Tape automation extends its 5-year run: advantages keep tape center stage.

For more than five decades, tape has provided the lowest cost per gigabyte of any data storage technology. This benefit is maximized when multiple tape cartridges are used in conjunction with automated media handling devices. More than 80% of the world's digital data is stored on tape. More importantly, tape is the central technology in all disaster protection architectures. The importance of tape in disaster protection and recovery cannot be overstated. Early-automated storage programs include:

[ILLUSTRATION OMITTED]

* IBM 2321 Data Cell (mid 1960s)

* IBM 1360 Photostore (1960s)

* Grumman Masstape (late 1960s/early 1970s)

* Precision Instruments Unicom (late 1960s/early 1970s)

* Ampex Terabit Memory (early 1970s)

* Xytex Automated Tape Library (early 1970s)

* Control Data Scroller (early 1970s)

* IBM 3850 Mass Storage System (mid 1970s)

* Control Data 38500 (mid 1970s)

* Fujitsu M861 (1980s)

The myriad types of storage media included reels, chips, strips, cylinders, and sheets of tape or film. Even the earliest of these historic subsystems were capable of access times of less than 10 seconds and capacities of more than 100GB. Reliability problems and high maintenance costs engendered by mechanical complexity conspired to limit the acceptance of most of these early devices. However, in 1988 StorageTek achieved a breakthrough in reliability for the industry when it introduced its legendary ACS4400 tape library. The tape library industry would never be the same after that.

Library Characteristics

The key components of tape libraries are: provision for two or more tape drives; racks or shelves to store media; one or more robotic subsystems to move media between drives and storage slots; media; an electrical interface and controller; and library management firmware/software.

Tape Drive Characteristics

Tape drives are a crucial building block of tape libraries. Tape drives utilized in tape libraries include 8-millimeter (Exabyte VXA, and Sony AIT), DLT and Super DLT (Quantum), LTO Ultrium (Hewlett-Packard, IBM, and Quantum), IBM-compatible half-inch 3490 cartridge, StorageTek 9840/9940, and Sony SAIT. These tape drives span a wide range of performance characteristics and cost, and produce an equally wide range of library characteristics and costs.

Media Characteristics

Most types of media cartridges are available in a variety of capacities and tape lengths and are widely distributed. Eight-millimeter data cartridges are similar to but are substantially more reliable than 8-mm consumer videocassettes. DLT, SDLT and LTO cartridges are similar in size and shape to IBM half-inch cartridges, enabling them to be easily integrated into previously existing robotic systems.

IBM-compatible half-inch cartridges are produced in high volume at low cost by a number of manufacturers. These cartridges remain a popular worldwide data interchange standard. Other half-inch cartridges--including StorageTek's 9840--are physically similar to IBM's cartridges, allowing easy adaptation and upgrades into existing half-inch robotic systems. Media used in helical scan SAIT drives are data-certified versions of those used in professional broadcast applications.

Library Scalability

Some library manufacturers develop a library family based upon a single drive and media platform. Members in such a library family range from very small boxes to very large boxes, each is designed to accommodate a specific number of drives and media of the selected tape technology. Each member of the family draws from a common set of components; the number of these components (drives, media, robots, storage racks, and interfaces) are varied to yield the desired levels of performance, capacity, and cost for a particular target market. This "vertical scaling" approach accelerates new model introduction cycles, simplifies manufacturing and support operations, and reduces overall costs.

Other manufacturers build a line of libraries based upon a single robotic and enclosure design--a single-design box approach. While the number of components in the library remains relatively fixed, the type of drives, media, and storage racks are varied to yield the desired performance, capacity, and cost levels. Many new library architectures embrace this approach to incorporate physically similar media cartridges into the same robotic design, even extending the concept to mixing and matching media types within the same library. For example, the similarity between DLT/SDLT cartridges, LTO cartridges, and half-inch cartridges is so close (and the drives have similar form factors) that essentially one physical library design accommodates all three technologies.

Many libraries are offered in partially de-populated models as standard configurations. These units can usually be field upgraded in response to increasing future capacity and performance requirements--a concept know as Capacity on Demand. In many instances, multiple units of the same or similar module can be grouped or stacked to yield higher capacity/performance library systems.

Library capacity is calculated by multiplying the specified media count times the specified capacity per media unit based on the selected recording format. Library models with fewer than the maximum number of cartridges will have proportionally lower capacities.

Library transfer rates are calculated by multiplying the transfer rate of a single drive times the number of drives. Library models with fewer than the maximum number of drives will have proportionally lower transfer rates. The maximum library transfer rates are often not achievable at the system level due to interface or network bandwidth limitations. Use of data compression typically doubles capacity and transfer rate.

Tape Standards

For more than five decades, tape has been the medium of choice for data interchange across all classes of mid-range and large computer systems. Compliance to standards is the foundation of this widespread acceptance. Standardization invites and guides new suppliers into a market, enabling a more competitive environment while providing a stable platform from which new products evolve. Users benefit by the resulting compatibility and media interchangeability.

Tape Library Software

Ultimately, the performance and capabilities of a library are as dependent on the library management software as on the hardware. Software techniques (such as caching of data and queuing of commands) can provide the user with a transparent, high performance, near-online storage subsystem. The library management component may provide some or all of the following functions:

* robotics control

* queuing of user requests

* control of media movement

* management of all data in the library

* interface control

* device status monitoring

* device status reporting and diagnostics

Library management software may be available from the library hardware manufacturer or from a third-party software developer.

Applications

Key characteristics (such as capacity, transfer rate, and cost) determine which types of libraries are best matched to a given type of application. With multiple or mixed drives and media or by partitioning a single physical library into multiple logical libraries, a library may serve several applications simultaneously within an enterprise, significantly enhancing the utility of the library.

Archival Storage

Tape is the overwhelming computer medium of choice for archiving. Ninety-nine percent of all archived data are stored on tape. Key requirements for archival storage include long-term data integrity, resistance to environmental contamination, very low media costs, and low shelf-space costs. Archived data are typically stored in a temperature and humidity controlled, secured, offsite environment.

ILM

Tape's low cost makes it the clear choice as the final destination of aged files. Compliance to recent government regulations often mandates long term storage of massive amounts of data.

Disk Backup

Disk backup is a key application for all tape subsystems, including libraries. The low cost per gigabyte of tape media is an important factor in this application. When used in a disk-to-disk-to-tape configuration, libraries operate essentially in an off-line mode, thus speeding the backup process.

Data Collection and Distribution

These applications include collection, storage, and transmission or transport of accumulated data, typically in very large sets. Industries served include telecommunications, energy, space exploration, and natural resource exploration. Types of data collected include seismic and geophysical, air quality, and environmental pollution, as well as digitized images.

Near-Online Storage

Tape is the preferred offline storage technology. When online, access time to data on tape is typically several seconds vs. milliseconds for data on magnetic disk. Automated tape storage subsystems represent an alternative to offline and online storage for high-capacity storage by providing much faster access to data than traditional offline shelf storage of tape.

In a typical near-online system, less frequently used data files are stored in tape libraries. The large differential in storage costs between tape and disk compensates for productivity losses caused by the slow access of tape. Compared to offline storage in which tapes must be manually retrieved and loaded, tape libraries operating in the near-online mode speed the retrieval process and eliminate operator errors.

Image Storage

Digital images, especially high-resolution color or motion video images, are storage-intensive. Very high tape capacities, high transfer rates, and low cost per gigabyte are important factors for these applications, which include storage of financial records, e-mail images, and long-term storage of medical images. Libraries facilitate this application by automating the retrieval of image files.

Remote Backup

Remote backup of very large disk systems--also known as electronic vaulting--is a form of disaster protection for mission-critical data. Data are transmitted to a secure offsite location for recording and storage. Remote backup can provide the primary backup copy or can provide a redundant copy of the data.

HSM manages the storage of the data within a computer system across a hierarchy of storage technologies according to a set of programmed rules. HSM has its roots in the mainframe environment and is yet to be popularized outside of the mainframe environment.

These rule-based expert systems automatically and transparently migrate files between various layers in the hierarchy. The rules of migration--established by the user or administrator--may be based on frequency of use, storage cost, performance, security, or other criteria. Each layer of the hierarchy is characterized by a set of unique attributes and is embodied by a storage device such as a tape library.

Recent and Future Developments in Tape

Here are the major milestones achieved by the tape industry in 2004 and other developments expected shortly:

* Third generation LTO products (400 GB, 60-80 MBS native) were launched in late 2004. Fourth generation LTO products (800 GB, 120-160 MBS native) are expected in 2006.

* Quantum's third generation SDLT products (300 GB, 36 MBS native) were introduced in the fourth quarter of 2003. Fourth-generation SDLT products (800 GB, 60 MBS native) are expected mid-2005.

* Sony's fourth-generation AIT products (200 GB, 24 MBS native) were launched late 2004. Fifth-generation AIT products (400 GB, 48 MBS native) are expected to be introduced in 2006.

* Sony's first-generation SAIT product (500 GB, 30 MBS native) was launched in 2003 and piqued user interest as the highest capacity tape solution available. Second-generation SAIT products (1 TB, 60 MBS native) are expected late 2005.

* StorageTek's third-generation 9840C drive (40 GB, 30 MBS native) was launched in 2003 as the industry's fastest access tape drive. Their second-generation 9940B drive (200 GB, 30 MBS native) was launched in 2003 as the industry's fastest access tape drive. Their second-generation 9940B drive (200 GB, 30 MBS native) was launched in 2002 as the industry's then-highest capacity tape drive.

* IBM's 3592J (300 GB, 40 MBS native) was introduced in late 2003 as the highest capacity/highest performance/fast access drive for the enterprise.

www.freemanreports.com

www.hp.com

www.exabyte.com

www.storagebysony.com

www.storagetek.com

www.quantum.com

Robert C. Abraham is principal analyst at Freeman Reports (Ojai, CA)