IBM this week disclosed a new automated tiered storage feature called Easy Tier for its DS8700 high-end disk array, designed to help customers move data more efficiently between tiers of spinning disk and solid-state drives (SSDs).
IBM won't issue a press release on Easy Tier until next week, according to an IBM spokesperson, but the vendor revealed details about the technology and other storage product enhancements Tuesday in letters to customers, and on the company website and executive blogs.
Easy Tier is a no-charge feature baked into the DS8700. According to IBM's website, DS8700 customers who don't have SSDs or the Easy Tier feature enabled can use another piece of software called the IBM System Storage DS8000 Storage Tier Advisor to "identify where and how much data in the system can benefit from SSDs."
Easy Tier builds on underlying DS8700 virtual volume architecture
IBM has made updates to the microcode of the DS8700 high-end disk array to take advantage of its underlying data layout and virtual volumes to move data among tiers of spinning disk and SSDs according to performance requirements.
According to a Taneja Group whitepaper posted on IBM's website, IBM's DS8000 series has long used virtual volumes, which IBM calls 1 GB extents based on groups of 64 KB stripes, beneath the logical volume or LUN seen by a host. From there, IBM can create a "heat map" showing all the extents and whether or not they're pushing the performance characteristics of each tier of disk within the array.
Looking at the heat map, customers can manually migrate data to ease performance constraints or set up Easy Tier policies to move data. Templates and custom policies will be available, and customers can select specific volumes for automated tiered storage or an entire pool. They can also set the performance sampling windows used to make up the "heat map" and schedule windows for data movement to take place.
While data storage administrators can see the potential value of automation, many are concerned with preserving control. "Our management is very interested in more tiering. And I like anything that can be automatic," wrote Bob Venable, manager of enterprise systems, BlueCross BlueShield of Tennessee, to SearchStorage.com in an email. "We need to control the automation to time windows of migration -- so production changes can be controlled."
"One GB chunks are less granular than some competitors," said Jeff Boles, a senior analyst and director, validation services at Taneja Group, as well as the author of the whitepaper. "But you have to think about how much granularity you can get before your algorithm is just creating noise as you're moving things around constantly."
IBM Systems & Technology Group exec Barry Whyte wrote in a blog post Tuesday that IBM is adding the same function to its SAN Volume Controller (SVC) storage virtualization device. "The same source code is being used for both DS8000 and SVC," Whyte wrote. "With SVC's variable extent size, you will be able to tune the granularity." IBM did not indicate when Easy Tier for the DS8700 will become generally available.
Automated tiered storage all the rage among storage vendors
IBM rival EMC Corp. has said that Version 2 of its Fully Automated Storage Tiering (FAST) software, which will add block-level granularity, is due out in the third quarter of this year. Easy Tier is preceded to market by two sub-LUN automated tiered storage offerings, Compellent Technologies Inc.'s Data Progression and 3PAR Inc.'s Adaptive Optimization.
"It appears that Compellent, 3PAR and IBM are leading the storage industry in delivering sub-LUN data optimization. At this rate, it appears EMC will be fourth or fifth by the time they release a sub-LUN option," said Analytico Inc. analyst Tom Trainer. "Further, it seems clear that sub-LUN optimization is only the beginning of a more fine-grained and autonomic data movement capability as data movement to the cloud becomes more of a requirement."
StorSimple Inc. came out of stealth last week with just such a feature — sub-LUN automated tiered storage the vendor claims is capable of passing blocks between on-premise and cloud tiers of storage. Another startup, Nasuni Corp., claims this feature at the file level.
"We'll see more storage vendors introduce not just array but infrastructure-based data optimization that'll move data from the data center out to the public and private cloud," Trainer predicted.
IBM's website also dropped details about more storage announcements coming next week, including new replication for the DS8000 and ProtecTIER data deduplication virtual tape library, as well as 2 TB drive support for XIV and LTO-5 support in its tape libraries. See our Storage Soup blog for more on those announcements.