The support forum

Straight one-time migration mode

ScottD-KMHA :

Apr 18, 2017

Is there a mode I can put Bvckup2 in so that it very simply just copies data from a source to a destination, without having to take the time to perform a full scan of the source/destination and then building the ID's and creating the snapshots, etc?

We have three uses for this product.  First, we use it for syncing data from server A to server B.  For that, we like the scanning and ID's and snapshots, etc.  For the second use, we utilize this for one-time migrations.  Most times, we still want to have at least the snapshot in place so that we can have "resume" capability.  The third use, is just a fast copy of a large data set from one drive to another.

In the this 3rd situation...it would be nice to skip all of the scanning, ID generating, snapshot creating steps and just instead start straight copying data from the get-go.

I realize we would lose the ability to start back where we left off, if there was a crash and we had to start Bvckup2 again...but sometimes we just want to move a chunk of data fast and take the remote risk of failure.

And here is I guess an honest question, if such a mode is possible (usage 3)...do you think copying say 1TB of data with Bvckup2 would be any faster than copying 1TB of data via plain old Windows copy?

I guess I'll ask, in summary, this way...we have two additional modes we would like to use this for (besides the normal sync mode):  migration and fast copy.

1) Migration would allow us to copy a large chunk of data (10TB for instance), but have resume capability if something crashes and we have to start Bvckup2 again.

2) Fast copy would allow us to copy a large chunk of data (500GB+) from one drive to another, without the need for resume capabilities (unless that is also possible without taking the additional time to scan, ID, and/or snapshot the source/destination).

What would you suggest we modify in Bvckup2 to support these two uses, if possible?  I know of some items that can be changed in the UI (like turning off snapshots/delta copying), but I have a feeling more would be needed to achieve what we are looking for.  Both of these uses only represent a small % of our actual use of this product, but we would like to replace our existing tools for things like this (Karens Rep, Beyond Compare, Windows Copy, Robocopy, etc).  

Or do you think we should just stick with plain old Windows Copy (Robocopy, xcopy, etc) for our migration/fast copy needs and only use Bvckup2 for our real-time syncing scenarios (which is mostly what we currently use it for now anyway)?

Alex Pankratov :

Apr 19, 2017

Is there a mode I can put Bvckup2 in so that it very simply just copies data from a source to a destination, without having to take the time to perform a full scan of the source/destination and then building the ID's and creating the snapshots, etc?


Scanning of the source is needed regardless of how the copying is made (not just with bvckup2, but with any tool). The difference with Bvckup 2 is that it's done upfront and other tools like robocopy do it as they go, so it ends up intermixed with the actual copying.

Scanning of the destination can be suppressed by using destination snapshots, but still requires a single full scan. *However*, in the data migrating scenario the destination is likely to be empty, so there's not much to be optimized here.

File ID retrieval *can* be suppressed. IDs are used for more/rename detection, so setting Backup Settings > More Options > Move detection to OFF will do the trick. This is absolutely worth doing for one-off migration jobs as ID retrieval is an expensive operation.

Snapshot creating - saving destination snapshot is very cheap, literally milliseconds.

If you were referring to the *VSS* snapshot of the source volume, this can be suppressed by setting Backup Settings > More Options > Shadow Copying to Disable. However the default is "As needed" meaning that the snapshot won't be created unless the app runs into a locked file. If you don't have locked files during a migration, so there's no overhead, but if you do, you will want the shadow copying, so, again, not much can be optimized here.

---

I guess I'll ask, in summary, this way...we have two additional modes we would like to use this for (besides the normal sync mode):  migration and fast copy.


For both cases you should get pretty close to what you are after using:

    1.  Manually run job
    2.  Use destination snapshot
    3.  Copy files in full
    4.  Disable move/rename detection

Running this sort of job will scan the source. On the first run it will scan the destination as well, but presumably it will be empty. It will then go straight to copying files to destination.

If the job aborts, the destination snapshot will capture the state of backup, so on the next run the app will only need to scan the source. As I explained the source is scanned regardless of how things are copied, but doing it upfront allows coalescing listing operation and doing them in parallel, so it's generally (much) faster.

The only edge case if the machine dies or the app itself gets killed/crashes. This will prevent a snapshot from being created (as it's saved after a run completes) and it will cause the app to do a destination scan on the next job run.

---

Now, all that said, I can see a point in having a simpler way to set this up. Let me see what we can do.

ScottD-KMHA :

Apr 19, 2017

Excellent.  Thank you.

New topic

Create
Made by Pipemetrics in Switzerland
Support


Follow
Twitter
Dev blog
Miscellanea Press resources
Testimonials
On robocopy
Company
Imprint

Legal Terms
Privacy