Re: SuperDuper! & Time Machine
- From: Fred Moore <fmoore@xxxxxxxx>
- Date: Wed, 04 Jun 2008 14:20:53 -0400
In article <1ihyvkt.1lvas6p1m8679uN%mikePOST@xxxxxxxxxxxxxxxxxxxxx>,
mikePOST@xxxxxxxxxxxxxxxxxxxxx (Mike Rosenberg) wrote:
Fred Moore <fmoore@xxxxxxxx> wrote:
Well, (and please correct me if I'm wrong; I don't have direct
experience) I recently read that a fellow with a moderate-sized HD, say
250GB, took 16 hours to do a TM restore.
Without knowing any details whatsoever, it's really next to impossible
to evaluate this. I did a test run with a backup containing about 125
GB total of current files (i.e. that's what needed to be restored) on a
500 GB SATA drive in a USB 2.0 enclosure, restoring to a 400 GB drive in
a Quad 2.5 GHz G5, and it took around 3.5 hours.
Right, Mike, I realize my comments call for a certain amount of
speculation. I'm just trying to get a handle on the relative performance
of TM vs SD because there have been some comments that TM has a great
deal of overhead which slows down a restore.
SD! or Disk Utility for that
matter could do the same restore in under an hour.
Moving 250 GB of data? You're grossly underestimating.
Well, let's run a few approximate numbers (and I don't want to get in an
argument about how many MB there are in a GB; these are
back-of-the-envelope calculations; specs are from Apple and Wikipedia):
** From the Apple Store I checked the advertised speed of the SATA disks
in a Mac Pro, that's Apple's current top-of-the-line tower. Their claim
is
3GB/sec or 10,800GB/hr = 0.023hr = 1.4min for 250GB of data
This seems grossly optimistic given the next item. Perhaps it's a max
for a RAID.
** The advertised max transfer rate for an Hitachi Deskstar 1TB drive is
1GB/sec or 3,600GB/hr = 0.069hr = 4.2min
This is exactly the kind of drive which one might put in a Mac Pro.
** But let's say reality (various overheads) is 1/2 the above
1/2GB/sec or 1,800GB/hr = 0.14hr = 8.3min
And let's compare to various other specs,
** eSATA is
300MB/sec or 1,080GB/hr = 0.23hr = 14min
** Let's say real eSATA is half that
150MB/sec or 540GB/hr = 0.46hr = 28min
** FW800 which _does_ get almost its spec is
98MB/sec or 353GB/hr = 0.71hr = 43min
** USB 2 spec is
60MB/sec or 216GB/hr = 1.18hr = 69min
** USB 2 reality is more like
30MB/sec or 108GB/hr = 2.31hr = 139min
** USB 1 spec (just for the hell of it)
1.5MB/sec or 5.4GB/hr = 46.3hr = 2778min
So according to these numbers, FW800 and faster _should_ be able to do a
restore of 250GB of data in an hour or less. Anything more would seem to
be system or app overhead. Even real USB 2 should restore your 125GB of
data in just over an hour, but you say it took TM 3.5 hrs total. I'm not
disputing your numbers. I'm just trying to get a handle on the what,
where, and why of the seemingly large overhead with TM.
But you cannot retrieve an accidently
deleted file or a corrupted file from a clone if you discover after you
last cloned that the file was deleted/corrupted.
SD! could be programmed to do a 'Smart Update' (just changed files)
every hour if you wanted to do that.
That's not the same as an archival backup by any means. Older versions
of those changed files are _not_ retained,
Yes, you're absolutely right. You'd handle this with multiple complete
backup with SD. Easy retrieval of recent docs is certainly one of TM's
strengths; quick restore of an operable system to get a crashed machine
up and running again apparently isn't.
nor are files that have been
deleted.
Actually, you can set a pref in SD to _not_ remove deleted files from
the backup, if that's what you want.
So tell me what you think of my analysis.
and
In article <1ihzdj6.bfd9tg1lghdpkN%nomail@xxxxxxxxxxxxxx>,
nomail@xxxxxxxxxxxxxx (Johan W. Elzenga) wrote:
My guess is that he used an USB 1.1 disk.
Based on the numbers above USB 1 would have take almost 2 _days_. :/
I recently bought a new
MacBook Pro and moved all the data from the old PowerBook by using a TM
backup, and I was up and running in about an hour. OK, that was 80 GB or
so, not 250 GB, but that would have meant 3 hours for 250 GB.
Johan, your results consistent with Mike's numbers, which reinforces the
question about TM overhead.
--Fred
.
- Follow-Ups:
- Re: SuperDuper! & Time Machine
- From: David Empson
- Re: SuperDuper! & Time Machine
- From: Richard Maine
- Re: SuperDuper! & Time Machine
- References:
- SuperDuper! & Time Machine
- From: Jim Higgins
- Re: SuperDuper! & Time Machine
- From: aRKay
- Re: SuperDuper! & Time Machine
- From: Johan W. Elzenga
- Re: SuperDuper! & Time Machine
- From: Fred Moore
- Re: SuperDuper! & Time Machine
- From: Mike Rosenberg
- SuperDuper! & Time Machine
- Prev by Date: Re: Lost faith in Mozy... :-( -- CrashPlan better?
- Next by Date: Re: SuperDuper! & Time Machine
- Previous by thread: Re: SuperDuper! & Time Machine
- Next by thread: Re: SuperDuper! & Time Machine
- Index(es):
Relevant Pages
|
Loading