Storage Management Meets Data Management

The emergence of low-cost disk-drive technology is making disk-to-disk-to-tape storage infrastructures more commonplace. In fact, the cost-effectiveness of disk-to-disk-to-tape infrastructures is a central pillar among many Information Lifecycle Management (ILM) strategies because it supports ILM's idea that companies can move data onto less costly storage as it ages, eventually archiving it on tape.

One of the primary challenges for companies that try to implement ILM is the data-migration management process. And with these challenges comes the question of how and when data should be moved. Typically, with ILM and its antecedent, Hierarchical Storage Management (HSM), data migration has been part of storage management. But in an intriguing announcement at COLLABORATE 06, Oracle staked a strong claim that the database itself is the answer to manage data migration for Oracle databases. The release of Oracle Secure Backup--a high-performance network tape-backup solution for Oracle databases and file systems--coupled with announcements over the last two weeks from IBM and Sun Microsystems, suggest that the line between data management and storage management is blurring. Data-management technology providers are increasingly including storage-management functionality in their products, while there are signs that storage-management vendors are moving more aggressively into what was once the domain of database and application vendors.

You could read the Oracle Secure Backup announcement in two ways. On the one hand, Oracle points out that Oracle Secure Backup is just a continuation of an effort that it started with Oracle Database 10g Release 1, which included disk-to-disk backup technology, and Oracle Database 10g Release 2, which included encryption. Oracle Secure Backup completes Oracle's database solution by enabling DBAs to manage the entire backup process from the database-management console.

On the other hand, Oracle Secure Backup isn't just evolutionary because for the first time, according to Oracle, backup logic is embedded in a relational database engine. Oracle says this provides faster, more efficient backups. Moreover, as Mark Townsend, senior director of Oracle's database product marketing, told me, to effectively implement ILM and to meet compliance requirements, enterprises have to better understand the data itself. The database, he says, sees the data as more than just bits and bytes.

Oracle isn't the only database vendor improving the storage aspects of its technology. IBM recently announced that its next release of IBM DB2, Viper, will include data-compression technology, code-named Venom. Venom will allow IBM to deliver mainframe storage compression in non-mainframe platforms. Venom features "row compression," which the company claims yields significant disk, I/O, and memory savings, particularly for large tables with repetitive data patterns. It's more efficient, IBM argued, than Oracle's table-compression technology. According to IBM, by using Venom, data that once required 1TB of storage might require only half a terabyte.

In addition, IBM is enhancing Viper's automated storage-management features, which currently require administrators to make numerous manual and incremental changes. The company is also quite clear that it hopes the Venom enhancements will position IBM to compete not only with its traditional rivals Oracle and Microsoft but also with storage heavyweight EMC. If database vendors are moving into what was once the storage domain, storage vendors might be moving into areas once reserved for application and database technology. In a major announcement of new NAS products, Sun Microsystems laid out its broad vision for future storage-technology developments with integrated data management as a key element. Sun's other new technology under development, Project Honeycomb, will feature a new type of platform that blurs the distinction between application server and storage, thereby giving application developers new flexibility in the way their applications are deployed by offloading low-level data services from the application server to storage.

Efficiently managing large-scale storage infrastructures has always been complex. Although simplicity is a common mantra among vendors, these new releases bring more concerns for IT. Not only must administrators worry about how to execute storage functions, they must now determine where in the application stack these functions take place. The boundaries between storage management and data management have definitely shifted.