# of differentials & full vs. differential backup

Re: Best way to remove old backup files?

Hello,

I'm new to TI and have 2 questions:

1. As I understand it you can have only ONE differential backup at any time. If you create another differential the old one is discarded. But in this thread I've read about many differential backups at any time. How come?

2. Somewhere else in this forum I've read that if you backup your system drive or partition it is almost better to use a FULL backup every time since the system drive or partition changes very often and so the full backup size is almost similar to the full and incremental or differential backups. Do you agree with that?

Thanks in advance for your help.

Albert

 

Re: Best way to remove old backup files?

Hello Albert

Welcome to TI and this Forum.

As the subject matter of your post deals with a completely different issue to the original content of this post may I suggest that you start a new thread.

The questions that you have asked are good questions which anyone new to TI may need help with. So a new thread with an appropriate Title may be helpful not only to yourself but other new users.

TiA

 

Re: Best way to remove old backup files?

Agreed; I have pulled these two posts out of the thread they were previously in and made the new thread.

 

Re: Best way to remove old backup files?

1. You can keep as many differential backup images as you want. Thus you could restore to a variety of dates rather that just the latest image. With incrementals all of them plus the base image need to be present.
At the present time if you delete any of the differentials between the base image and the latest you will not be able to run the additional verification. This quirk is supposed to be fixed in an upcomming build.
2. As I have enough room on my backup drive for the number of whole images that I need I keep it simple and do not use incrementals or differentials.

 

I have never been able to see any benefit from either dif or inc backups and always do a full back up of either the operating system and program partition or my data partition. Doing full backups keeps things simple. Perhasp there are benefits to dif or inc backups - but I have never really been able to figure out why it might be better to not do a full back up every time.

 

The only advantage is space and to a certain extent time. When you create an incremental and include validation, the whole archive is validated not just the last incremental. Due to a bug, you can only validate differentials if all the intervening differentials are present so this takes a lot of time. Also, if you have a bad incremental then everything from the last good incremental onward is lost.

I do the same as you, just do fulls. IMO, less chance of something going wrong due to the number of files or inadvertent deletion of an intervening file.

If I were to do incrementals, I would only go as far as a full and then 6 incrementals (1 week if you image each night); then start over with a new full...

 

Wouldn't it then be good to have the possibility to tell the system that the rest of the backup is already validated and only the new incremental has to be validated?

 

Agreed completely, this has been put forward frequently to Acronis in these columns and also on the the Wish List. The way the Secure Zone is totally re-validated each time one tries to validate an addition to the zone is ludricous in the extreme. After all a validated image in the zone is supposed to be secure and yet it gets validated again and again.
From my point of view, rather than give up in disgust, I re-thought the way I used TI and now have a much better way of securing my backups which does not use validations as provided at all.

 

What way is it?

 

If I knew how to give you a link to several posts I have made Ad Nauseam on the subject I would.
The basis is to image and then restore right away to a swappable hard drive. This gives a perfect replica of your hard drive and has advantages over doing a clone.

 

A reasonable alternative is to just not validate at all. Since starting with Acronis 6 I can recall 2 builds which gave significant problems. I don't really see the need to validate provided full system images are made frequently with known good builds. should a restore fail - very unlikely - the previous imaqe can be called upon. should that image fail.... and so on.

 

IMO, once you have established that TI is working properly with your hardware, be it the motherboard, memory, internal/external drives, or memory then it is not very likely to fail unless you develop a HW problem.

It seems that good hardware gone bad is most likely to be a memory problem. If the archive was written using good memory and the memory goes bad then the archive is good but it will not validate. If the archive was written with bad memory then you could have a problem because good memory will not arrive at the same validation checksum; a serious problem. I think writing an archive to a bad disk area is a much lesser problem.

Naturally, restoring from time to time will verify that things are as they should be but if restorations are not done then validations should be done at some interval. Letting it amass unvalidated, untested archives for a long period of time could lead to a real problem particularly if you don't keep a good supply of old archives on hand.

On the other side of the coin, a successful archive validation also provides confidence that your related hardware is in good shape. Sort of a fringe benefit to using TI .

 

I once went for just over a week without needing to restore
and even though I accept your general point I still think that what TI does is a kind of magic - which seems to either work completely or fail completely.

Just installed an intel core 2 6300 and a couple of 160 gig sata drives. Making an image of C: ( xp and a few programs) to E: took 1 minute 52 seconds ..... sure beats a full reinstall.

 

I also do full images only, since they are so easy to manage.

When my full images grow and I still want to keep several points in time avilable to restore to, I'll most certainly opt for incrementals and not for the multi-differential strategy. Validation bug aside, the multi-differential strategy is a waste of space in my opinion, since (assuming that files on HD are in permanent net growth, as they usually are), every differential, except the first one, will also contain changes already recorded in the previous differential files. In a string of 5 differential images the changes recorded in the first differential will be re-recorded 4 more times, the changes recorded in the second differential will be re-recorded in the following 3 differentials and so on. A buildup of redundant data.

For the same number of points in time to keep, the incremental strategy is much more space-conscious. I'm afraid I see no point in doing multi-differentials instead.

I'm aware, though, that I may be missing some particular application, when multi-differentials do indeed make sense.