When SuperDuper! does an incremental backup (using Smart Copy), does it
have to look at some "Yes, I've been changed since the last backup, or
no, I haven't" indicator on each and every one of the 270,000 files on
my HD?
Or can it just start moving down the nested folder hierarchy from the
top, looking at some marker associated with each _folder_ which tells whether anything inside that folder has been changed or not? -- so that
if that marker says nothing has been changed inside the folder, it
doesn't need to look at anything any deeper inside that folder?
Reason for asking: Incremental backups using the SuperDuper! "Smart
Copy" feature and similar-sized LaCie PocketDrives seem to go much
faster on my iBook G4 than on my wife's PowerBook G4, despite my usually having modified many more files between weekly backups. I'm wondering
if it's because my HD has a deeply nested multi-layer arrangement of
nested folders, and she keeps all her files in just a few chaotic
top-level folders.
Or -- I suppose more likely -- maybe it's that I'm backing up to a
FireWire PocketDrive and she's still using a USB-only PocketDrive.
More likely is the speeds of the 4 different disks.
Your iBook disk and your Firewire disk.
Your Wife's Powerbook disk and her USB disk.
A lot of time is going to be spend seeking on each disk looking
for modification dates. The faster the disk, the sooner each seek
will complete. So if one or more of your disks has a faster
rotational speed (5400 RPMs or 7200 RPMs vs 4200 RPMs), then you
will tend to finish sooner. Track to track head movement seek
times can also be involved, but that info is not generally printed
on the packaging, where as RPMs generally are.
When SuperDuper! does an incremental backup (using Smart Copy), does it
have to look at some "Yes, I've been changed since the last backup, or
no, I haven't" indicator on each and every one of the 270,000 files on
my HD?
Or can it just start moving down the nested folder hierarchy from the
top, looking at some marker associated with each _folder_ which tells
whether anything inside that folder has been changed or not? -- so that
if that marker says nothing has been changed inside the folder, it
doesn't need to look at anything any deeper inside that folder?
Thanks very much for instructive reply. As reward, another query: Any
idea why Retrospect's incremental backup process seems to take very substasntially longer in the preparation phase than SuperDuper's Smart
Copy, when backing up the same iBook computer, with the same change
history, and two similar LaCie FireWire pocket drives as targets?
,Doc O'Leary <droleary.usenet@2q2006.subsume.com> wrote:
In article <siegman-27BADF.10294316042006@news.stanford.edu>,
AES <siegman@stanford.edu> wrote:
Thanks very much for instructive reply. As reward, another query: Any idea why Retrospect's incremental backup process seems to take very substasntially longer in the preparation phase than SuperDuper's Smart Copy, when backing up the same iBook computer, with the same change history, and two similar LaCie FireWire pocket drives as targets?
I'd wager it's because Retrospect has crufty old code that was the
result of a limited Carbon port where SuperDuper seems built for OS X. Consider, for example, just these two methods of finding files modified
in the last week:
mac64:~ droleary$ time find ~ -mtime -7
real 0m13.309s
user 0m0.239s
sys 0m4.806s
mac64:~ droleary$ time mdfind -onlyin ~ 'kMDItemFSContentChangeDate > $time.today(-7)'
real 0m7.480s
user 0m0.169s
sys 0m0.387s
So using Spotlight for just that simple query I can almost halve the
time, and who knows what other attributes/factors come into play. There
are all sorts of inefficiencies that have probably compounded over the
years with Retrospect.
Sysop: | Gate Keeper |
---|---|
Location: | Shelby, NC |
Users: | 790 |
Nodes: | 20 (0 / 20) |
Uptime: | 41:10:38 |
Calls: | 12,115 |
Calls today: | 5 |
Files: | 5,294 |
D/L today: |
72 files (9,959K bytes) |
Messages: | 564,934 |