When solid-state drives (SSDs) first started showing up in storage arrays three years ago, the lack of automatic storage tiering was cited as one of the major barriers to deployment -- along with cost. SSDs are still far from ubiquitous in storage arrays, but you can no longer blame a lack of automated tiering options for that.
Auto-tiering software is commonly available in today’s storage arrays. These include Dell Inc.’s Data Progression for Compellent systems, EMC Corp.’s Fully Automated Storage Tiering (FAST) for its major storage platforms, Hewlett-Packard (HP) Co.’s Adaptive Optimization for 3PAR arrays, Hitachi Data Systems’ Dynamic Tiering and IBM’s Easy Tier. These applications vary in the number of tiers they support and how much control they give customers over the process, but they all do sub-LUN tiering.
There are other ways of tiering data for SSDs. Avere Systems Inc., NetApp and XIO all move data from SSDs to spinning disk either by caching or building the process into the controller.
Automatic storage tiering can result in better storage efficiency and lower costs by minimizing the amount of the most expensive media in storage arrays. The process makes it easier to move less frequently accessed data off SSDs or 15,000 rpm Fibre Channel (FC) drives onto cheaper SATA or near-line SAS disk.
Barry Blakeley, infrastructure architect at Irvine, Calif.-based Mazda NA, uses Compellent’s Data Progression to move data across three tiers. Mazda’s tier one is 1.06 TB of solid-state drives, tier two is 27 TB of 15,000 rpm FC drives and tier three is approximately 72 TB of 7,200 rpm SAS drives.
Blakeley said before using Data Progression, Mazda migrated data manually from primary to secondary storage. The decision was made by committee, he said, and took a lot of time.
Now Data Progression will move rarely used data down to a more cost-effective tier, usually the SAS drives. With only four people on his team handling storage, Blakeley appreciates the automation.
“With Compellent, it’s all automatic,” Blakeley said. “The data migrates at the block level as it becomes stale and I don’t run into performance problems. I don’t have to decide by committee which data to move. It’s one of the things you set and forget.”
While most data storage administrators see the benefits of storage tiering, the amount of automation they want or need will vary. Some may be willing to completely trust the software or array to make all the right tiering moves, while others will want more control over the process.
“I think it’s too early in the game to say whether fully automated wins or not,” said Jeff Boles, senior analyst and director, validation services at Taneja Group. “I think the predisposition of the storage administrator out there is typically manual. We’d like more control. So can vendors develop enough trust to win over the skeptical storage administrator into a fully automated solution?”
According to a July 2011 Enterprise Strategy Group Market Landscape Report on storage tiering, “tiered storage is becoming more crucial . . . due to a simple matter of economics.”
“It’s been a reactive event in response to customers’ demands,” Taneja Group's Boles said of the rise of auto-tiering products. “So now we have every other vendor standing ready with another auto-tiering technology. And there's certainly a lot of marketing push behind it.
“It's not a technology that’s going to fade and go away some day," he continued. "This is going to become an integral technology in the storage system in the future.”
(Senior News Director Dave Raffo contributed to this story)