File-based storage has been standard for years, but now object-based storage, also known as content-addressed or content-aware storage (CAS), is making inroads on the file-based market.
"The market for object model archiving alone is over half a billion dollars," said Brad O'Neill, senior analyst for the Taneja Group, a storage research company in Westborough, Mass. "If you add in object model approaches to backup and restore, we're going to be looking at a billion dollar plus market for all object model storage in 2006."
The interest in object-based storage is driven by several important differences in the technologies.
Traditional file systems organize data via a hierarchical scheme of directories and files tied closely to application and operating systems. This complex and tightly coupled hierarchy makes it difficult to dynamically access the data across multiple applications and operating systems -- a rigidity that O'Neill believes will prove increasingly unacceptable to end users. On the other hand, object-based storage systems use specific meta data "tags" that are unique to each content element. "This content-based addressing schema enables object-based systems to create digital assets that have been processed by an object-based addressing technology and enhanced with meta data that enable the asset to be utilized as an independent resource," O'Neill said. What this ultimately boils down to is that IT managers can use the meta data, which describes attributes such as type of data or who owns the data, to more efficiently identify, track and access data regardless of application or operation system association.
Object-based systems include
HP, RISS, EMC, Centera, IBM, DR 550, Sun Microsystems, IntelliStor, DataCenter Technologies, DCT Content Director, Archivas, ARC, Permibit, Permeon, Nexsan, EverTrust
File systems' performance erodes as they approach the limit for address space available for storing data and migrating a production environment to a new file system is time consuming, hard work. The need to continually manage and migrate file systems can pose a significant challenge. Object systems, on the other hand, can expand easily to high capacities across multiple applications and storage media by using their CAS capacity.
Moreover, object storage uses single instance storage, in which multiple meta data tags, each tailored to a specific user's requirements, can point at the same piece of unique content. This can lead to a hefty reduction in the amount of storage needed overall.
File systems and the operating systems on which they depend are designed and certified for deployment with specific disk types and protocols Object-based systems are neutral to the storage media in use, making it much easier to migrate to new storage without disturbing the integrity of archived materials.
These attributes have driven a fair amount of market interest in certain niche areas that need to store fixed content for long periods, said Tony Asaro, senior analyst at the Enterprise Storage Group, a storage research company in Milford, Mass. "You have more control over that data than you would have normally," he said. "And because the applications are so tightly linked [through proprietary APIs], you can retrieve quickly as well."
Object storage applications
E-mail archiving remains the most popular use of object-based storage, as industries dealing with regulatory compliance requirements, such as the Sarbanes-Oxley Act or the Health Insurance Portability and Accountability Act , find object based storage's use of meta data tags very helpful in terms of data security. "For example, one could ascribe a meta data tag that restricts access to certain data only to the accounting department," said Michael Karp, senior analyst at Enterprise Management Associates in Boulder, Colo. "Somebody without the seal of approval won't have access. Because there's more intelligence built in with the data, you know more about the data that's stored."
It also helps automate storage policies and procedures by providing meta data for the policy engines to access and act upon, which can help storage managers set up tiered storage strategies.
With such market potential, small wonder that vendors are jumping in with both feet. EMC Corp.'s Centera system has dominated the market for several years, but that's no longer the case. IBM and Hewlett-Packard Co., among others, have competitive systems on the market (see side bar). And, Asaro said, although traditional NAS systems can't scale to the level that object-based storage can, next-generation NAS systems will address some of those issues bringing in new competition. For example, storage managers can use Snaplock API to deploy Network Appliances Inc.'s NAS in the same manner, Asaro said. "It allows them to do what EMC does with Centera," he said. "What the NAS guys have said is that they can turn their stuff into CAS."
In the long run, however, vendors aim to add their presence to a new market, not cannibalize the old one. "In disk-based archiving for unstructured content, I'm willing to bet that we will see object-based approaches essentially eliminate traditional hierarchical file system approaches within the next 36 months in large enterprises," O'Neill said. However, he urges IT folk to keep things in perspective. "Production storage environments will be leveraging file systems for the foreseeable future," he said. "Don't expect a paradigm shift in production data anytime soon." Asaro agrees. "CAS absolutely will not replace primary NAS storage," he said. "They're used for different things."