Hi,

First note that slices can grow to at least 4GB, or larger depending on your virtual disk size.

There's many reasons for using a split disk over a single monolithic file.

The virtual disk split file is really the best choice even while it may seem a bit unlogical at first.

With split files you:

• get a better response, faster results when doing any VMware disk operation such as snapshots, shrink, compact, extend and defragment.
• have less chance of getting into a "all data lost" scenario when something bad happens such as running out of disk space as the split disk file scenario is much easier to repair than a growing single disk.
• don't need to have the full virtual disk size on free disk when committing snapshots

There seems to be a notion of "but it will be less performant", but in practice that is just not true.

Another side note is that, using your method, you can _only_ make backups when the VM is shut down.

If you try to make backups with the VM running then due to how the guest OS works, the backups can be corrupt as the guest OS is still writing to file allocation tables and files during the time it takes you to make the backup.

--
Wil

| Author of Vimalin. The virtual machine Backup app for VMware Fusion, VMware Workstation and Player |
| More info at vimalin.com | Twitter @wilva

Hi Wil,

Thank you for your answer, can you please clarify this:

1) You mean that my backups may come out corrupted if I try to backup a live running VM (which is obvious, of course) or is it also true for a paused VM? (i.e. when I back it up while the hypervisor is shut down)

2) What happens if I have a paused split file VM consisting of ten 2GB files which I run, then browse 2-3 websites from it, and then pause it again: will the hypervisor write data to just a few of those ten 2GB files, or will it write to all of them? (again, all things being equal) I am asking this question because I learned that the backup software I use (Owncloud desktop client) cannot analyze the contents of files for changes and upload only those changes to the server. Instead, it calculates file hashes to find out if the files were changed since the last check, and if they were, it simply re-uploads them entirely anew. Right now I have to re-upload my monolithic 30GB file VM each time after I use it and I am wondering if splitting it into 2GB files would make things a bit faster. What do you think?

3) If I change my mind after some time and decide to return to a monolithic file VM from a split file VM, would that be possible?

Thank you!

Thank you very much, Wil, this answers my concerns completely, I will now close this thread, many thanks.

When you power off a VM all open files will be closed and all filesystem information will be written to the vmdk.
When you only suspend a VM some of this information will be stored in the vmem-file instead.
This means that a backup of a suspended vmdk always is a little bit corrupt - most of the times that will be harmless but it can go wrong as well.
So if you have the choice - backup your vmdks when the VM is powered off.

2. You can easily answer that question yourself - just check the timestamps after you power off a split vmdk. Only one or a few slices will have the latest timestamp.

3. as long as your vmdk is healthy you can convert between split and monolithic at any time you want

Thank you very much for your reply. Yes, I know that it is important to backup all files associated with a VM (incl. vmem's), not only the vmdk itself. Thank you for your offer, but I don't want to waste your time, really. I guess I just need to split my VM into multiple files and see how it goes, since it is easy enough to convert it back to a monolithic file, should something goes wrong. Thank you!