Against which full backup are differentials based?
Acronis True Image Home (ATIH) v11
Windows XP Pro SP-3
Say I have the following tasks define in ATIH:
Task 1: "C: - Full backup"
Runs: Tue-Sun @ 5AM
Type: Full backup of C: drive
Task 2: "C: - Differential backup"
Runs: Mon @ 5AM
Type: Differential backup of C: drive
Backup location for both tasks is E:\Backups\TrueImage (which has been recorded as a backup location with quotas); that is, for both these tasks, the path to store the backup files was defined as a backup location. My assumption was that the Tue-Sun differential backups would be based off the Mon full backup. That is, a differential would be based on the latest full backup, so I'd get something like:
Week 1 Mon: Full_A Tue: Diff_A1 Wed: Diff_A2 Thu: Diff_A3 Fri: Diff_A4 Sat: Diff_A5 Sun: Diff_A6Week 2 Mon: Full_B Tue: Diff_B1 Wed: Diff_B2 Thu: Diff_B3 Fri: Diff_B4 Sat: Diff_B5 Sun: Diff_B6
I had expected Diff_B1 to Diff_B6 would be based off Full_B, not on Full_A. What has me concerned is the wording when defining a task as an incremental or differential image backup. I don't want to define a differential backup that creates an initial full backup and creates another full backup every N days. I want the full backup to occur on a specific day of the week with a differential on all the other days of the week. What I'm wondering is the phrase when defining a differential backup that says:
A differential backup contains all changes against the initial backup.
Well, if I don't select to define an differential backup where the first one would be effectively a full backup and another full was created every N days, what just is the "initial backup" of which the description speaks? Would the differentials be based against only the full backups that were created initially and every N days? Or would the differentials be based on whatever was the latest or just-prior full backup? I had assumed the later case.
Pat L,
TI 11 (the version before TI 2009) did things differently than the newer versions. Using the same folder had a much better chance of success. Also, the Backup Locations feature let you specify rules for the location (it didn't always work, though).
---
VanguardLH,
If the second task is using the Full created by the first task, it would work for one pass. After that, I think the first task would create a new Full, but the second task would still be pointing to the first Full. In other words, I think you would end up with all the Differential backups being based on the first Full (which wouldn't be good).
Using Chain2Gen would give you a lot more control over when the backups get created and what types.
Crap. I was hoping that the differentials would be based on the latest full backup. I don't want to use the "cyclic" setup for differentials where:
1. First differential is actually full backup.
2. Each differential thereafter is an actual differential backup.
3. After the configured N differentials have been made then another full backup gets created. Go back to step 2.
The reason why I don't like this scheme is that the full backups won't be on a specific day of the week, like Monday early morning. The first differential that becomes a full backup occurs on the first scheduled run of this backup task. Well, you could schedule it to occur on the next Monday. Alas, if the job gets postponed because the host is busy or was down then Monday's backup job might not run, so the N+1 backup would be on Tuesday or whenever next the host was usable for backup jobs.
So when I define a backup job, one could be for a full backup and another could be for differentials. It appears the differentials would be based on the first differential that was effectively a full backup and not based on the full backups that are scheduled weekly. The problem with cyclic differential backups that are based off the first "initial" backup is that the chain becomes more fragile over time. Although there are only 2 backups maximum involved (full + 1 differential), the dates between the full and selected differential could be immense, like weeks, months, or even possibly years. Over time the differentials could exceed the size of the full backup. If the initial [full] backup were unusable (deleted, corrupted, lost, whatever) then ALL those differentials are lost. If the inital backup were months ago, you would lose all those months of differentials, including your recent ones. All your backups would become unusable because just one of them, the initial one, because unuable.
So I guess the best choice with ATIH11 is to use cyclic backups where a full gets generated every N days and forget about the full backups happening on a specific day of the week. At least that would let me walk back through the backups without relying on just one full backup for all the differentials.
About Chain2Gen, where do I find that? Is it an editable script used with the jobs that get saved when you define a backup task? Or is it an executable that operates separately or as a plug-in to ATIH11? And will it work with ATIH11?
Chain2Gen can be downloaded from this thread. You might also want to check out some of the links in Grover's thread (Section 5 deals with Chain2Gen). It has many options available and you can set it to create a new Full on a specific day of the week.