Tips for Risk-Free Thin Provisioning
Experts and users mull the risks and rewards of this technology
July 11, 2007
![Network Computing logo Network Computing logo](https://eu-images.contentstack.com/v3/assets/bltde8121fc52c5c8f3/blt3f3d0318f746b1c2/65a530e4187606040a1d8b8c/placeholder.png?width=1280&auto=webp&quality=95&format=jpg&disable=upscale)
Thin provisioning: The term denotes a range of techniques for faking out a storage system in order to avoid overinvesting in disk space. And while vendor argument over the fine points has generated a media hypefest, users seem captivated by the concept of thin provisioning and eager to see if it really works. (See Revisiting Thin Provisioning's 'Firsts' and Real Thin Provisioning.)
Apparently, it does. Companies like Compellent, Datacore, Pillar, and 3PAR credit thin provisioning for increased sales. Among large system suppliers, Hitachi Data Systems has created thin provisioning for its Universal Storage Platform V that's been OEM'd by HP (in its EVA XP24000) and resold by Sun. (See Hitachi Bulks Up and Pillar Pushes Provisioning, Capacity.) NetApp claims to have its own take on the technique for file stores, in addition to a partnership with 3PAR. (See 3PAR, NetApp Join Ranks.) EMC has had thin provisioning for file and iSCSI on EMC Celerra systems since early 2006. And iSCSI vendors are getting in on the act. (See EqualLogic to Add Thin Provisioning and EqualLogic: Thin Is In.)
But thin provisioning comes with caveats. Even David Scott, CEO of 3PAR, has warned users about the importance of monitoring capacity thresholds in any implementation. (See Thin Is Definitely In.) Early adopters say thin provisioning can lead to problems if not done right.
The question is, what constitutes doing thin provisioning right? There are customers who claim to be succeeding with it, including Michigan-based law firm Dickinson-Wright PLLC (an EqualLogic user), Priceline.com (a 3PAR user), European satellite systems operator SES ASTRA (Datacore), the U.K.'s Sunlight Service Group (3PAR), and United Airlines (HDS).
Successful deployments have a few things in common. To find out what these are, we explored the issue with a range of suppliers, experts, and customers, and came up with the following list of thin provisioning best practices:
Don't believe the hype. "Just because a solution claims to be transparent, doesn't mean it is," says Greg Schulz of the StorageIO consultancy. "The term can mean virtually anything. Ask the vendors what they really do. Make them demonstrate it. See how it scales. If a product doesn't address your particular environment and how you use storage, it's not for you."
Be aware of architectural differences, but don't prejudge. Vendors differ in their approaches to thin provisioning. 3PAR, for instance, claims to carve physical disk arrays into "chunklets" of storage. Users request a specifically sized volume at a given quality of service, and the system carves out the capacity automatically. The hitch? You need 3PAR's arrays to make it work.In contrast, HDS (and by extension HP) create pools of storage from RAID groups of hard drives. As LUNs are provisioned, the user assigns them to one or more pools of storage. As actual disk size increases, operators must make adjustments to their volume assignments. You get to keep your HP or HDS investment, but thin provisioning could be more manual.
Match a solution to your infrastructure. In today's world, what kind of storage you have will determine the kind of thin provisioning you choose. As noted, with 3PAR's solution, you need to use 3PAR's hardware. You can stick with your own storage preference if you're an HDS or HP user, but your administrative tasks may include a bit more manual intervention to map LUNs on and off storage pools. Then again, you could pick Datacore and use a variety of third-party arrays -- as long as you like its software features and functions.
Check the operational impact. As noted, thin provisioning solutions vary in the levels of manual intervention and administrative supervision they require. Thing is, they may not differ enough to make a difference in your choice. Claus Mikkelsen, chief scientist and industry expert at HDS, thinks all storage provisioning requires storage know-how. "Whatever it is, you'll always have up-front work to allocate thresholds." HDS helps, he notes, by offering default thresholds for storage pools, LUNs, and arrays. Administrators would have to set those in the course of normal provisioning. Thin provisioning streamlines the work significantly, he maintains.
Know your data. A good implementation of thin provisioning involves a solid understanding of how data consumes storage, sources say. The key is to understand the growth rate of an application and be able to anticipate when thin provisioning will have to give way to actually supplying more storage.
"If an application actually must periodically touch the blocks [of data], as happens in defragmenting or reorganizing a database or in table optimization, you could see a spike in demand," notes Greg Schulz. To avoid unpleasant surprises, it's important to know how data is likely to perform and what its storage requirements really are.
Check the integration with storage management. Thin provisioning, tied as it usually is to a specific vendor's storage, needs to be checked for its ability to integrate with existing storage management applications. In many instances, your favorite monitoring or SRM application may not be usable once you implement thin provisioning. Test, to avoid disappointment. (See Sunlight Service Group.)
With so many caveats, it's easy to see why some users are waiting for thin provisioning to perform awhile longer before taking the plunge. (See The Skinny on Thin Provisioning.) Today's solutions are generally tied to a specific vendor's hardware, and it's tough to sift some vendor claims from actual fact.
Nonetheless, as data continues to grow and budgets don't, this technology will no doubt play a crucial role in storage networking. Stay tuned for further developments.
Mary Jander, Site Editor, Byte and Switch
Compellent Technologies Inc.
DataCore Software Corp.
EMC Corp. (NYSE: EMC)
Hewlett-Packard Co. (NYSE: HPQ)
Hitachi Data Systems (HDS)
Network Appliance Inc. (Nasdaq: NTAP)
The StorageIO Group
3PAR Inc.0
You May Also Like
Maximizing cloud potential: Building and operating an effective Cloud Center of Excellence (CCoE)
September 10, 2024Radical Automation of ITSM
September 19, 2024Maximizing Manufacturing Efficiency with Real-Time Production Monitoring
September 25, 2024