Bringing data backup into practical application: hierarchical storage management offers storage solutions for the dynamic enterprise. (Storage Networking).

Hardware storage solutions have matured exponentially in the last several years to the point where once cutting-edge technologies, like storage area networks (SAN) and network-attached storage (NAS), are converging and tape library capacities have greatly increased. Thanks to new applications that allow greater access to data, rich media functionality, and unreserved file sharing, companies are more productive than ever, and storage technologies have successfully kept pace with the change. But with growth comes complexity, and improved storage capacities are meaningless without the ability to manage them effectively. While the average person might consider storage as simply backing up a server to disk, is this really using storage to its fullest capacity?

The easy answer is no, it's not. The more complete answer, however, involves a different question: How can businesses effectively utilize data, not just store and protect it, within an enterprise network? One solution that's opening new doors for network managers is hierarchical storage management (HSM), a concept that's been reinvented thanks to new network technologies and an increasing use of data in new applications.

Hierarchical Storage Management Defined

HSM is a software-enabled storage management method that is based on pre-determined policies that govern data migration. HSM systems automate file backup and archiving by managing data movement among tiered storage devices, such as SANs, disks, and tapes. Based on preset parameters of percent of usage, time, size, significance, and more, the driving idea behind HSM is to move rarely used files from networks or hard drives to more cost-effective archival media, such as tapes, in order to reduce the cost of storage management. HSM also saves money spent on replacing expensive media, since the process moves less-timely files to less-expensive media. This data movement is transparent to the user, who can quickly retrieve files whether they reside on a backup storage medium or an online network.

Thanks to the development and maturation of storage networking and crossbred systems, an affordable, flexible three-tier structure is replacing the previous HSM model, which was built on disk storage devices, optical jukeboxes and tapes. Today, the three-tier design is more efficient and provisional: SAN/NAS for online storage, followed by near-line disk storage and finally reliable, affordable tape storage for archiving and backing up files.

As opposed to other tiered storage system strategies, HSM promotes automated, hands-off data storage management, which is a crucial benefit for enterprise administrators who need to sustain data availability and limit, or eliminate, system downtime. According to an October 2000 Meta Group study, companies that are most dependent on automated systems--such as energy energy and telecommunications enterprises--accrue an average of nearly $3 million in losses for every hour of unexpected downtime, based on lost revenue and employee idling. Even industries that are less dependent on IT infrastructure--such as health care, media, and hospitality--lose between $330,000 and $636,000 of revenue per hour. The chance of unplanned downtime is not one a company can afford, arid an HSM solution reduces the risk.

Storage Networking Advances Usher in New HSM

Essentially, the goal of HSM is to align storage technology with application functionality, resulting in easier administration and better business. With new storage networking architectures, such as NAS, SAN, virtualization, InfiniBand, and iSCSI, creating an ideal web of networked users with seamless data sharing and storage capabilities, the HSM model emerges as an ideal choice in backing up all levels of data residing in multiple network locations. However, confusion exists about what HSM is and how to implement it. In fact, industry analysts consider general confusion in the marketplace one of the biggest obstacles to its adoption, explaining why the concept has not been more widely accepted outside the mainframe marketplace.

Although data backup is often considered the primary function of HSM, the need to effectively and efficiently use lower-cost storage media via HSM software is also a driving force behind the management approach. HSM has existed for about a decade and, until recently, its primary use was for specialized applications and system design. At the time, the approach of many administrators was simply to add more RAID as storage space was depleted. However, major corporations quickly learned that merely adding RAID was expensive and perpetuated the need for additional administration to determine what resided on it and how to properly control the expansion of the server system. As many administrators and companies now understand, adding RAID does not resolve capacity or management issues, and the search is on for a better approach, such as HSM.

Putting Theory to Work

An HSM system offers businesses time and cost savings that can be attributed to reduced management hours, maximum storage exploitation and cost-effective media usage. How-ever, in order to build an effective system, network managers must coordinate several aspects of an enterprise's overall business structure.

The initial goal of planning an HSM system is to devise data management policies that are in line with a company's business objectives. Network administrators must first look at their company's storage system as a microcosm of the entire company's needs. As the pre-set parameters of HSM define when data is moved to and from various storage devices based on factors like size, usage frequency, last access date and more, the network administrator becomes a digital content archivist of sorts, defining what's essential for today's business and how to recall yesterday's information. Network administrators must ask questions such as: What type of data is being stored on a daily basis? How many email messages does each user receive per day, and how often do they access their archives? Which data is most critical to our company, and where does it reside? The upfront decision-making process about these policies is the essential step in bridging the digital side of an enterprise with the human side: The HSM design must align with day-to-day end-user needs.

Once policies are defined, technological considerations must be addressed. As mentioned, HSM software is only beneficial if it is transparent to the end-user, but it must also be invisible to other software applications. It is middleware that has to be implemented in an open environment and cannot interfere with or affect how business applications access data. Data within an accounting database, for example, must be moved by the policies within the HSM software to the appropriate storage medium without affecting how that data is integrated into billing or other accounting procedures.

In fact, approaching HSM integration in this top-down manner-that is, looking at the design from the application layer down to the storage network, rather than the other way around--actually proves the point of why it's such an efficient system. Application-specific HSM installations allow businesses to move highly specific pieces of data, which may be essentially wasting valuable server space, to more cost-effective media. For example, a company with hundreds of email users has thousands of email attachments that are essential to business but waste time and server space when they are downloaded or freeze. Using a designed HSM system with connected servers, software and Sony Petasite TLU systems, businesses can automatically move email attachments to near-line storage while keeping email online, thus freeing server space, relieving email maintenance concerns and improving overall efficiency. Best of all, the user doesn't know the difference.

Bringing HSM to Life

Possibly without even realizing it, particular vertical industries have business objectives that perfectly map into an HSM design. Project-oriented markets, such as the rich media, medical, scientific, seismic exploration, and pharmaceutical fields, generate massive amounts of data used for certain periods of time and then stored as reference material for up to decades. For example, instead of backing up important medical information to disk, the pre-set parameters programmed by the network administrator will move this data offline to a less expensive tape format. While this data is still accessible to the user, its exact location is transparent and seamless upon retrieval. Hawaii's Kuakini Medical Center applies storage technology in this way, backing up radiological images, training videos, patient medical records and research data to Sony's tape-based PetaSite system library, which remains instantaneously available to every user.

As demonstrated by the oil industry in real-world scenarios, the use of HSM for seismic exploration systems can save RAID space, personnel time, and budget, and it can also make information available faster to enable intelligent decisions for oil drilling. By employing a time-and data-efficient HSM system, oil companies such as Conoco have been able to increase the hit ratio of their oil drillings and save millions of dollars in potential costs along the way.

Additionally, companies that deal with rich media content such as video, photos, and audio are beginning to integrate narrow and broadband data into a single file. For example, a piece of video for today's news might also have a written script attached to it that becomes tonight's Web content and tomorrow's newspaper copy. Storing this content as if it's a simple computer file, and thus simply backing it up to disk, is not the optimal approach. Instead of tying up the server with high-bandwidth video, HSM allows the video to be stored offline, while enabling the entire file to remain instantaneously available. For example, Telemundo is using Sony's PetaServe HSM software to interface file movement to its video servers. In another application, Sony's PetaSite system library provides near-line storage for the Starz Encore network, transferring files to play-to-air servers and allowing the network to digitize and repurpose its assets for multiple channels of distribution.

Complete Storage-Management Integration

In addition to the storage system itself, implementing HSM brings overall benefits to an enterprise's network by freeing bandwidth, releasing bottlenecks in the LAN and minimizing infrastructure complexity. With businesses re-purposing data through techniques, such as Six Sigma and customer relationship management, digitizing rich media audio/visual materials into multifaceted files and generally growing storage needs at an exponential rate, integrating HSM into a business IT system makes more sense now than ever. The inefficiencies of simply backing up servers to disk creates a solution that's no longer affordable or safe for enterprises that want to move forward by leveraging storage technologies.

HSM can empower businesses with automatically managed storage, while keeping pace with the latest advancements, so administrators can simplify data management as storage networking architectures continue to branch out and become more intricate.

www.sony.com

Thomas Yuhas is director of data systems at Sony (San Jose, Calif.)