Windows Server 2012R2 screenshot
-
Freshly formatted volume, somehow deduplication is saving me 734GB on a data volume where nothing is stored.
-
I was gonna say WinSxS but... if that's 734 GB you got bigger problems.
-
You are a master of understatement.
-
The bigger mystery, dedup was not even turned on for that volume...
-
Is this perhaps on a VM? You can (sometimes) get odd stuff like that on VMs.
-
Nope. Bare metal install. It happens somewhat frequently on 2012R2. It always resolves itself without intervention. This is the first time I thought to take a screen shot and post it.
-
Addendum: I believe this same Windows Server bug is what sometimes causes a web front end that we wrote for an in-house product to display some strange values. We have a backup service that we sell, that will also take full image backups for disaster recovery. The web front end will occasionally show negative values for backups, or backups that are larger than the available storage on the server itself. It is usually a negative for those though.
Yes, I know, we could fix it. It has never been an issue that did not resolve itself on the next scrape though so it is in the Backlog with very low priority.
-
It's possibly displaying the max savings possible?
I'm not a server admin I have no idea WTF Deduplication is.
-
Deduplication is like compression, sort of.
It should not show any savings until there are files on the volume, because there are no savings to be had. Dedup goes over the entire volume and finds portions of it that are duplicated and replaces them with a library reference to a single copy of said data. That is oversimplifying it though. It gives you a rough idea of it.
Kind of like zipping an entire volume. Sort of.
We use Server 2012R2 because it is able to dedup within and across VHDs. In some cases, we have seen over 80% savings on file storage. We try not to push it too much, but it is fairly easy to fit 60-70TB of VHDs on a 20TB volume with room to grow.
-
When I see "server 2012", my first thought is SQL Server. I was confused for a few posts there
-
Sorry, the "R2" should have tipped you off. I suppose for clarity I should have said "Windows Server 2012R2" though.
-
I'm not exactly compus mentis at the moment. Just got a bit confused
-
@Intercourse said:
Dedup goes over the entire volume and finds portions of it that are duplicated and replaces them with a library reference to a single copy of said data.
Dude, you've got a lot of zeroes on that disk. But maybe not as many as you thought.
-
Maybe we could just replace the whole lot by a count of the number of zeroes and ones there? Think of the compression ratio!
-
Or we could just sum all the bits while dropping remainders. We could reduce an entire disk to either 0 or 1. One single bit, no matter the initial size. That is nearly infinite compression.
-
I was just going to get that by applying the compression algorithm to its own output…
-
The magic of dedup:
-
So, what, you curate your porn collection by copying to various folders instead of some kind of tagging system? What's on there that's so duplicated?
-
In this case it is one of the folders that holds a client's backups. So for the image of the OS, nearly all of those files are parts of other backups, so they do not count towards the size on disk, plus dedup only holds one copy of the parts of files that are replicated all over the place (MS Office files are bad about this), etc.
I am not entirely sure why this particular one dedup'd so massively, but it did. For your average office, you might be surprised just how much dedup can help. For our backup service, we usually see somewhere around 80-90% dedup rates. This one is exceptional in that it is more like 99%.
So, what, you curate your porn collection by copying to various folders instead of some kind of tagging system?
Also, yes.
Seriously though, video does not dedup very well. 20-30% maximum. Unless I guess you sorted your video collection like that instead of creating hard links or sym-links. ;)
-
@Intercourse said:
In this case it is one of the folders that holds a client's backups.
Ah. I guess that makes some sense.
-
The best part is, they get billed for 1TB and it only costs me 11GB of disk space. Sort of.