Now, I concede that in the (unlikely) event that corruption occurs among the child backups but not in the full, then you would be probably have access to more recent backup data that is uncorrupted, if you were doing regular differential backups. Thus, the probability of corruption preventing your ability to restore to the state that was most recently backed up (e.g., yesterday's backup) should be about the same for a differential that was created yesterday and a chain of incrementals that was most recently added to yesterday. Second, in situations where one cares primarily about being able to restore the most recent daily backup (and less so about the ability to go back in time to a specific date), the amount of storage space on the destination drive is going to be approximately the same for the most recent differential as it is for the chain of incrementals up to and including the most recent one.
#Macrium reflect free incremental full#
Therefore, if bit rot occurs, it is much more likely to occur in the full backup than in one of the child incrementals and if it is the full that is corrupted, then the arguments about differential vs. Furthermore, unless the inter-backup (e.g., daily) changes to your source data represent a significant fraction of your backup selection (e.g., source partition), then the full backup file will be much larger than the full chain of incrementals. The arguments about corruption in long chains of incrementals, while in some cases (but not always) technically correct, are in my opinion misleading.įirst, the probability of corruption in storage (bit rot) can reasonably be assumed to be relatively uniform across the entire backup set. Ask yourself though, "How many backups do you make compared to how many restores you do." In my case the answer is many, many, many more backups per restore so I'd be doing incrementals.
There shouldn't be any problem switching to differentials. The old "different strokes for different folks" situation. For various reasons I only do full images of my OS and apps with data files being done using a different program and I only do the OS and apps somewhat infrequently because I'm not concerned about data that exists nowhere else. Then you get into whether you are doing the consolidation methods which I never use myself. If I were doing it, I'd probably do new fulls on a week or 2 week schedule. I don't think there is a hard rule on what is a too long chain but IMO 30 is in the ballpark. Data corruption can throw a wrench (spanner for you UK folks) into either scheme but an often considered weakness in a long chain of incrementals is that any incremental from the bad one onward are useless. OTOH, the incremental is seen to be best for backup because it only has to backup the data changes from the last incremental and is a faster not having to go back to the state of the full. In simple terms, the differential often is considered the best option for restoring because you only have to deal with the full and the differential.
0 kommentar(er)
