One question about Time Machine
I just thought about it and realised that Time Machine seems to require me to tether my notebook to a hard drive, at least occasionally. Anyone think this would be much cooler if combined with Amazon S3? (http://akatsuki.co.uk/article/apples...-opportunities)
Comments
That said, if you use JungleDisk or a similar tool that lets you mount Amazon S3 as a local volume through WebDAV, you could probably use that as the target for Time Machine to backup to, and that'd be just what you wanted.
I just don't personally believe it's useful yet. You'd be looking at large bandwidth bills from Amazon, and hours and hours of backing up.
As has been commented on elsewhere, copying the whole file to backup when a small part is changed is dumb. Don’t do it.
Binary diffs are extremely difficult to implement transparently. Very few file systems can do that at this point. It can be done, and maybe we'll see it by 10.6, but nobody could realistically expect it.
It sure is nice to see the files assorted by date, but you know what would be better? How about something similar to Microsoft’s version control on documents? Give me a preview window with red strike-throughs and green additions for changes between this document and the one before it. Show me the changes.
That can be done already; it's up for applications to do this through Time Machine's API.
Rsync is your friend.
No, rsync is not as sophisticated. It has binary diffs, but they're not transparent, so you don't end up with complete files, but only incremental versions. That's very user-unfriendly. It can check for changes and not copy what hasn't been changed, but it has to do so manually, which is a lot slower than Time Machine's automatism.
Permanently tethering my laptop to another HD to take advantage of this feature is dumb.
No, it's not. It means that you can safely put your external HD somewhere else, so if your laptop or desktop blows up, the house is set on fire, or the thing is stolen, you have a backup.
Sign a deal with Amazon S3 or Google’s GDrive and add it to the .mac package.
Google's Gdrive is for employees. It's not an external service, and won't be. Maybe they'll launch a different service with the same name, but what they have right now is something completely internal.
Have Time Machine back up to off-site, redundant storage handled by professionals. Have it do it automatically and invisibly. Store changes when there is no network connection to upload later. Wow, a HUGE and compelling reason to sign up for .mac.
Yes, and hugely expensive and slow. Sorry, but that's about as realistic as asking people to send their backups in, burned on DVDs, via snail mail.
ZFS is also your friend if you will let it be.
How so? What does ZFS do in terms of backups? Its redundant storage doesn't help for most backup situations.
If I “Secure Delete” a file, will Time Machine also scrub every previous copy? Obviously very important.
Obviously, it is.
Can I use Time Machine on two PowerBooks with ONE external drive? I have one external drive and I hope I can use it for both my fiancée's PowerBook, and my PowerBook.
I have another question.
Can I use Time Machine on two PowerBooks with ONE external drive? I have one external drive and I hope I can use it for both my fiancée's PowerBook, and my PowerBook.
I have not tried that, but let me tell you why I think it might be possible:
Time Machine stores its backups in a folder on the destination volume called "Backups.backupdb". Inside that folder is another one with the name of a machine, e.g. "Lundy's Power Mac G5". Inside that folder is the following:
Attachment 131
So you can see that there COULD be two folders for two different machines. Whether it actually supports that, I can't test.
You would have to set each machine to backup manually, so it would not go looking for the missing drive.
Time Machine backs up changed files every hour or so, and consolidates the changes every day. If it hits a thrashing situation, where the file keeps changing as it tries to back it up, it will time out and try again later.
ZFS has versioning built in. It also has the ability to add storage seemlessly, so your drive just gets larger.
I am not convinced that bandwidth would be a huge issue as far as backing up goes except initially which would be painful. My computer is connected to the internet all the time, and the computer could trickle data from a cache upstream although this would have inherent inefficiencies in getting the most recent data backed up.
I guess I could always set up a local wireless NAS for home use and just deal with the fact that backup is not going to be done when I travel or go to work (again recent backups will not be present).
If Google does launch Platypus/GDrive, then I definitely think this is the way to go.
It also bothers me how inelegant backing up the whole file every time there is a change is. When editing and writing I tend to save a lot! Like every sentence or two. I will probably be generating hundreds of saves in a session. Is Time Machine really going to save every single copy until it runs out of space? Or if it does it every hour, am I going to risk losing an hour's worth of work which can cost quite a bit in law and other fields (although I admit this is better than the current setup). If I retag MP3s, am I going to end up with lots of duplicates when I don't really care to have versions, just the most recent?
Sorry, but having to tether my laptop takes the mindlessness out of automated backup... If I have to do that, then I can manually run SuperDuper myself. I certainly don't need Time Machine at all then.
And what's the alternative? To back up to a different partition on the internal hard drive? How useless is that?
ZFS has versioning built in. It also has the ability to add storage seemlessly, so your drive just gets larger.
I don't see how that helps this particular backup application.
I have not tried that, but let me tell you why I think it might be possible:
Time Machine stores its backups in a folder on the destination volume called "Backups.backupdb". Inside that folder is another one with the name of a machine, e.g. "Lundy's Power Mac G5". Inside that folder is the following:
Attachment 131
So you can see that there COULD be two folders for two different machines. Whether it actually supports that, I can't test.
You would have to set each machine to backup manually, so it would not go looking for the missing drive.
Time Machine backs up changed files every hour or so, and consolidates the changes every day. If it hits a thrashing situation, where the file keeps changing as it tries to back it up, it will time out and try again later.
Ooh, do you have access to Leopard Developer Builds? Can you give us a high-res screenie of the Time Machine vortex thingy in action, I'd love to see it up close but the the screenshot on Apple's website is quite low-res. That must be one stunning piece o' eye candy.
With one pricier model w/ wireless -- that would automatically connect to your laptop at home and back it up via Time Machine?
I'd actually wager good money on the former; the latter is something of a pipe dream.
And what's the alternative? To back up to a different partition on the internal hard drive? How useless is that?
The alternative would be at least caching backups onto .Mac until that drive is plugged in... And once you have an online mechanism built-in, then it is not too big of a step to online backups.
I don't see how that helps this particular backup application.
You should probably read a bit about ZFS then... Check-summing, snapshots and incremental versioning. Seems to address a lot of the issues you would want in a backup system. Basically a lot of the heavy lifting for incremental versioned backups is already done (to address your earlier comment regarding binary diffs).
I have no idea where you live, but after the initial backup, I don't think the backup would be too burdensome on bandwidth on an individual basis (the aggregate might be a lot of traffic though).
Certainly Apple might be saving this for 10.7 as an update to the Time Machine concept. As for me, I think I might just pay for S3 if I can get Time Machine to work with JungleDisk or another mounted solution.