In-Depth

TapeLabs: More than Just VTL

VTS: three new (and important) letters in the storage world

• By Jon William Toigo
• 10/18/2005

In its quest to find purpose and meaning in “multi-tiered storage,” the industry has resurrected the Virtual Tape Library (VTL) from its mainframe-era slumber. Vendors are advising consumers to use their second tier of disk (typically SATA arrays because of their generous capacity and comparatively low cost) as a virtualized automated tape library.

In current industry parlance, a VTL is defined as “an archival backup solution that combines traditional tape backup methodology with low-cost disk technology to create an optimized backup and recovery solution.” Some vendors further define VTL as “an intelligent disk-based library that emulates traditional tape devices and tape formats.”

VTL backups are alternatively presented as a secondary backup stage on the way to tape, or as their own standalone backup/restore solution—promising faster restore of individual files accidentally deleted by careless end users. Either way, a VTL “solution” generally consists of an appliance or server operating software that emulates traditional tape devices and formats. Some vendors include disk arrays with the tin-wrapped software to deliver a “one-stop shop,” while others sell only the server or software, enabling consumers to use legacy disk or new arrays of their own choosing to create a repository.

Operationally, backup data is streamed to the VTL (disk drives) just as it would be to a tape library, but presumably with greater speed because of the rotational latency advantages of disk. This is less and less the case with the increased speeds and feeds of today's tape drives, leading some vendors to shift their argument for VTL to focus on the number of tape drives that can be emulated by their solution. The suggestion here is that the more drive targets you create, the faster your backups will occur (which is also a problematic claim).

Differentiators in this product class are exceedingly thin, and so are the profit margins. Some vendors encrypt the data streams that are written to disk, while others compress the data once written. Before his departure from OTG Software, and during his time at Quantum as CTO, Michael Del Rosso was endeavoring to add much more value to the story. He envisioned VTLs and Tier 2 storage generally as being a platform for adding data management and data security to I/O. He might have wanted to talk to Tape Laboratories (TapeLabs) in Los Angeles, CA.

Unless you have been operating in the world of Tandem Non-Stop Computing, you may not have heard of TapeLabs. They are well-known and their value proposition is well established. Founder and CTO Allen Ignatin, whose personality blends equal parts curmudgeon and visionary, is one of those people who thought up much of the technology that goes into VTLs today—and he is by far the biggest critic of the VTL wares in the market.

Ignatin notes that the original premise of VTL in the early 1990s was to optimize the utilization of once-expensive and (by contemporary standards) very low capacity tape media in mainframe shops. Products such as StorageTek’s Iceberg, among the first VTL solutions, enabled backup data sets to be staged and combined into an image that would completely fill each expensive and bulky tape cartridge. The payoff: fewer tapes to manage and lower overall tape management costs.

The story of VTL advanced when backups became a problem in the distributed world. When a company grew large enough to justify the centralization of backups and the acquisition cost and use of a common tape library, VTL resurfaced as an answer to the problems of resource scheduling and slow tape device operating speeds. The latter, by the way, was partly a function of the ratio of disk and tape media capacities in the 1990s that required tape cartridges to be changed in the middle of a backup, and partly the result of slower tape drive performance relative to disk at the time. So, the concept of VTL was enlarged to provide a way to get backups done in limited operational windows—to emulate a tape library and to buffer the real-world tape device with a few ranks of disk.

That’s where things went awry, in Ignatin’s view. Today, most VTL solutions, he complains, are hitting boundaries that are based in large part on their designer’s decision to emulate a specific library. Just as you might max out a real tape library, Ignatin notes, some consumers are discovering to their chagrin that it is also possible to max out their emulated subsystem.

Hearing this criticism coming from one of the founders of the technology intrigued me so much that I felt compelled to accept TapeLabs’ invitation to LA to learn more about its move into the overcrowded field of VTL vendors in the “open systems” space. By the end of the day-long meeting, I was glad I did.

TapeLabs doesn’t sell VTL: it sells VTS—a virtual tape system. Add this to your overloaded dictionary of storage acronyms because the difference is important. As Ignatin describes it, VTL is an emulation of a real-world library. A virtual tape system or VTS, by contrast, is very different.

TapeLab’s VTS offering might be best described as a hardware-agnostic data management system aimed at upper mid-tier to large-scale enterprises. It provides a range of data management services that enable companies to comply with regulatory mandates, to protect their most irreplaceable asset against loss, to recover rapidly from unplanned interruptions and data-corruption events, and to selectively archive key data, while providing necessary privacy and security guarantees to data along the way.

So how is this different from a VTL? First, with TapeLabs VTS you aren’t emulating a library at all. You can simply create tape drives of the brand or type that you need whenever you need them (restricted only by the limitations of memory and disk). Second, the TapeLabs VTS is a software platform for doing things that go beyond traditional store-and-retrieve (though it also provides these services with alacrity). There is a full set of media management functionality to aid in cataloging tapes and tracking media expiry, plus policy-based automation that can help you identify what to back up, at what intervals and to which device (VTS supports writing to disk as well as directly to tape). In this area, TapeLabs may actually have the edge on competitors, since their product also enables you to construct policies for deleting data from your backups and archives—something that is required to keep backups and archives in sync with regulation-driven policies for orderly data deletion in production storage.

Encryption has always been a feature of the product, a natural given TapeLab’s “dark past” in the field; among Ignatin’s first efforts was a board-level encryption product. You can optionally encrypt data being stored to disk or tape using the product, so you will avoid the embarrassment that has recently touched many companies that have lost their backup tapes off the truck transporting them to off-site storage.

Data compression is another VTS option that effectively reduces the amount of disk storage capacity used by up to 90 percent (10:1 compression ratio), although more typical compression rates are in the 2:1 to 4:1 range. The software compression engine efficiently compresses files as they are being written "on-the-fly" to disk storage resources. The compression algorithm, specifically developed by TapeLabs, exacts no overhead and in many cases increases performance and throughput by decreasing the volume of data written to the disk resources.

All in all, porting VTS to the “open systems” world and using it in place of traditional VTL offerings in that space seems like a good idea. It holds out the promise of using second-tier storage as more than simply a tape surrogate and for making it a more meaningful part of a managed data infrastructure.

Your thoughts are welcome. Write to jtoigo@toigopartners.com.