Ultrium Reality Check.

Moore's Law still seems to work. In fact, in many places, Moore's Law has been surpassed.

Moore's Law originated from an observation that the number of transistors in an integrated circuit doubled every 18 months. This has been extended to apply to many electronic devices. Hard drives are doubling at a much more rapid rate than Moore's Law predicts. CPUs continue to increase in complexity, delivering higher performance and lower power consumption, again, at rates that are usually higher than Moore's law predicts.

The real problem with Moore's Law is that it applies to something that already exists. If you start from zero, doubling nothing every 18 months will leave you with nothing a century later.

In some ways, such is the case with Linear Tape Open. Members of the LTO Consortium have made tentative announcements of shipping and production schedules, but so far haven't actually delivered a working product. At Comdex last November, LTO drives were being shown at vendor booths and were seen working in libraries and changers offered by various automation vendors. Quantity shipments of Ultrium drives were expected to start in January 2000.

At the end of February, when this article was written, none of the three drive manufacturers in the consortium; Seagate, IBM, and Hewlett-Packard, had shipped functional drives. However, both Hewlett-Packard and Seagate parted the veil during February.

In mid-January, Hewlett-Packard invited me to visit their Boise facility at which they planned to brief me on their Ultrium drives and to show me some of the work in progress.

Hewlett-Packard's Boise facility is a large complex of mostly three story buildings. The buildings sport large numbers and are numbered in a somewhat illogical order, possibly representing the order in which they were built, rather than proximity to buildings with adjacent numbers. Buildings in the campus are interconnected and all seem to lead to the company cafeteria.

The meetings offered few surprises, enhancing data that was already available from HP, but also enhancing the HP approach to Ultrium. Clearly, Ultrium was a technology whose basic elements were agreed to by the three hardware partners. Just as clearly, one of the strengths of such a unified effort was the competition between the three manufacturers to distinguish its products from those of the others. Thus, the hardware partners in the LTO Consortium are faced with a two edged sword--the drives must all provide absolute compatibility of the basic features (a tape recorded by a Seagate drive must be readable by an HP or IBM drive, for example); but each company must also differentiate its drives from the competition's drives in order to earn market share.

In some ways, all three vendors share the same risk: if the first release of Ultrium drives should fail in some notable way, all three vendors could be seriously damaged. For example, if one company's drive should be sent into the market before it was really ready, what the public will remember is that the Ultrium drives fail, not that company X's Ultrium drives fail. This is only one of the reasons why the manufacturers want to be absolutely certain that their drives work flawlessly--the damage can be much more widespread than a hit to one company's reputation.

Hewlett-Packard has taken what may be the most interesting approach to differentiation--they are offering two different Ultrium drives. The drives, code-named Einstein and Newton are targeted at different price points and different markets.

Einstein, the company's high performance drive, is designed to compete with other Ultrium drives and should be priced to compete with SDLT once Quantum ships SDLT drives. The drive will be a full-height, 5.25-inch drive. It won't be as deep as the current DLT drives, making it easier to integrate the drive into automation devices or into enclosures. Configurations that include a front panel LED for system monitoring will be offered. The Einstein drive will have a native transfer rate of up to 20MB/sec.

Newton is a somewhat surprising offering. Newton is a half-height version of the Ultrium. It will be able to read and write tapes written by any other Ultrium drive, but it will do so at half the speed. This drive will have a native transfer rate of up to 10MB/sec.

Pricing of the Newton will place it near that of the DLT7000. The Newton will, thus, be a high capacity challenger to the DLT7000 and DLT8000 (which may be reduced in price to accommodate SDLT) while providing higher performance than the DLT7000. Newton will clearly be a competitor to the current high end of DLT, while also offering a lower cost alternative to competing Ultrium drives from HP, Seagate, and presumably, IBM.

Further, Newton will also be competing with AIT2 and Mammoth 2 technologies. The competitive niche being staked out by the Newton may give Hewlett-Packard an advantage over other Ultrium offerings by providing a clear alternative to current technologies at a price point that other Ultrium offerings won't be able to match. Clearly, where HP's Einstein and Seagate and IBM's Ultrium drives are Cadillacs, offering a Chevrolet that will still store as much data as any other Ultrium drive and be fully compatible with Ultrium may provide a strong market for the Newton.

A wide range of component tests was demonstrated to me during my visit to the HP labs. Although the LTO Consortium is using a lab that they've contracted with to perform compatibility testing, HP (and presumably the other manufacturers), is also performing various media compatibility tests, as well as tests of the various mechanical and firmware components of their drives.

Hewlett-Packard will fully support the Memory In Cartridge (MIC) capabilities of the Ultrium tape cartridge. The MIC component of the Ultrium cartridge will be readable using a radio frequency component. Hewlett-Packard's drives will support the basic set of data fields that were agreed upon by the LTO Consortium. Although the MIC component is severely limited to just a few kilobytes, HP is considering adding data fields to the core data set.

The utility of MIC was clearly demonstrated at the HP labs. For automation, especially, MIC is extremely useful. Using a handheld scanner or a scanner integrated into an automated picker, the unique identification of each cartridge-can-easily be read. A catalog can be made in minutes, merely by passing the reader past the cartridge. With a robust database built into a backup application, tracking each cartridge and the contents of the cartridge should be easy.

Conceivably, a handheld reader combined with a label printer would enable a user to scan the unique ID of new media and print a label reflecting the data. Conceivably, a tape cartridge can be tracked from birth (at the media manufacturer) through shipments to the ultimate customer, through its usage cycles, and finally, to retirement.

The initial Ultrium offerings from Hewlett-Packard will be available with SCSI interfaces. Hewlett-Packard plans to begin volume shipments in late second quarter or early third quarter of this year. Fibre Channel versions of the HP Ultrium drives will follow the initial SCSI drives by a few months.

I left Hewlett-Packard with considerably more information than I had when I arrived there in late morning. I also had a blank 50GB cartridge (yes, lower capacity cartridges will be offered by media manufacturers), a basic marketing brochure that included the technology roadmap, and a promise to provide additional information about HP's Ultrium once the information became available.

Following my visit to HP, I wanted to learn more about the Ultrium efforts of the consortium's other drive manufacturers. Seagate's Kevin Perry provided a briefing on the company's tape products a week ahead of the company's formal announcement of its Ultrium drive, the Viper 200.

The Viper 200 will provide native 100GB uncompressed data capacity. A native data transfer rate of 16MB/sec will provide a compressed backup rate of 1.92GB/mm, assuming 2:1 compression. The Viper 200 will include a feature that Seagate calls FastSense technology. FastSense automatically adjusts the tape speed in order to match the tape speed to the rate of the data coming from the server. Perry noted that FastSense optimizes the backup speed, throttling the Viper 200 to match the data speed. With FastSense, any of five speeds can be used, providing a match for a wide range of data rates.

Additionally, intelligent data compression will automatically detect data that is not compressible and turn off compression for such incompressible data files. This is a useful feature because when previously compressed files are recompressed, the result is often a larger file than the original compressed file. Writing such recompressed files takes up more media and may take longer to compress and write than would a compressed file that is not compressed a second or third time.

This feature won't be unique to Seagate. Hewlett-Packard will also be building this feature into its Ultrium drives.

The Viper 200 will have a serial connection that will provide support for remote diagnostics. The diagnostics are Java-based and platform independent. Additionally, the Viper 200 will have a feature called Dynamic Powerdown. This feature stores enough power in on-drive capacitors to smoothly power down the media in the event of a power failure. This feature enables both reels to be slowed simultaneously, maintaining tension on both spools and reducing the risk of damage to the tape or the cartridge. Seagate's Viper 200 will begin shipping in June. The current status of IBM's Ultrium will be explored in a future issue.

As the three Ultrium manufacturers continue their crawl towards final testing of pre-production models and eventual shipment of production units late second quarter or early third quarter, the ongoing progress will be reported in these pages. Conceivably, first units will be put into general use by June. At that point, Moore's Law will really kick in and the clock will start for second generation Ultrium drives, an effort that is already reportedly progressing well.