Tape: the media is the message
In this roundtable discussion, representatives from leading tape media manufacturers tackle technology issues.
By Mark Ferelli
Saying that a lot depends on your recording media is a gross oversimplification. Media holds your data, and the reliability of the media can make a difference between data security and data loss.
Backup and restore are so vital that the tape industry continues to introduce new formats, technologies, and startup companies in rapid-fire succession. But discussion of drives and libraries is the rule, while discussion of media is the exception. This roundtable focuses on tape recording media.
We assembled representatives from five leading manufacturers to discuss some of the pertinent issues surrounding tape media. The roundtable was moderated by Mark Ferelli, InfoStor editor-at-large. Participating in the discussion were:
- Mike McCorkle, national technical support manager, Fujifilm
- Durkee Richards, magnetic recording scientist, Imation
- Richard Gadomski, national marketing manager, Maxell
- Larry Jones, vice president, quality management, Sony
- Mark Rogers, tape product marketing manager, Verbatim
FERELLI: Which recording methodology produces more tape wear: helical scan or serpentine recording? Why?
RICHARDS: Tape wear is a function of many factors, of which tape format (helical or serpentine) is just one. Others include rotating vs. stationary heads; tape path, speed and formulation; head formulation; design of the head-tape interface, etc. Wear can occur both on the recording surface, where the tape interacts with the head-tape interface, and also along the edges, where it interfaces with the tape path. There is currently insufficient data to state categorically that either helical scan or serpentine recording produces greater tape wear.
The head-tape interface of the linear Travan NS platform produces extremely low tape wear. In addition, its short, simple tape path makes it resistant to misalignment, which can produce excessive tape-edge forces and accelerated tape wear.
JONES: Different recording methods produce varying amounts of tape wear, but media manufacturers develop formulations that are optimized for helical and longitudinal recording.
McCORKLE: Materials are worn from the media surfaces in all tape recording technologies. Many factors, including tape path design, tape running tension and head design affect media wear for both technologies. However, media wear is more significant with helical-scan technologies. In helical-scan devices, the rotating read/write heads actually penetrate the tape media to create a stable head-to-tape interface. These heads impact the media and cross the width of the recording surface at an angle.
In addition to the head-to-tape interface, wear can occur at contact points in the drive`s tape path. Helical-scan devices place the data recording surface in contact with guides and a spinning head-drum surface. Many of these devices also employ a pinch-roller in contact with the data recording surface of the tape media.
In contrast, tape media used in linear serpentine technologies is virtually free from wear. In linear devices, the recording surface of the tape media comes into gentle contact with a stationary read/write head. Wear and tear on the tape is less than for a drive with a more complicated tape path and rotating heads.
GADOMSKI: In general, helical scan recording produces more tape wear because it has a very sophisticated system of guides and rollers that allow for precise data recording on a relatively small tape area. The benefit of this method is a physically smaller tape cartridge, thus reducing cost and storage space.
ROGERS: Of the many tape recording methodologies, linear recording continues to offer--for many users--the best balance of capacity, reliability, performance, and cost. The most compelling advantage for linear recording is its simplicity. Linear recording technology has been used in many tape products, such as open reel, 3480, and QIC cassettes and cartridges. All of these products are still in use today, proving the long-term reliability of their basic design.
In linear recording, the tape path is generally in the plane of the tape, with a minimum of guide rollers, minimizing tape wear and debris. In addition, most stress components on the tape are in the direction of tape motion, vs. perpendicular to the direction of tape motion, minimizing tape edge damage and debris generation. With higher and higher capacity media solutions using thinner and thinner media substrates, minimizing the complexity of the tape path has obvious advantages from a mechanical point of view.
Linear recording generally puts less mechanical stress on the tape media, as opposed to helical methods that involve an aggressive physical environment (drum rotating against tensioned tape at several thousand rpm) and high media wrap angles (large contact patch of head to media). In fact, most helical scan cartridges are rated to 2,000-pass lifetime, vs. 5,000 to 15,000 pass lifetimes for QIC-type technology.
FERELLI: In spite of MTBF figures, how long can you really go before having to refresh a tape?
RICHARDS: MTBF figures are generally applied to the drive mechanisms. EOL (end of life) for a tape cartridge is most often measured in number of passes before a tape cartridge is replaced.
Usage patterns have a tremendous influence on tape life. Tapes that are subject to short, frequent read/write passes experience greater wear and more uneven wear across the media than a tape that undergoes full end-to-end read/write passes on a less frequent basis. Factors such as how often the drive is cleaned also have an influence. Is the tape being used in a challenging physical environment, i.e., extremes in temperature, humidity, airborne particulates, etc.? Users may also use different measures than total number of passes for EOL, such as number of mounts, total data written to or read from, error rate increases, or other factors, depending on their particular needs and usage patterns.
McCORKLE: Refreshing refers to re-writing to restore magnetic signals (data) to full strength. When MP tapes are stored under good conditions, refreshing isn`t needed. Even after 30 years, signal loss is minimal. Tapes based on ATOMM technology have longer life than traditional thicker-coated MP media. For example, the archival specification for DLTtape IV is over 30 years with less than 5% demagnetization loss. This can be compared to a thicker coating product with less than 7% loss after 30 years. Both 5% and 7% are insignificant in terms of magnetization loss, and playback would not be affected.
Users could plan to rewrite archived data to the same or new media at some point after 30 years. Refreshing the magnetism is prudent; however, it shouldn`t be necessary for MP media. Conversion to a new storage format will occur long before refreshing the data is necessary.
FERELLI: Even with decades of archival life, how about something like re-tensioning?
GADOMSKI: When operated and stored within the environmental specifications of the hardware and media manufacturers, metal tape has an archival life of 20 to 30 years. To maximize this longevity, a common practice is to re-tension the tape every 12 months or so depending on technology and environmental factors.
ROGERS: It used to be a common practice to refresh reel tape because the magnetically encoded info would "print through" to adjacent wraps of tape. With today`s coatings, it`s considered a non-issue, so information can basically be stored indefinitely on magnetic tape.
JONES: The archival life is highly dependent on the format. Variables such as coating type, coercivity, recording density, storage, and operating environments greatly affect media life. As such, the various recording standards must be consulted to determine the useful operating life.
FERELLI: Which has more reliability: single- or dual-capstan tape. Why?
GADOMSKI: Dual capstan tape systems seem to be more reliable because issues of belt slippage are eliminated. However, single capstan technologies have been around for a long time and continue to thrive today. The single capstan system reduces cost.
McCORKLE: From a media viewpoint, there`s not a big difference between single and dual reel products. From a drive viewpoint, we believe a single reel type is easier for designing tape path and running tension. With a single reel type, the second reel is located in the drive, so the tape path is shorter and simpler. Dual reel drives require fine-tuning of tape path for stable tape running. So a single reel drive system is more reliable than a dual reel system from a tape running viewpoint.
JONES: Both are reliable since each is designed for a specific application. For instance, dual hub cartridges are well suited for high-speed access and typically use shorter lengths of tape. Shorter tape balances the greater stress of high rotational speeds, whereas single hub cartridges contain longer lengths of tape, but fewer moving parts.
RICHARDS: With tape drives that have two drive motors and capstans, as opposed to one, reliability will be reduced for one simple reason--more moving parts means more potential failure points. Dual motor and dual capstan systems that are engineered to a high degree of reliability will tend to be more costly than single motor and single capstan systems of equal reliability.
FERELLI: Is the addition of intelligence to the tape cassette worth the investment?
McCORKLE: Yes, RAM chips in dual-axis data cartridges will enable a greatly expanded role for tape. However, new applications will require coordinated development in operating system software, tape storage management software, and multi-sector tape media formats to tightly integrate the potential enabled by data cartridges with RAM.
GADOMSKI: The addition of MIC [Memory In Cassette] can be very useful for applications where rapid file access is required. MIC allows file location data to be read by the drive without actually mounting the tape.
RICHARDS: Improving the efficiency of data management, particularly in large data centers, is one of our major goals. However, it must be done in a cost-effective manner to produce adequate ROI for the associated cost delta. Currently, the cost of MIC is quite high relative to the amount of data that can be stored. As such, at the current time we do not believe the technology justifies the costs.
JONES: Data storage on magnetic tape has grown exponentially. We see new applications such as multimedia files like MPEG movie data and MP3 audio that will even further increase the amount of information contained on data media. With the addition of memory in cassette, one can instantly scan the cassette`s memory and determine the content of the cassette and other important usage information. The benefit of memory in cassette is especially evident in library applications where the cassette can be scanned before it is "picked" for loading.
RICHARDS: It does depend on the application, but quicker seeks to data is what it buys you. So the answer depends on your application. If you`re storing data online in an interactive application, you`ll want faster seeks. If you just backup and archive, or store primarily very large files, seek time is not a primary value.
FERELLI: Are there new media formulations in the offing? Will metal particulate ever yield to cobalt chrome or another formula?
McCORKLE: ATOMM uses dual-coated metal particle media. We continue to support R&D efforts to formulate even thinner coated metal particle media. This thinner magnetic layer media will be able to support the next generation of high-capacity magnetic media disk and tape products currently being developed.
GADOMSKI: Technological advancements in magnetic recording mediums continue. However, the next several years will probably be focused on advanced metal evaporated particles because the potential for this technology has yet to be fully exhausted.
JONES: Metal (cobalt) evaporated media is a good example of media that is on the cutting edge of magnetic recording. The performance of the latest generation of advanced metal evaporated media already exceeds that of metal particulate media.
RICHARDS: Particularly for linear tape systems, there are still considerable improvements in capacity and performance to be leveraged from existing metal particle (MP) formulations. For example, linear platforms such as Travan NS20, 9840, 3590, and Magstar MP still utilize first-generation metal particle formulations, such as MP+, which have a proven track record of stability, reliability and durability. In addition, MP+ is made from readily available materials, making it highly cost-effective. Advanced metal particle formulations, such as MP++ and MP+++, hold enormous potential for further improvements in areal density, tape capacity and performance.
Clearly, metal evaporated technologies are being used in select applications, but their broad applicability in different tape paths is questionable. Coatings similar to those used on rigid disks (for example, cobalt-chrome) might offer enhanced performance at very high recording densities, but cost issues would be very serious barriers.
ROGERS: I would expect current technology to give way to superior formulation technology as it becomes cost effective to implement it. Each basic media formulation has its strengths and weaknesses, so as advances are made with formulations, expect both incremental and breakthrough changes down the road.
Total revenue for all tape media types is expected to top $1 billion before leveling off.
Q&A with David Bunzell, Santa Clara Consulting Group
Q: Is media keeping up with drive technologies? Which is the tail and which is the dog?
A: I would say that drive technology is lagging behind media potential and capability. There has been a lot of development in substrates and coating technologies that provide sufficient "real estate" to increase capacities, but they have to have the hardware, drives, and heads enabled to support the additional real estate.
Q: There`s a real variety in tape media out there. What`s the bestseller and what`s lagging behind?
A: The bestseller in 1998 revenues was DLT media. It represented more than half the revenues of the tape products that we cover. There are transitions going on. All the tape formats except DLT are facing declines of various degrees. It`s partly because, as things move up, other technologies have not been able to have products available for the market that meet some of the higher capacity storage requirements that DLT has. They are catching up, but it`s a game of catch-up. 8mm and 4mm vendors have products coming in that reach certain parts of the market. But if you`re looking at revenue, it`s difficult to compete with the momentum that DLT has.
Q: How do you view some of the market dynamics?
A: As I mentioned, DLT has momentum. You also have 8mm alternatives competing at similar capacity points--Mammoth and AIT. LTO, which will not be available until next year, has significant players behind it--IBM, HP, and Seagate. LTO has potential if they release it on schedule. 4mm continues up in capacity, but will not necessarily be reaching the capacity points that some of the other technologies have, but it definitely has a position that can be maintained.
Q: How about challengers like Tandberg Data`s SLR?
A: They`ve attracted interest from users who are familiar and comfortable with 5.25-inch QIC technology, and they`ve been able to upgrade some of their customer base. But they have also hedged their bets by licensing DLT from Quantum. What isn`t clear is: Will they put more support behind their home-grown technology, or provide equal or greater support for DLT?