Can SAN Give HSM A Face lift?

According to Framingham, MA-based industry analysts at IDC Research, total storage requirements worldwide are expected to reach 3 million TB by 2000. This compounds key storage management challenges exponentially. In addition to needing a solution that offers scalability, high accessibility, and availability, as well as reliable recovery, new technical challenges have brought new meaning to "catch 22."

It seems obvious that vast amounts of data must be physically stored and backed up, but that data clogs storage channels and consumes large chunks of network bandwidth. Today's heterogeneous corporate IT systems consist of a wide variety of hosts, servers, client systems, and networks. Users demand immediate access to data and the new age of the Internet has thrust the world into global, 24/7 operations, leaving zero downtime for backup and storage maintenance. It's no wonder anxiety-ridden administrators consider their need for a better storage management solution to be at a crisis level.

Of all the new buzzwords and acronyms, perhaps none is more talked about (and less understood) than SANs, or Storage Area Networks. Is it the next generation of HSM (Hierarchical Storage Management)? Not really. Can SAN boost the performance of your HSM? Yes.

SAN 101

In theory, SAN is simple to describe--a dedicated highspeed network, separate from general network bandwidth, used to move information between the users and the storage resource. Information that was previously shared across conventional SCSI (Small Computer System Interface) connections (which limited data transfer to a few yards), can now be shared across highspeed Fibre Channel. By combining Fibre Channel and switching technology, organizations can connect servers and storage over distances as great as 10 kilometers at speeds of 100MB/sec or faster. The corporate network is freed from storage-related chores and communications because the servers are only attached to the storage for specific tasks.

SANs are inherently scalable. Through the use of switches and hubs, organizations can add storage capacity completely independent of the servers. Switching also enables truly hot-pluggable subsystems, allowing IS managers the freedom to add and replace storage without affecting operations. An organization's SAN is their connection to their data; it's the artery between the heart and the brain. How that data is stored and managed is another issue.

By 2002, the Palo Alto Management Group (PAMG) predicts the average data warehouse will reach 6.5TB and support as many as 100,000 users. But not all data is equally important and new applications such as Enterprise Resource Planning (ERP) and new data types such as MPEG2 video are constantly changing our perception of what constitutes vital information. It wasn't long ago that a novelty called e-mail was thrust into the critical data resource category.

The SAN/HSM Advantage

By marrying SAN with HSM, administrators can finally have the golden challis: scalability, accessibility, availability, recoverability, and economy. In the past, the one great hindrance to the success of HSM has been the speed of access to data. Traditional HSMs relied on SCSI interface, which offer speeds of only 10-20MB/sec and limit connectivity to within a few yards.

Ideally, an organization's storage management solution should be composed of complementary software, systems, and peripherals that are integrated, scalable, and modular. Such a flexible solution addresses both mission-critical, enterprise-wide requirements and business-critical desktop needs without sacrificing fast, reliable access to data. A SAN-empowered HSM solution meets those requirements with an exceptional level of price/performance and data security.

A SAN/HSM can provide organizations with three things that separate solutions cannot:

1. Lower total storage management cost;

2. Faster access to high priority data;

3. Greater data security through prioritized or regulated data access.

Traditional SAN solutions store data on RAID systems connected to the network via Fibre Channel. RAID offers users data redundancy, fast access, drive hot-swapping capabilities, as well as a high cost per megabyte, making a RAID-only SAN solution too expensive for implementation into many of today's cost-conscious organizations.

HSM Economy, SAN Efficiency

The average HSM solution efficiently manages data stored on magnetic disks, optical disks and less-expensive tape, CD, or DVD. However, for some response-driven organizations, delayed access to data can make HSM a more costly solution than a RAID-only SAN solution. Together, SAN/HSM can provide organizations with ultra-fast access to data stored economically on a three-tiered system accessible to, but separate from, the network.

HSM combines storage mediums to manage data more efficiently and more economically. Storing to a hierarchy of user-defined policies, the HSM automatically moves data between the various levels to eliminate manual storage maintenance, resulting in savings of time, storage space, and money.

For example, this month's sales figures would be a high priority and nearly daily access would be required. Access delays for such high-priority data would be costly. Those files would be stored on RAID to give users the fastest access to the data, while data mirroring would provide a safety net against that file being lost. The average cost per megabyte of a RAID-based first-tier HSM solution is about $0.11 to $0.15 (36GB\$4,000).

After the month-end report, the sales information would still be important, but immediate access wouldn't be required. MO solutions offer access speeds of up to 20msec, as well as a high-speed data transfer rate of over 5MB/sec. Industry standard 5.2GB MO media is virtually impervious to contamination or data corruption and provides a data life of more than 50 years. Leading MO suppliers offer jukeboxes ranging from gigabytes to terabytes that can be configured with multiple drives to provide on-line access to gigabytes of stored files. With an average cost per megabyte of less than one cent, MO is an ideal second-tier HSM solution.

At some point, less frequently accessed data can be moved to the final hierarchy level, utilizing a slower, less-expensive storage solution such as tape, CD, or DVD. The data here is still a vital resource to the organization, but, because file access is minimal, a compromise of slightly longer retrieval time is more acceptable, considering the cost per megabyte savings these technologies offer over the first two-tier solutions.

Although tape still offers the lowest cost per megabyte at less than one cent, many organizations are choosing to spend a little more for the random access offered by CD and DVD. CD solutions offer users exceptional flexibility and up to 24x playback at $0.02 per megabyte; but for some, the 5.2GB double-sided capacity of DVD with its superior performance read speed is worth the additional cost.

Data Security

Finally, a SAN/HSM solution can provide greater data security than an ordinary SAN solution through prioritized or regulated data access. As more and more organizations reengineer their businesses, an enormous amount of company-sensitive data is distributed and available over the network to everyone. However, the Regional Sales Manager doesn't need the same access to data as the senior vice president of sales and accounting isn't as concerned with units sold as it is with the amount they can expect to receive from each customer. The president and CEO, on the other hand, would likely require access to everything.

Fully integrated HSM solutions allow system administrators to monitor and regulate data access by assigning access levels to everyone on the network. A Fibre Channel-based SAN with its fast, OS-separate access to data can provide just the facelift HSM needs to carry such a valuable tool into the next century. Together, users have uncompromised price/performance and greater data security.

Mitch Cipriano is the vice president of marketing and sales at Maxoptix Corporation (Fremont, CA).