What’s a Dynamic Disk?

Something you may have seen, but probably don’t have

It seems like a disk would be a really simple thing. You put some data on it, add a little organization around it to find that data and you’re done. Right?

Dynamic disks are a little more complex, but the good news here is that most folks really don’t need to worry about dynamic disks. But they are kind of interesting, and I’ll go into some detail on the different things they can do.

Dynamic disks

Dynamic disks are different from basic disks in their ability to handle features like merging, spanning, and mirroring volumes across multiple physical drives. Most home users have basic disks and likely don’t need dynamic disk support unless they require advanced functionality.

What makes a dynamic disk different?

Basic disks are the disks you’re familiar with already. A basic disk pretty much matches the relatively simple idea that I described. The basic disk typically contains one or more partitions, each partition contains one or more volumes, and those volumes are formatted using one of several different types of file systems like FAT32 or NTFS.

While there are several layers to this scheme, the approach is actually pretty straightforward.

Dynamic disks differ in several different ways. For example, rather than having a partition table to determine where all those partitions physically are on a disk, they actually have a partition database. That database implies that you might be able to do a few more things with dynamic disks than you would with a basic disk.

Merging space

Consider this really common scenario. You have a basic disk with two partitions and some unused free space. It’s partition A next to partition B, followed by that free space.

With a basic disk, if you want to use that free space to increase the size of partition A, you must first rearrange things so that the free space is actually next to partition A. That typically means moving partition B so that you end up with, instead, partition A next to that free space, followed by partition B:

You could then merge the free space into partition A next to it so that partition A takes up the total. You’re left with a larger partition A followed by partition B, and that’s it.

There are several partition-management tools out there that can do exactly what I’ve just described, and in later versions of Windows, the disk management tool itself can usually do some of this kind of thing.

“Merging” space

Dynamic disks are slightly more flexible in terms of partitioning.

Rather than move anything, you can create a partition, of sorts, out of the free space and then simply say that partition A is made up of the area in front of partition B, and the area behind it. Your disk then looks like the first part of partition A, followed by partition B, followed by the second part of partition A:

To your computer it looks like two partitions, pretty much as normal. But on the disk the partition is actually broken up and lives in two different places.

Now, the technical term for those parts of a partition is “extent”. And as we’ve seen, extents don’t have to live next to each other to be used to create a single volume. So, that partition A that we just talked about is made up of two extents: one before and one after partition B.

Spanned volumes

Extents don’t even have to live on the same physical drive. The simplest case is taking two hard drives, setting them up as dynamic disks and then creating a single partition that encompasses both disks. Your system now sees one partition. The fact that it’s physically implemented as two separate drives is hidden. This is what they call a spanned volume since it spans multiple, physical disks:

The downside? You’ve doubled your risk of failure, as if either disk fails, the entire partition could be lost.

Striped volumes

Striped volumes are like spanned volumes except that instead of putting, say, the first part on one drive and the second part on another, the data is actually intermixed. Chunks of data alternate between one drive and the other. While this sounds confusing, there’s a very good reason for it. It turns out that when you distribute your data evenly across the drives like this, it’s faster to read and to write the data to and from the combination of drives:

(“sections” is actually arbitrary and undefined – exactly how data is divided up is an unspecified implementation detail.)

Once again, though, you’ve doubled your risk of failure, as if either disk fails, the entire partition could be lost. This is often referred to as “RAID-0”.

Mirrored volumes

Mirrored volumes are conceptually the opposite of striped. Two drives are used, but the amount of data represented is equivalent to one drive. Why? Well, when you write data to the drive, it’s written to both so that if one fails, its mirror can continue on without data loss. Again, two drives, but to the system it looks like a single partition:

This is often referred to as “RAID-1”. The downside is that it might be slightly slower, and of course it takes twice as many disks to get the storage you need.

RAID-5

RAID-5 is a little bit of everything so far. A RAID-5 setup can be implemented using dynamic disks. Three drives are used: two for data and one for error checking. As it turns out, when configured properly, any one of those three drives can fail without data loss:

RAID implementations can get significantly more complex, but this touches the surface of what you can do with dynamic disks in Windows. In practice, most RAID implementations don’t rely on Windows at all, but rather dedicated hardware RAID controllers.