second partition spare so you can
install a second OS. It won;t amount to much out of 1/2 TB.

Cheers

Tim

I would say about the oposite to that. I dont split discs and dont lose
document files when installing or reinstalling an OS. Using a sensible
backup routine gives backui[p protection. So far I've lost a total of
one object as a result of storing it in an odd place and forgetting to
include it in any backup, then a hdd failed. But partitioning wouldnt
have helped that in any way.

I understand your reasoning. But it has some problems (which you may decide
to live with, which is fair enough):

a) It leaves you vulnerable to a greater risk of filesystem corruption
(which is more often than not constrained to one instance of a filesystem,
unless you are an FS/device driver developer ;- ); If you monitor your
disks, failed disks sometimes (more often than not in my experience) go
through a failure period where some cluster of blocks become inaccessible,
before the entire drive crashes. Splitting filesystems reduces the impact
and gives greater chance of recovery.

b) OS reinstalls *are* more difficult. Your solution is conditional upon
good backups - but backups are easier to set up if you arrange your
filesystems sensibly.

c) It makes dual/multi boot harder to implement.

d) Multiple filesystems provide rudimentary quotaring, so one errant user
can't fill up the system filesystem, which is usually bad. Then again if
your whole household logs in as "Administrator" this point is moot.

I've used all sorts of partitioning schemes, but they always revolve around
the separation of user data from program/OS files. No professional sysadmin
would do it any other way - it was good practise in the 70's (initially for
different reasons) and it still is good practise.

Remeber the days of 500M and 1G drives? Remeber why partitioning was
such abad idea then? 500G seems like a vast amount of storage now, but
it will ni time become a 'small disc', and however you partition it now
its guaranteed to be split alll wrong.once space starts getting low.

Put in a couple of 40-60GB partitions for OS + spare - that leaves nearly
400GB for "user data", which is going to take a while to fill, even with a
digital camera. By then, expect a second 500GB disk to cost 70 quid (as the
Seagate Barracuda SATA 300GB does right now).

If necessary, use second disk as a decant area to re-layout partitions on
first disk, but with 40-60GB for the OS/programs, you are likely to be good
for a fair few years unless you install Matlab + Xilinx with every library
possible along with everything else going.

Another solution is to use logical volume management, but it depends on the
OS support for it and sometimes the simplicity of simple partitions buys
extra robustness.

I remember when a 70M hdd was gargantuan, and none of us could figure
out how we could ever fill it.


Yes. I remember when 20MB was the dogs Doesn't change anything though.

The one remaining question: how the heck do you backup a half terabyte
disc? Only 100 dvds I spose, or less after compression.

The only practical answer is another disk. Unless one is rich enough to buy
a SAIT tape drive and some media which will do about 800GB per tape
typically. Several grand for the drive and 100 quid per tape. Ouch.

DVDs suffer from the same problems as CDRs - as in how long are they good
for? 1 year, 5 years, 10 years? I don't trust them for the only copy of
valuable data.

What I do in my home server is double up on the disks - I have 4. Each disk
carries various data (home directories, /vol/ areas, upto about 40% of the
disk capacity), the rest of the disk is used as a backup area for data from
other disks. Backups using rsync run several times a day automatically.
Won't help is the box catches fire, but it guards against disk failure,
which I've had twice in 4 years on that server. For the box catching fire
scenario, I back up my really important data to a spare disk in my work PC,
using rsync it doesn't take long nor use much bandwidth. But an old mouldy
PC in the shed/garage with a big disk and a bit of cat5 (even WIFI) would
work well too.

Cheers

Tim