Data Centers

Thin provisioning overview for storage efficiency

The storage purchasing process can be complicated because of the number of available options. Learn why thin provisioning is likely the best way to get the most out of your storage investment.

Thin provisioning is a technology that is available on some storage processors but not on all products. Thin provisioning on a storage area network (SAN) or a network attached storage (NAS) system basically consumes the usable storage to the level utilized with the storage subscriber's operating system. The best way to draw an analogy is the pie chart representing a Windows Server system's drive (Figure A). Figure A

In this example, the nearly 70 GB drive is only using 17.2 GB of the drive; in a thin-provisioned SAN environment, this would consume 17.2 GB on the logical unit number (LUN). In most situations, there is a small amount of overhead and/or metadata that manages the thin provisioning, as well as a block of incremental growth. Speaking in terms of Gigabytes and Terabytes, that overhead will be negligible.

Thin provisioning is available on some storage products, and too many times administrators shop for storage simply in terms of dollars and Terabytes. Software features with the storage should be an important part of the decision, but this is difficult; I've adopted a saying that "You can always find lesser expensive storage." (For a geek's delight, check out this cookbook from BackBlaze to build a 67 TB system for less than $8,000, and this related TechRepublic post on petabyte storage.)

Thin provisioning on a SAN system exists in many forms, but fundamentally, it allows the disks to be better utilized. If we stack a number of thin provisioned workloads on a SAN, we are potentially increasing the amount of hot spots on a disk. In the Figure A example, free space makes up approximately 75% of that disk allocation. Free space is great for SAN performance, but it is a waste of the storage investment dollar. This contention is not necessarily a problem, but it is an indicator to purchase faster drives (SAS over SATA, for example) if you need to keep performance levels high. Figure B shows a good way to visualize this stacking of LUNs on an individual disk. Figure B

This representation of Dynamic Provisioning from Hitachi Data Systems is used for the company's USP and AMS series of storage products, and it puts the LUNs on top of all of the disks in use. Storage administrators can carve this up in many ways, and each thin provisioning solution will offer different capabilities.

Over-provisioning

The risk with thin provisioning technologies is over-provisioning. If the SAN has more available storage, in most situations, each LUN is simply given more storage; but if the SAN runs out of space, there can be a hard stop approaching. Individual operating systems may also perceive that they have space available, complicating their behavior. In some situations, thin provisioning only grows — it does not shrink if data is deleted on the volumes. If the LUN or workload is moved at the storage processor level and rewritten based on its current size and contents, the decrease in size can be realized.

Summary

Thin provisioning is probably the best way to get the most out of your storage investment. Everyone's requirements vary, but if you work your scenarios and modeling out enough, you may be able to make the case for SAS disk solutions in favor of slower SATA disks if the overall storage requirement is lessened.

How do you manage thin provisioning? Share your comments in the discussion.

About Rick Vanover

Rick Vanover is a software strategy specialist for Veeam Software, based in Columbus, Ohio. Rick has years of IT experience and focuses on virtualization, Windows-based server administration, and system hardware.

Editor's Picks

Free Newsletters, In your Inbox