After provisioning the Disks for VM, we normally tend to 'Allocate' space and create Logical Drives inside the VM.
ReFS seem to be a good bet in terms of being the latest and having the feature of auto-correcting in case of any failures. NTFS has been there for quite some years and ReFS is touted as next best technology for file storage & access.
It so happened that, one of the ReFS drives in one of our VMs got corrupted somehow and we could not recover the contents. Ten days effort of deploying an application for a demo, went for a complete mess. In NTFS, there is a mechanism to run CHKDSK and possibly recover the bad sectors. But in ReFS, running it is of no impact, as the system internally tries to do it by itself.
Finally, we had to recreate everything and from this incident onward, I started choosing NTFS for all drives that I create.
There is a way we could have possibly recovered the data had we taken a backup of either the content or the drive itself to a Storage Account. But being (a) new to Azure, (b) not knowing internals of how technology works, (c) when you are the sole person doing everything, (d) demo set up in a short time and (e) especially when we do these in small / start-up work places, these kind of processes may not be practically executed.
So, if you choose ReFS, go ahead, but make sure you backup the contents as per the changes made. Else, choose NTFS. In either case, taking backup is a good practice. NTFS has ChkDsk way of possible recovery while ReFS does not seem to be proved (as in our case).
No comments:
Post a Comment