|
 |
|
General Information
|
Add To My Personal Library |
November 7, 2008
• Vol.30 Issue 45
Page(s) 1 in print issue
|
Stop Storage Overload
Putting The Brakes On The Growth Of Primary Storage
|
| Key Points
• Multiple data-reduction technologies on primary storage can be leveraged to significantly reduce overall disk capacity.
• Smarter storage systems allow for better provisioning and management from the data’s birth.
• Compression, archiving, and data deduplication offer cost savings and increased efficiency by freeing up primary storage in favor of inexpensive long-term storage. | | The growth in storage is unprecedented, but what is growing fastest is not the active data set—the data we are using this week, for example—but rather the amount of data that we retain. The problem is that all of this data, active and inactive, is being kept on primary storage.
The good news is there are solutions to address the problem, ranging from primary storage optimization via thin provisioning and data progression to data reduction through compression or deduplication to eliminating the inactive data through migration and archiving.
 Storage Optimization
Storage optimization will require new intelligence at the storage controller, typically a virtualized solution. Bob Fine, senior product marketing manager at Compellent Technologies (www.compellent.com), says, “According to analysts, the average utilization of storage is typically 25 to 40%.” This is the amount of assigned capacity of the volume that is actually being used and does not take into account the amount of data that is old or inactive. According to Fine, although resizing a volume is a capability most of today’s SAN solutions possess, the problem is it’s not always straightforward and always requires storage administrator interaction. As a result, administrators do not have the time or availability to always be on hand to resize a volume, so the alternative is to give the volume plenty of excess capacity—capacity that will take years to consume and more than likely will never be used.
Thin provisioning solves this by allowing the storage administrator to allocate large storage volumes, but the system only consumes storage as the actual disk space is used and does so automatically.
The next step in optimization is block-level data migration driven by the storage controller. These technologies can move data within a storage system to different classes of storage based on age or type. For example, as data becomes inactive, it can move from a high-speed Fibre class of storage to a higher-capacity, lower-cost (albeit slower) SATA class of storage. “It’s important to note that this is block-level data movement, so even old sections of databases can be migrated.” says Fine, who believes this technology can reduce storage costs by 70 to 80%.
Data Reduction
With the storage optimized, the next step is to reduce that data’s size. There are two methods available for this today: compression and data deduplication.
“Compression is the logical first step in data reduction, and it sets up data deduplication,” says Peter Smails, vice president of marketing for Storwize (www.storwize.com). Real-time compression of primary data allows for all data on a file system to be compressed and decompressed as needed with no impact on performance. These technologies sit on the network logically between the users and the file data being served.
With the file data compressed, the next step is to deduplicate it. According to Larry Freeman, senior marketing manager of storage efficiency solutions for NetApp (www.netapp.com), “Users should deduplicate where it makes sense—for example, focusing on VMware VMDKs, user home directories, or other data sets that contain a high degree of similar data.”
Compression and deduplication can work well together and are complementary: Compression works on all data one time; deduplication works on certain sets of redundant data multiple times. For example, it is not typical to have redundant copies of a database, but it can be large and will compress well. VMware image files, on the other hand, have a large degree of redundancy between them and will show high levels of efficiency from a deduplication solution as well as achieve a high level of compression.
Data Elimination
Chris Santilli, chief technology officer at Copan Systems (www.copansystems.com), believes primary storage is not growing at the same pace as the data that is being retained. “One of the latest Gartner studies [found that] 80% of data in most data centers has not been accessed in the last 90 days, and while data reduction technologies can make that data smaller, it still creates a burden on the environment,” he says. The problem is that this storage sits on expensive, space- and power-consuming primary storage and needs to be archived.
According to Santilli, archiving has a reputation of being a dark and gloomy storage vault that is full of dust and is not accessible. The challenge is that this data, while not needing to be on day-to-day active storage, does need to be on a storage platform that is accessible for responding to legal requests or even user requests for the data. The response to this need has been active archives typically based on disk technology.
Disk-based platforms bring better accessibility, ease of use, searchability, and response time. Additionally, moving data is a simple move command from one file system to another. “On the downside, traditional disk also brings lack of scalability and a hefty power appetite. By using technologies like power-efficient drives and dense packing of storage, a disk-based archive can bring much of the same cost and power efficiencies of tape [as well as the] accessibility and searchability that organizations need.”
These techniques can be implemented into a storage platform, with the user selecting the implementation that provides the most return for his expenditure and bandwidth. The user can then expand other data reduction techniques onto a disk-based archive as needed. 
by George Crump
The Effects Of Primary Storage Data Reduction
Reducing data on primary storage can have a number of effects on your data center. Here are three such effects.
Eased backup load. If, for example, an SME moves 80% of its data to an active archive, that means 80% less data will need to be backed up regularly by the backup application. Then, compressing the remaining data means a reduction in the data that has to be sent across the network to the backup server.
Cost savings. By implementing a primary storage reduction strategy, costs savings can be achieved upfront as well as during the life span of the data. Better storage provisioning and management by the storage solution means less storage is needed upfront, and archival, compression, and deduplication lead to less storage needing to be purchased during the life of that data.
Green benefits. Primary storage data reduction provides immediate power and cooling cost savings with few or no changes to the day-to-day business process. |
|
|
