On 2009-03-08_12:58:14, Steven Demetrius wrote:
Paul E Condon wrote:
I'd like some confirmation, or refutation, of some reasoning:
I have a USB external hard drive. It came with vfat fs, but I want to
write an ext2/3 fs on it. All my internal HD are ext3, but should this
one be ext3, also? Doesn't ext3 essentially write everything twice,
first to the journal, and then to the actual target location? This is
OK with an internal bus interface from the CPU to the HD, but USB is
not so fast. So I think I should not use ext3 for this HD. Is this
correct?
TIA
Basically ext3 is ext2 with Journaling. Journaling basically safe-guards
against power failure and system crashes. It is well suited for system
partitions and partitions that are being used most of the time your
computer is on.
Journaling uses significantly more disk space and does not allow for
deleted file recovery. IT uses more resources that ext2. Journaling does
not write everything twice. It keeps track of the file system which makes
recovery fast and more reliable than file systems without Journaling.
I recommend the following:
ext3 - for system partitions and data partitions which are in use most of
the time (/, /home, /var, etc. if they are separate partitions or
drives).
ext2 - for backup, removable, partitions rarely used, etc.
If your USB external is for backup or file transfers then I recommend
using the ext2 file system on it. Logic being that if your USB external
data gets corrupted then you still have a copy of the data on another
partition.
FYI:
Some people confuse backup with archiving. They will make copies of their
data and store it away until they have data problems with the system. This
is archiving.
Backup is a never ending routine whether done once a week or one a month
and also include regular data integrity checks of the backups.
I turned 76 last Dec. I've followed digital electronic computing since
I was in high school in late '40s. That was way back when digital
computers would seldom run for more than a few hours or a day without
crashing. Back then, people regularly ran what they called 'check
points'. These were records of the current state of the unfinished
computation in a format that was suitable for restarting the computer
after the offending vacuum tube was found and replaced. Often, the
last check point record was unreadable, whatever the recording medium,
and they had to find the last *good* check point. Then, they would
'back up' to that last good check point and resume the calculation
from that poing. Over time, the jargon has changed. The word backup
now means what was once called 'writing a check point'. Back then, I
think there was little idea of archiving as it is thought of today,
namely a permanent historical record. Check points were written onto
flaky (literally) magnetic tape, or punched paper tape. Any idea of
permanence of such records seemed rediculous.
I show my age by calling my nightly backups check points. They are
written on a separate HD on a separate computer. Now, I am working on
a system for archiving my check points onto an external HD.
Your comments are helpful reinforcement of my inclinations, but see
Sven's earlier comment. The current default for mount of ext3fs seems
not to be so costly as you or I have supposed. There is a very costly
option, but it is not the default.
Thanks.