Although 4Kn disks solve the overhead problem admins encounter with 512n disks, not all OSes and hypervisors have...
been updated to the new format. Until that happens, organizations can make do with 512e disks.
Historically, hard disk drives have been designed to organize data by segregating spinning platters into concentric tracks and then breaking down each track into constituent sectors. Each sector traditionally holds a 512 byte data payload, and a sector is then the smallest possible unit of storage on the disk.
For example, if a file is 100 bytes, it will still be allocated an entire sector, though a typical multi-megabyte file -- such as an image file -- will consume many sectors. Later electronic storage devices, such as solid-state drives and nonvolatile memory express devices, don't use the same mechanical composition as hard disk drives, but still use the same sector-type formatting to organize data for compatibility.
All of these traditional disks have no distinct industry designation, but they have recently been dubbed 512n disks to indicate that they use natural or native 512 byte sectors.
The problem with 512n disks
The problem here is that modern, high-capacity hard drives are suffering from efficiency limits -- not capacity limits. The issue is that each sector doesn't just hold data, it also holds error correction, a sector address and other associated formatting overhead.
This overhead is multiplied by the number of sectors. As disks keep getting bigger, the amount of space used by this overhead increases, too -- the storage becomes less efficient. In addition, most files use multiple -- sometimes many -- sectors and can lead to increased file fragmentation and greater sector error rates.
4Kn disks attempt to solve the problem
Designers realized that it could be far more efficient to simply make the sectors larger. Rather than using the traditional 512 byte sectors, disks could use 4 KB -- 4 kilobytes or 4,096 bytes -- sectors instead. This lowers the total number of sectors for the same disk capacity, and because each sector still uses the same basic overhead, storage efficiency -- the proportion of storage to overhead -- improves. Disks that employ this International Disk Drive Equipment and Materials Association (IDEMA) Advanced Format are dubbed 4Kn -- or 4 KB native -- disks.
But there is another issue. OSes and hypervisors have long relied on this traditional 512 byte sector format, making it difficult -- even impossible -- for organizations to move to the newer 4Kn disks. Real migration to 4Kn disks won't be possible until OSes, hypervisors and legacy applications with direct storage access are all updated to support the IDEMA 4Kn Advanced Format. The good news is that major OSes such as Windows Server 2008 and later and Hyper-V already provide support, but it's always important to test and verify compatibility before deploying 4Kn disks in production.
The introduction of 512e disks
To bridge this gap between 512n and 4Kn formats, designers created an emulated disk product that enables disks with 4 KB sectors to emulate disks with traditional 512 byte sectors. These are dubbed 512e -- or 512 byte emulation -- disks.
These 512e disks are indeed 4 KB sector products, but the disks internally translate between the formats and present themselves to the host system as if they have traditional 512 byte sectors. The advantage is that legacy environments can use the newer disk hardware.
Even with 512e, the move to 4 KB sectors isn't guaranteed. For example, VMware didn't support direct-attached 512e disks until the release of vSphere and vSAN version 6.5, though this will require Virtual Machine File System 6.0 for proper 512e support. By comparison, Hyper-V in Windows Server 2016 will support 512e, and 4Kn disks with virtual hard disk (VHD) files use the advanced VHDX format. Older VHD files might need to be upgraded to VHDX format to use advanced disks properly.
A 512e disk might experience performance degradation upon writing because the 512 byte sector that is written to the 512e disk must first be translated to the disk's internal 4 KB sector format before the data can actually be committed to the media. That is, the 4 KB sector of the disk is first read into an internal buffer -- one 4 KB sector can hold up to four traditional 512 byte sectors -- the 512 new bytes are written to the corresponding portion of the 4 KB buffer, and then the whole 4 KB internal buffer must be rewritten to the underlying disk.
Microsoft documentation refers to this process as read-modify-write (RMW). In addition, writes from an OS are generally not aligned to the disk's natural 4 KB boundaries, so multiple RMW processes might be needed to execute a single write to a 512e disk.
Because of these potential inefficiencies, emulated disks are intended as a stopgap measure, and they should be replaced with 4Kn disks as soon as OSes and other software allow.
Dig Deeper on Server hardware and virtualization
Related Q&A from Stephen J. Bigelow
Just because software passes functional tests doesn't mean it works. Dig into stress, load, endurance and other performance tests, and their ... Continue Reading
Don't neglect form factor as part of your data center server selection. Instead, figure out what type of environment you need and learn which server ... Continue Reading
Learn how load balancing in the cloud differs from a traditional network traffic distribution, and explore the different services available from AWS,... Continue Reading