Hi - I think of an incremental as "everything created after <date><time>". I will keep fulls and differentials for system restore, but incrementals catch files that might have been deleted in-between two full backups.
What i don't understand from the manual is what is meant by IFW "referencing" a prior backup before creating an incremental (and a warning not to have "too long a chain"). Can someone explain this to me - is it doing anything more complicated than noting the date of the "referenced" backup? (and if so, why?)
Thanks!
/j
Question about Incremental Backups
-
Bob Coleman
- Posts: 788
- Joined: Fri Aug 12, 2011 5:58 pm
Re: Question about Incremental Backups
jeffw_00 wrote:
> Hi - I think of an incremental as "everything created after
> <date><time>".
At least in the context of programs being discussed here an incremental is "everything changed after the last backup".
IFW references a prior backup because, by definition, the incremental backup being created will be everything that has changed since the referenced backup.
The distinction between differential and incremental backups can be confusing. The only difference really is whether or not the "referenced' backup is a full backup or a previous incremental backup.
The intended use of differential backups is create a full backup, create a differential backup based of the full backup, create another differential backup based on the full backup, create another differential backup based on the full backup. Repeat as many times as desired before creating a new full backup. Each differential backup depends only on the full backup upon which it is based.
The intended use of incremental backups is create a full backup, create a first incremental backup based on the full backup, create a second incremental backup based on the first incremental backup, create a third incremental backup based on the second backup. Continue this incrementing as many times as desired before creating a new full backup. Each incremental backup depends on the entire chain of incremental backups between it and the original full backup.
The first differential backup and the first incremental backup are actually the same thing. The only thing that absolutely characterizes a backup as incremental is that it is based on a previous backup that is not full.
> Hi - I think of an incremental as "everything created after
> <date><time>".
At least in the context of programs being discussed here an incremental is "everything changed after the last backup".
IFW references a prior backup because, by definition, the incremental backup being created will be everything that has changed since the referenced backup.
The distinction between differential and incremental backups can be confusing. The only difference really is whether or not the "referenced' backup is a full backup or a previous incremental backup.
The intended use of differential backups is create a full backup, create a differential backup based of the full backup, create another differential backup based on the full backup, create another differential backup based on the full backup. Repeat as many times as desired before creating a new full backup. Each differential backup depends only on the full backup upon which it is based.
The intended use of incremental backups is create a full backup, create a first incremental backup based on the full backup, create a second incremental backup based on the first incremental backup, create a third incremental backup based on the second backup. Continue this incrementing as many times as desired before creating a new full backup. Each incremental backup depends on the entire chain of incremental backups between it and the original full backup.
The first differential backup and the first incremental backup are actually the same thing. The only thing that absolutely characterizes a backup as incremental is that it is based on a previous backup that is not full.
Re: Question about Incremental Backups
Hi Bob - thanks for writing. I actually mis-spoke. I also think of an incremental as everything since the last backup, as dictated by the Modified flag on the file. It sounds like if I do a full on Sunday, and then 6 incrementals before the next full, that when I do the 6th incremental, IFW has to index all 5 prior incrementals, as well as the full, to determine what to backup. Is that really the case? (their comment in their docs about "chain length" seems to imply it). I guess that's more accurate than the Modified flag, but also more cumbersome.
Thanks!
/j
Thanks!
/j
Re: Question about Incremental Backups
On Sun, 25 May 2014 13:54:49 PDT, jeffw_00 wrote:
>Hi Bob - thanks for writing. I actually mis-spoke. I also think of an incremental as everything since the last backup, as dictated by the Modified flag on the file. It sounds like if I do a full on Sunday, and then 6 incrementals before the next full, that when I do the 6th incremental, IFW has to index all 5 prior incrementals, as well as the full, to determine what to backup. Is that really the case? (their comment in their docs about "chain length" seems to imply it). I guess that's more accurate than the Modified flag, but also more cumbersome.
>
Although there is some awareness of files in IFW, in general the
changes-only imaging is done by looking at blocks which have changed
on the disk, rather than changed files.
So if only part of a file has changed, only the block(s) containing
the change are imaged, rather than the whole file.
In your scenario above, for the 6th incremental run IFW will need to
check the prior incrementals + the full, unless you have turned on the
'Speed up changes-only' flag. In this case, special hash files are
written for each image that indicate the state of the drive or
partition at that time, so IFW would only need to compare the latest
hash file with the current state of the HD to determine the changed
blocks, and would not need to reference the previous images.
>Hi Bob - thanks for writing. I actually mis-spoke. I also think of an incremental as everything since the last backup, as dictated by the Modified flag on the file. It sounds like if I do a full on Sunday, and then 6 incrementals before the next full, that when I do the 6th incremental, IFW has to index all 5 prior incrementals, as well as the full, to determine what to backup. Is that really the case? (their comment in their docs about "chain length" seems to imply it). I guess that's more accurate than the Modified flag, but also more cumbersome.
>
Although there is some awareness of files in IFW, in general the
changes-only imaging is done by looking at blocks which have changed
on the disk, rather than changed files.
So if only part of a file has changed, only the block(s) containing
the change are imaged, rather than the whole file.
In your scenario above, for the 6th incremental run IFW will need to
check the prior incrementals + the full, unless you have turned on the
'Speed up changes-only' flag. In this case, special hash files are
written for each image that indicate the state of the drive or
partition at that time, so IFW would only need to compare the latest
hash file with the current state of the HD to determine the changed
blocks, and would not need to reference the previous images.
Re: Question about Incremental Backups
good point about blocks over files. Now i understand why restores can be more cumbersome (and risky).
Thanks
/j
Thanks
/j
-
Bob Coleman
- Posts: 788
- Joined: Fri Aug 12, 2011 5:58 pm
Re: Question about Incremental Backups
jeffw_00 wrote:
> good point about blocks over files. Now i understand why restores can be
> more cumbersome (and risky).
>
> Thanks
> /j
I'm not sure what you mean by "more cumbersome". More cumbersome than what? If "cumbersome" refers to what has to occur internally in the program, maybe.
I just want to clarify that there is nothing cumbersome from the external/UI point of view. Just select the most recent incremental as the thing to be restored and IFW, IFL, or IFD handles everything.
If by "cumbersome", you mean the need to insure that all previous items in the chain are available, that's correct.
> good point about blocks over files. Now i understand why restores can be
> more cumbersome (and risky).
>
> Thanks
> /j
I'm not sure what you mean by "more cumbersome". More cumbersome than what? If "cumbersome" refers to what has to occur internally in the program, maybe.
I just want to clarify that there is nothing cumbersome from the external/UI point of view. Just select the most recent incremental as the thing to be restored and IFW, IFL, or IFD handles everything.
If by "cumbersome", you mean the need to insure that all previous items in the chain are available, that's correct.
Re: Question about Incremental Backups
On Mon, 26 May 2014 13:22:59 PDT, just as I was about to take a herb,
Bob Coleman disturbed my reverie and wrote:
>If by "cumbersome", you mean the need to insure that all previous items in the chain are available, that's correct.
I try to avoid mixing incrementals and differentials in a given backup
chain. as space is not an issue, I use differentials exclusively.
--
Cheers,
DrT
** Stress - the condition brought about by having to
** resist the temptation to beat the living daylights
** out of someone who richly deserves it.
Bob Coleman disturbed my reverie and wrote:
>If by "cumbersome", you mean the need to insure that all previous items in the chain are available, that's correct.
I try to avoid mixing incrementals and differentials in a given backup
chain. as space is not an issue, I use differentials exclusively.
--
Cheers,
DrT
** Stress - the condition brought about by having to
** resist the temptation to beat the living daylights
** out of someone who richly deserves it.