Just connect the HD to your MB whenever you like. TM will notice it was attached and backup to it. If you shutdown the MB while TM backup is active you will be warned and given option to allow it to complete or you can quit and shutdown. When MB restarted after interrupting a TM backup the next TM backup will try and resume where it left off and complete the backup it started prior to you shutting down your MB.
oh ok, thanks for the info....but if you connect the external HD whenever you want does it still know what to backup???
Doing a backup just involves sending new files up to .Mac (a background task, but yeah it requires a good broadband connection). Looking into the past, afaik, is done either
1) on a per-folder basis - retrieving individual folder information is not bandwidth intensive
or
2) a spotlight search - and Leopard allows you to send a search request to another Mac, meaning .Mac would do a search itself and merely send the results to your machine.
I think .Mac backups have HUGE potential I don't know if Apple will enable it, but it seems to be the same technology as what will be used for local server backups anyway (just slower). It'd be great if Mozy offered the same thing.
ps. Anyone know if Timemachine compresses the backups?
I'm currently using Superduper for my backup needs, doing the Copy Different option to an external drive, so apart from all the 'Point in time' niceness in Time Machine, and the hard linking stuff, the data I've got on my backup is sort of the same as what Time Machine would be keeping.
What I've noticed, and what I hope they've put in a mechanism for, is the problem of differential backups and podcasts. My external drive filled up, and I couldn't work out why until I found thousands and thousands of multi-meg files in iTunes under podcasts.
You never notice them on your own machine if you have the setting 'keep last 3' or similar, but the differential backup keeps everything, and eventually drowns in them.
Anyone else have a solution for this, or am I going to be impressed to find it's one of the standard Ignore folders in TIme Machine ?
Sure. But I assume it won't make 'temporary backups' while the external HD is offline, so of course you lose all hourly, daily or even weekly backups for the time you're away, right? Theoretically, this could be avoided (as hinted) by making all those backups to a temporary maybe hidden directory on your local HD, but I'm pretty sure TM won't do that.
Sure. But I assume it won't make 'temporary backups' while the external HD is offline, so of course you loose all hourly, daily or even weekly backups for the time you're away, right? Theoretically, this could be avoided (as hinted) by making all those backups to a temporary maybe hidden directory on your local HD, but I'm pretty sure TM won't do that.
Philotech
TM backup will execute only if the configured backup Volume is mounted.
This article is very well done and enjoyable to read.
I have Leopard and had fiddled with TM-- It was OK, but no big deal.
While reading the article, I experimented... and learned a lot!
TM, in fact, is a very big deal!
Without violating the NDA, I can say this much:
TM is typically Apple OS X-- does what you want/expect, transparently, with minimal overhead. It is flexible and easy to use-- just let it do its thing!
I have lots of questions on Time Machine.. Apple should make a FAQ or something because it's one of the features I think will cause more doubts!
-> Background: 99% of the time I use my Mac either at Work or at Home. At Home i have an Airdisk set up to automatically mount.
From what I understand, Time Machine will only work if the drive is mounted, otherwise TM won't perform the hourly backup. This means that if I chose to backup to Airdisk, when i'm at work no hourlies will be backed up.
a) Can I choose to both make a back up to my Aidisk AND an ENCRYPTED backup to a Windows samba share on a windows server at work?. Won't TM cofuse the two destinations?
I think this is weird since it might occur that a file is backed up to the server at work but not to my Airdisk and then If I try to recover it how could TM tell which one to look for? this confuses me a lot!
b) This is about the hourly/weekly backups.
For example:
Oct 12 7pm : newly created File A is backed up to TM
Oct 12 8pm : file A was previously (7.30pm) delted so it doesn't appear on this hourly backup.
Oct 13 12am: daily backup from Oct 12 is backed up to TM. It doesn't contain file A, but file A is backed up in hourly 7pm from oct 12.
Now... a few days later, it comes the time to delete hourly backups... how does TM manage to keep file A in the TM pool for me to recover at a later time?
Wild speculative guess: Your file A will lost because the hard link feature does not provide a way to easily take care of it.
To keep it, it would require a consolidation of all 24 hourly backups. Then you'd have for each file 24 potential conflicts because you could have 24 versions. You'd say never mind just keep the latest, ie the 12am version. Fine that's what TM does right now. Unfortunately, the latest version of your file A is the deleted version, ie no file at all... But really that's no difference to a file that was created the day before and then amended a few times but finally deleted on that day: all changes that did not persist at 12am are lost, you only have the previous day's backup IF the file existed then already.
So you'd have to adjust the rule: for the 12am daily backup take the 12am snapshot of the file system UNLESS there were files existing before that have been deleted between 12am the previous day and 12am today. That would require full file checking and everything an ordinary backup program would do, and all the easiness and beauty of TM and hard links was gone. Not gonna happen. You need an ordinary backup programm for that and fire it off before deleting files.
The easier way to explain the difference between soft and hard links, in my opinion, is like this:
A file consists of an actual file data, and directory entry, known as inode, referring to the data.
A symbolic link is a file, referring the inode of the original file, so to access the file data a system has to travel through this path:
symlink => inode => file data.
...
Thanks a lot, that was much easier indeed. As much as I like the original article, the explanation of how links and everything work including the pictures is pretty much incomprehendable.
For consistency's sake, the depiction above should read (IMHO):
A question for those who have tested it: how large should a typical Time Machine backup drive be? I understand that this will vary with our use etc., but let's assume that I have an iMac with a 320 GB disk, about 100 GB of it available. WHat size drive should I consider for backup?
Given that 500GB drives can be had now for under $100, get a 500GB & never think about it again (for a few years anyway - storage gets consumed, it's the nature of the thing)
My guess is no. You would probably need to reinstall the system and then retrieve your files from the backup..
I think the multiple meanings of "backup" are mixing us up: backing up files to a storage disk, from which they can be retrieved, is one sort of backup; maintaining a current bootable partition on a different drive from your main boot disk is another kind. neither replaces the other; they are complementary, and each is important.
As to whether one can restore a bootable partition from Time Machine, I suspect it does not, for the reason that it would involve "crossing the streams" Or, as an old Irishman once said, "when I drink whiskey, I drink whiskey, and when I drink water, I drink water."
I think the multiple meanings of "backup" are mixing us up: backing up files to a storage disk, from which they can be retrieved, is one sort of backup; maintaining a current bootable partition on a different drive from your main boot disk is another kind. neither replaces the other; they are complementary, and each is important.
As to whether one can restore a bootable partition from Time Machine, I suspect it does not, for the reason that it would involve "crossing the streams" Or, as an old Irishman once said, "when I drink whiskey, I drink whiskey, and when I drink water, I drink water."
TM will restore a complete system backup and it will be bootable afterwards. Pop in the DVD Installer disc, select the Volume you wish to restore to, select the complete TM backup you want to restore from, wait for it to complete the restore operation and then you will have a bootable system once again. Works just fine, easy to do and without any real fuss.
This is a good background article, but it misses the point of Time Machine, I think. Backing up files is old news, and I'm sure Time Machine does it as well as anyone. The real magic of Time Machine, though, is information retrieval. It's something that Shadow Copy and rsync can't hope to do.
In a traditional file backup solution, if you discover one missing record in your Address Book, you'd have to go get all your incremental backups, restore each file in turn, look for the record, and if you find it export it manually, then restore the current version and import it.
With Time Machine, you just search back through time using the same search field you do normally. It will go back in time and find the first instance of the search criteria and allow you to restore just that record to the present. THAT is revolutionary. The rest is just another implementation of an old concept.
The Shadow Copy service in Vista can do just that. In fact, Sharepoint and Exchange implements this feature extremely well. I can pull back any one piece of data (a calendar event, email, contact etc.) without restoring or exporting the entire record.
Even with Previous Docs you can open up an old folder and just re-instate one file out of that shadow copied folder. The API lets you go even further detail by allowing (or disallowing) certian items to be restored together AND NOT separately. For instance, if you have a bank record of a deposit into one account (item 1) from another account ( item 2) you can make it so it is impossible to reverse item one without reversing item 2 at the same time. That's a simplistic example but it shows the power of A.C.I.D. transactions in Vista/NTFS6.
Quote:
The data backup features related to Shadow Copy are only useful if a Windows machine is running in an environment with a server backing them up. Shadow Copy is not in itself a backup system, although it can present a listing of duplicated files that were captured by the shadow copy service. Without a dedicated backup system, Previous Versions only shows local shadows of a file. It does not copy files to an external disk for safekeeping, and its shadow copies can't be browsed through by the user in the file system by date or by query. Shadow Copy is certainly not an easy to use consumer backup solution (nor is intended to be), which is what Time Machine expressly is.
That's incorrect. Vista comes with a backup program (that uses Shadow Copy) that can copy incremental backups to external media and hard drives as well as network connected drives. Previous Docs also allows one to search and browse through backups using Windows Explorer by date, time, and numerous other criteria. In fact, if you open a folder from the previous docs window you will see that the address bar shows you the name, date, and time of the folder snapshot in naturall language and that you can do a natural language search over that data or use the breadcrumb bar to see other times and snapshots registered in the file system.
That's incorrect. Vista comes with a backup program (that uses Shadow Copy) that can copy incremental backups to external media and hard drives as well as network connected drives. Previous Docs also allows one to search and browse through backups using Windows Explorer by date, time, and numerous other criteria. In fact, if you open a folder from the previous docs window you will see that the address bar shows you the name, date, and time of the folder snapshot in naturall language and that you can do a natural language search over that data or use the breadcrumb bar to see other times and snapshots registered in the file system.
Be careful when pointing out inaccurate statements about Windows here, because you might get told something like "If you know so much about Windows, then why don't you just buy a PC?". If the feature does not work exactly like the Mac, then it doesn't exist.
This is a good background article, but it misses the point of Time Machine, I think. Backing up files is old news, and I'm sure Time Machine does it as well as anyone. The real magic of Time Machine, though, is information retrieval. It's something that Shadow Copy and rsync can't hope to do.
And that is where you need to be careful. Other than the amazing visual experience offered by Time Machine and the integration with individual applications, rsync is *very* much capable of achieving *exactly* the same backup strategy. Tiger already supports hard links as discussed in the article and rsync supports syncing content using hard links to avoid storing redundant copies of the same file. Sure, it's not going to be as user friendly, but it would actually work.
Quote:
In a traditional file backup solution, if you discover one missing record in your Address Book, you'd have to go get all your incremental backups, restore each file in turn, look for the record, and if you find it export it manually, then restore the current version and import it.
With Time Machine, you just search back through time using the same search field you do normally. It will go back in time and find the first instance of the search criteria and allow you to restore just that record to the present. THAT is revolutionary. The rest is just another implementation of an old concept.
With rsync and hard links there would be no "restore" as such, you'd just change to a folder on your backup drive that corresponds to a backup date. That folder would appear as a full backup.
I agree that rsync isn't as good an option as Time Machine appears to be from a usability point of view - but lets not forget that it is actually capable of achieving a very similar result in terms of file storage and retrieval.
Yeah, and that's like and antivirus program which the OS takes a hit from. If the article is correct and the browser cache is one of those places TimeMachine will ignore, then it shouldn't be too bad.
FSEvents has to deal with less information than Spotlight (its a subset) so the hit should be less than Spotlight, which you have now.
And that is where you need to be careful. Other than the amazing visual experience offered by Time Machine and the integration with individual applications, rsync is *very* much capable of achieving *exactly* the same backup strategy. Tiger already supports hard links as discussed in the article and rsync supports syncing content using hard links to avoid storing redundant copies of the same file. Sure, it's not going to be as user friendly, but it would actually work.
With rsync and hard links there would be no "restore" as such, you'd just change to a folder on your backup drive that corresponds to a backup date. That folder would appear as a full backup.
I agree that rsync isn't as good an option as Time Machine appears to be from a usability point of view - but lets not forget that it is actually capable of achieving a very similar result in terms of file storage and retrieval.
This is all correct, but somewhat beside the point. It may actually be that rsync is the engine that Time Machine uses (I don't know this and am just too lazy to try and find out right now). Apple does this often, take the existing BSD utility(s) and make it usable to the ordinary user (Activity Monitor, Disk Utility, etc.) I've used rsync and agree with you in principle but it takes either a lot of experience and/or time to get rsync to work this way reliably and with no user intervention.
Comments
Just connect the HD to your MB whenever you like. TM will notice it was attached and backup to it. If you shutdown the MB while TM backup is active you will be warned and given option to allow it to complete or you can quit and shutdown. When MB restarted after interrupting a TM backup the next TM backup will try and resume where it left off and complete the backup it started prior to you shutting down your MB.
oh ok, thanks for the info....but if you connect the external HD whenever you want does it still know what to backup???
Why too slow?
Doing a backup just involves sending new files up to .Mac (a background task, but yeah it requires a good broadband connection). Looking into the past, afaik, is done either
1) on a per-folder basis - retrieving individual folder information is not bandwidth intensive
or
2) a spotlight search - and Leopard allows you to send a search request to another Mac, meaning .Mac would do a search itself and merely send the results to your machine.
I think .Mac backups have HUGE potential
ps. Anyone know if Timemachine compresses the backups?
No compression done.
oh ok, thanks for the info....but if you connect the external HD whenever you want does it still know what to backup???
TM knows what to backup.
What I've noticed, and what I hope they've put in a mechanism for, is the problem of differential backups and podcasts. My external drive filled up, and I couldn't work out why until I found thousands and thousands of multi-meg files in iTunes under podcasts.
You never notice them on your own machine if you have the setting 'keep last 3' or similar, but the differential backup keeps everything, and eventually drowns in them.
Anyone else have a solution for this, or am I going to be impressed to find it's one of the standard Ignore folders in TIme Machine ?
Cam
TM knows what to backup.
Sure. But I assume it won't make 'temporary backups' while the external HD is offline, so of course you lose all hourly, daily or even weekly backups for the time you're away, right? Theoretically, this could be avoided (as hinted) by making all those backups to a temporary maybe hidden directory on your local HD, but I'm pretty sure TM won't do that.
Philotech
Mel, You are correct about reading the entire thing. When I saw "users," I did not equate it with different machines. I appreciate your post. Thanks
Didn't mean to be bitchy, but this happens so often.
The other thing is someone asking a question right after ten people have answered it for someone else.
Sure. But I assume it won't make 'temporary backups' while the external HD is offline, so of course you loose all hourly, daily or even weekly backups for the time you're away, right? Theoretically, this could be avoided (as hinted) by making all those backups to a temporary maybe hidden directory on your local HD, but I'm pretty sure TM won't do that.
Philotech
TM backup will execute only if the configured backup Volume is mounted.
I have Leopard and had fiddled with TM-- It was OK, but no big deal.
While reading the article, I experimented... and learned a lot!
TM, in fact, is a very big deal!
Without violating the NDA, I can say this much:
TM is typically Apple OS X-- does what you want/expect, transparently, with minimal overhead. It is flexible and easy to use-- just let it do its thing!
-> Background: 99% of the time I use my Mac either at Work or at Home. At Home i have an Airdisk set up to automatically mount.
From what I understand, Time Machine will only work if the drive is mounted, otherwise TM won't perform the hourly backup. This means that if I chose to backup to Airdisk, when i'm at work no hourlies will be backed up.
a) Can I choose to both make a back up to my Aidisk AND an ENCRYPTED backup to a Windows samba share on a windows server at work?. Won't TM cofuse the two destinations?
I think this is weird since it might occur that a file is backed up to the server at work but not to my Airdisk and then If I try to recover it how could TM tell which one to look for? this confuses me a lot!
b) This is about the hourly/weekly backups.
For example:
Oct 12 7pm : newly created File A is backed up to TM
Oct 12 8pm : file A was previously (7.30pm) delted so it doesn't appear on this hourly backup.
Oct 13 12am: daily backup from Oct 12 is backed up to TM. It doesn't contain file A, but file A is backed up in hourly 7pm from oct 12.
Now... a few days later, it comes the time to delete hourly backups... how does TM manage to keep file A in the TM pool for me to recover at a later time?
Wild speculative guess: Your file A will lost because the hard link feature does not provide a way to easily take care of it.
To keep it, it would require a consolidation of all 24 hourly backups. Then you'd have for each file 24 potential conflicts because you could have 24 versions. You'd say never mind just keep the latest, ie the 12am version. Fine that's what TM does right now. Unfortunately, the latest version of your file A is the deleted version, ie no file at all... But really that's no difference to a file that was created the day before and then amended a few times but finally deleted on that day: all changes that did not persist at 12am are lost, you only have the previous day's backup IF the file existed then already.
So you'd have to adjust the rule: for the 12am daily backup take the 12am snapshot of the file system UNLESS there were files existing before that have been deleted between 12am the previous day and 12am today. That would require full file checking and everything an ordinary backup program would do, and all the easiness and beauty of TM and hard links was gone. Not gonna happen. You need an ordinary backup programm for that and fire it off before deleting files.
The easier way to explain the difference between soft and hard links, in my opinion, is like this:
A file consists of an actual file data, and directory entry, known as inode, referring to the data.
A symbolic link is a file, referring the inode of the original file, so to access the file data a system has to travel through this path:
symlink => inode => file data.
...
Thanks a lot, that was much easier indeed. As much as I like the original article, the explanation of how links and everything work including the pictures is pretty much incomprehendable.
For consistency's sake, the depiction above should read (IMHO):
inode => symlink => inode => file data
A question for those who have tested it: how large should a typical Time Machine backup drive be? I understand that this will vary with our use etc., but let's assume that I have an iMac with a 320 GB disk, about 100 GB of it available. WHat size drive should I consider for backup?
Given that 500GB drives can be had now for under $100, get a 500GB & never think about it again (for a few years anyway - storage gets consumed, it's the nature of the thing)
My guess is no. You would probably need to reinstall the system and then retrieve your files from the backup..
I think the multiple meanings of "backup" are mixing us up: backing up files to a storage disk, from which they can be retrieved, is one sort of backup; maintaining a current bootable partition on a different drive from your main boot disk is another kind. neither replaces the other; they are complementary, and each is important.
As to whether one can restore a bootable partition from Time Machine, I suspect it does not, for the reason that it would involve "crossing the streams" Or, as an old Irishman once said, "when I drink whiskey, I drink whiskey, and when I drink water, I drink water."
I think the multiple meanings of "backup" are mixing us up: backing up files to a storage disk, from which they can be retrieved, is one sort of backup; maintaining a current bootable partition on a different drive from your main boot disk is another kind. neither replaces the other; they are complementary, and each is important.
As to whether one can restore a bootable partition from Time Machine, I suspect it does not, for the reason that it would involve "crossing the streams" Or, as an old Irishman once said, "when I drink whiskey, I drink whiskey, and when I drink water, I drink water."
TM will restore a complete system backup and it will be bootable afterwards. Pop in the DVD Installer disc, select the Volume you wish to restore to, select the complete TM backup you want to restore from, wait for it to complete the restore operation and then you will have a bootable system once again. Works just fine, easy to do and without any real fuss.
This is a good background article, but it misses the point of Time Machine, I think. Backing up files is old news, and I'm sure Time Machine does it as well as anyone. The real magic of Time Machine, though, is information retrieval. It's something that Shadow Copy and rsync can't hope to do.
In a traditional file backup solution, if you discover one missing record in your Address Book, you'd have to go get all your incremental backups, restore each file in turn, look for the record, and if you find it export it manually, then restore the current version and import it.
With Time Machine, you just search back through time using the same search field you do normally. It will go back in time and find the first instance of the search criteria and allow you to restore just that record to the present. THAT is revolutionary. The rest is just another implementation of an old concept.
The Shadow Copy service in Vista can do just that. In fact, Sharepoint and Exchange implements this feature extremely well. I can pull back any one piece of data (a calendar event, email, contact etc.) without restoring or exporting the entire record.
Even with Previous Docs you can open up an old folder and just re-instate one file out of that shadow copied folder. The API lets you go even further detail by allowing (or disallowing) certian items to be restored together AND NOT separately. For instance, if you have a bank record of a deposit into one account (item 1) from another account ( item 2) you can make it so it is impossible to reverse item one without reversing item 2 at the same time. That's a simplistic example but it shows the power of A.C.I.D. transactions in Vista/NTFS6.
The data backup features related to Shadow Copy are only useful if a Windows machine is running in an environment with a server backing them up. Shadow Copy is not in itself a backup system, although it can present a listing of duplicated files that were captured by the shadow copy service. Without a dedicated backup system, Previous Versions only shows local shadows of a file. It does not copy files to an external disk for safekeeping, and its shadow copies can't be browsed through by the user in the file system by date or by query. Shadow Copy is certainly not an easy to use consumer backup solution (nor is intended to be), which is what Time Machine expressly is.
That's incorrect. Vista comes with a backup program (that uses Shadow Copy) that can copy incremental backups to external media and hard drives as well as network connected drives. Previous Docs also allows one to search and browse through backups using Windows Explorer by date, time, and numerous other criteria. In fact, if you open a folder from the previous docs window you will see that the address bar shows you the name, date, and time of the folder snapshot in naturall language and that you can do a natural language search over that data or use the breadcrumb bar to see other times and snapshots registered in the file system.
That's incorrect. Vista comes with a backup program (that uses Shadow Copy) that can copy incremental backups to external media and hard drives as well as network connected drives. Previous Docs also allows one to search and browse through backups using Windows Explorer by date, time, and numerous other criteria. In fact, if you open a folder from the previous docs window you will see that the address bar shows you the name, date, and time of the folder snapshot in naturall language and that you can do a natural language search over that data or use the breadcrumb bar to see other times and snapshots registered in the file system.
Be careful when pointing out inaccurate statements about Windows here, because you might get told something like "If you know so much about Windows, then why don't you just buy a PC?". If the feature does not work exactly like the Mac, then it doesn't exist.
This is a good background article, but it misses the point of Time Machine, I think. Backing up files is old news, and I'm sure Time Machine does it as well as anyone. The real magic of Time Machine, though, is information retrieval. It's something that Shadow Copy and rsync can't hope to do.
And that is where you need to be careful. Other than the amazing visual experience offered by Time Machine and the integration with individual applications, rsync is *very* much capable of achieving *exactly* the same backup strategy. Tiger already supports hard links as discussed in the article and rsync supports syncing content using hard links to avoid storing redundant copies of the same file. Sure, it's not going to be as user friendly, but it would actually work.
In a traditional file backup solution, if you discover one missing record in your Address Book, you'd have to go get all your incremental backups, restore each file in turn, look for the record, and if you find it export it manually, then restore the current version and import it.
With Time Machine, you just search back through time using the same search field you do normally. It will go back in time and find the first instance of the search criteria and allow you to restore just that record to the present. THAT is revolutionary. The rest is just another implementation of an old concept.
With rsync and hard links there would be no "restore" as such, you'd just change to a folder on your backup drive that corresponds to a backup date. That folder would appear as a full backup.
I agree that rsync isn't as good an option as Time Machine appears to be from a usability point of view - but lets not forget that it is actually capable of achieving a very similar result in terms of file storage and retrieval.
Yeah, and that's like and antivirus program which the OS takes a hit from. If the article is correct and the browser cache is one of those places TimeMachine will ignore, then it shouldn't be too bad.
FSEvents has to deal with less information than Spotlight (its a subset) so the hit should be less than Spotlight, which you have now.
And that is where you need to be careful. Other than the amazing visual experience offered by Time Machine and the integration with individual applications, rsync is *very* much capable of achieving *exactly* the same backup strategy. Tiger already supports hard links as discussed in the article and rsync supports syncing content using hard links to avoid storing redundant copies of the same file. Sure, it's not going to be as user friendly, but it would actually work.
With rsync and hard links there would be no "restore" as such, you'd just change to a folder on your backup drive that corresponds to a backup date. That folder would appear as a full backup.
I agree that rsync isn't as good an option as Time Machine appears to be from a usability point of view - but lets not forget that it is actually capable of achieving a very similar result in terms of file storage and retrieval.
This is all correct, but somewhat beside the point. It may actually be that rsync is the engine that Time Machine uses (I don't know this and am just too lazy to try and find out right now). Apple does this often, take the existing BSD utility(s) and make it usable to the ordinary user (Activity Monitor, Disk Utility, etc.) I've used rsync and agree with you in principle but it takes either a lot of experience and/or time to get rsync to work this way reliably and with no user intervention.