Need advice on testing v7 in parallel with v6

[cparker4486]

Hi,

I've installed V7 on a new Server 2012 VM and have been testing it by backing up a few minor VMs that weren't already included in a V6 job. I'd like to do a full run of a big job from V7 this weekend.

The V6 job uses CBT and is in reversed incremental mode. I'm thinking I should disable changed block tracking on this new V7 job to minimize the effect on the existing V6 job. Is that correct?

[tsightler]

CBT will not be affected by the V7 job, they will both work and use CBT just fine, although you don't want to run both jobs at once.

[Starman]

I had to do this with v6.5 as I created a server 2012 build for Veeam 7.0. Both were backing up the vCenter fine as I moved jobs from 6.5 to 7.0

[cparker4486]

CBT will not be affected by the V7 job, they will both work and use CBT just fine, although you don't want to run both jobs at once.

Then I think I have a misunderstanding of exactly how CBT works. I thought it behaved as a sort of "archive bit" in that once Veeam (or anything that uses CBT) has made a backup the CBT record is reset to begin tracking changes to be used at the next backup. Please set me straight.

Ultimately, what I'm concerned about is that after my V7 test job runs the next V6.5 job* will either be invalid because the CBT was reset by another job or it will be treated as an active full because of confusion over CBT status.


* I am actually using V6.5.

[chrisdearden]

Think of CBT like a journal of blocks. As I understand it each process can create its own pointers in the journal.

[tsightler]

I think of it like a version control system for blocks, although it doesn't have the content, just the knowledge of which blocks changed. Every time VMware takes a snapshot it tags that "version" of the disk, the next time a snapshot is taken creates a new version any blocks changed since the previous version are tagged with the new "version" info. This gives me the ability to ask for a "diff" between any two versions, (i.e. any two snapshots) so the chain looks something like this:

1. 1st V6.5 Backup -> Snapshot V1 - Full Backup
2. 1st V7.0 Backup -> Snapshot V2 - Full Backup
3. 2nd V6.5 Backup -> Snapshot V3 - Ask for list of changed blocks since snapshot V1
4. 2nd V7.0 Backup -> Snapshot V4 - Ask for list of changed blocks since snapshot V2

And it continues on, otherwise even 6.5 would only be able to have one job per VM so you couldn't have say a replication job running every hour and a backup job running every day, because they'd step on each other. This isn't intended to be a technical explanation of exactly how CBT works, just a conceptual way to think about it.

[cparker4486]

Thanks guys!