Two Decades Of Tape.

AN INTERVIEW WITH FREEMAN REPORTS' ROBERT ABRAHAM 

The year 2000 marks the beginning of the third decade for WestWorld Productions, Inc. and the editors felt that this provided a good reason (some may call it an excuse) for taking a look backward to major storage milestones of the past twenty years and a peek ahead at where it appears to be going. In this article, executive editor Mark Brownstein spoke with Bob Abraham, president of Freeman Reports, who began looking at tape even before WestWorld started publishing.

MARK: Let's start at the beginning. Twenty years ago, when you said tape, you were really talking about 9 track. Right?

BOB: Right. At that time, the rage was 6250bpi, otherwise known as GCR (Group Code Recording). You can put an entire 180MB of information on one of the 10.5-inch reels, unformatted. When formatted, [the capacity] typically got down to 120 or 140MB.

There was almost no other tape on the market. There was definitely no other mainstream tape on the market.

There were some other interesting things from CDC (Control Data Corporation) and maybe Fujitsu. The point I'm making is that, when you thought of tape, it was just like in the movies--you saw a reel of tape in the background, rotating, and saw two reels. The drives were very expensive, typically tens of thousands of dollars, but the media was fairly cheap, $10-20 or so, in that kind of range, although this seemed expensive then.

In that timeframe, tape was already more than a quarter century old. If we go back another 26 years, in 1953 the first commercially available computer tape product shipped from IBM and it was also a half-inch reel product. In essence, this was the only viable tape technology from 1953, when tape was introduced, to 1979 and it remained the only viable product for another five years.

In 1984, there was a milestone: IBM introduced half-inch cartridge technology with the specific intent to replace the half-inch reel. They were very successful in converting their customers quickly, but non-IBM customers didn't go along quite so quickly.

[Today] half-inch reel products are still shipping. The first, original, and oldest technology is still shipping today. Fourteen companies are shipping and making money, producing half-inch reel technology products. These are all for legacy applications.

The half-inch cartridge was introduced around 1984. IBM converted its own customer base, but it took approximately a decade to do.

The 1980s were a very busy decade for tape. Back around 1979, there were no other real, viable tape technologies, but we were right on the threshold of some pretty exciting stuff. Around 1981 or so was the quarter inch tape introduction. It took off fairly slowly, but not terribly slowly, considering what it was. It was a breakthrough in size.

Compared to 8-inch form factors [of half-inch products], the compactness was good and the cost was much lower than big, bulky computer grade tape. [Quarter-Inch tape] started serving a whole series of new needs. These were not yet for PCs, but were for smaller [than mainframe] systems. As quarter-inch technology developed, it splintered and went many different directions in terms of format and size, in spite of 3M's efforts to hold it together.

MARK: I remember a lot of quarter-inch stuff out there and none of it seemed compatible with anything else. It seemed to be pretty confusing and something of a gamble to choose one over the other. Do you remember it that way?

BOB: A number of companies used proprietary formats and some were very successful, especially Irwin. It was QIC [the Quarter-Inch Committee] that pulled this together and got it wrestled under control. The QIC organization was active from that point until literally a year or two ago.

MARK: What caused the QIC committee to shut down? Aren't they the same ones doing Travan?

BOB: Travan was something of a successor to QIC, although QIC embraced both the mini cartridge and the full-sized cartridge. It took a fair amount of time for the market to form for that size cartridge. In the early '90s, mini-cartridges really caught on. Again, there was a lot of fragmentation, proprietary interfaces, and a lot of infighting in the industry. There was a lot of industry consolidation--at this time, what's left of the quarter inch industry is two or three companies with full size data cartridges, primarily Tandberg, and a number of companies with mini-cartridges: HP, Seagate, Tecmar, and one or two others.

The mini-cartridge aspect has revolved around the Travan technology, which is sort of a spin-off of QIC. Travan belongs to Imation--they're really the driving force here.

[Tape technologies undergo an] up and down cycle. [For QIC] it's not quite down yet, but appears to be going down cycle, for quarter inch has spanned these last two decades.

MARK: What about other formats, like 8mm?

BOB: In 8mm, Exabyte has really owned that technology. From the Exabyte standpoint, it has been an up and down cycle, and 8mm is now barely one decade old.

There's sort of a subset of that, developed by [Exabyte's] co-developer [Sony]. Sony AIT is on an uptick. AIT was introduced four years ago and, in terms of mind share, AIT is on a par with Exabyte for 8mm. In terms of a market position, [AIT] may not be quite there yet, but momentum is favoring AIT. I do not discount Exabyte. With Mammoth2 very close [to release], assuming they can execute and pull it off smoothly, they'll be back in the ballgame.

MARK: And DLT?

BOB: In 1984, DEC (Digital Equipment Corporation) was anticipating that IBM was going to release a half-inch cartridge product, wanted to pre-empt IBM, and announced Compact Tape. This was a nifty little 5.25-inch [cartridge] but a little oversized. It was used in the (DEC) VAX machines and was very successful in the digital environment, but didn't get out of the house. It was designed for in-house, captive use anyway. It evolved into a family of products.

About the time [DEC's Compact Tape was] getting successful, they sold it to Quantum. [At the time] Quantum didn't know what they were getting. It turned out that they got the crown jewel of the technology. At the time, the capacity was 10GB up from 90MB in 1984. [Compact tape had] gone up to 10GB. [Quantum named the technology Digital Linear Tape (DLT)] and the rest is history now.

[DLT has been] the most successful tape technology ever, measured in the short term, in terms of market acceptance, mindshare, and revenue share. It has done phenomenally in the last four or five years.

MARK: Let's not forget 4mm.

BOB: DDS and DAT were introduced in 1988. HP and Sony were its champions. HP has dominated that market every year and certainly for most all of the recent years. The first [DAT] drives shipped in 1988 and HP is still producing them in the same format.

There was a format war in the early '90s and, for two or three or four years, this was a big war. Ray Freeman [of Freeman Associates] was a facilitator for the DDS format. This was not a trade Association, but was a manufacturing group. It was a facilitator of the DDS format. The DDS format is the [DAT] format winner. The data DAT format disappeared--it never got off the ground, never achieved any critical market mass.

Sony, I think, is the most prolific tape company and probably even in optical. In terms of available technologies, they're involved in so many drive technologies, media technologies, and sub-technologies--raw materials and component stuff. Sony is really the most prolific of all of the tape industry players, but they don't have the market share except in AIT. They don't really shine; they have been a disappointment for DAT. In .25- inch, they've done media, but not drives.

DAT is a story all by itself. It has peaked and is in the process of peaking in terms of market share.

When we look at what's going on in tape technology, many--many we look at are at their peak now--or maybe a year or two away, technologies like AIT and some of the very new technologies like AIT, if you talk about AIT being a subset of 8mm, collectively it is a slow growth and low growth situation. Some would conclude that there's no action here.

If you look at the long-term picture, if you consider that it took forty-some years to get to this point, tape won't disappear. Tape has a huge amount of momentum and inertia and it won't disappear quickly.

[With regards to the] original DAT format, DDS[1] is still shipping today. It's in the eleventh year of shipment and I predict that it will ship another year at least. That's 12 years for one format. It took seven years to get to its peak and another five years to get to the bottom. DDS2 came on the scene and has peaked. DDS3 was subsequently introduced and is probably at its peak point now. DDS4 has been introduced over the last few months and is just beginning to climb.

DDS5 should be out within the next two years or so. We see no signs of weakness on HP's part. They claim they are working on lab technology specimens. I believe that DDS5 will happen. If we put this all together, it means that DDS will have its own 20-year cycle, if not longer. That's in light of all the others that will come after it--VXA [from Ecrix], OnStream, and others.

Why would anybody want to be in this business? If they manage it well, there is a lot of legacy business there, although you can be outflanked by competitors, response times, and time to market are not as critical as [they are] in the magnetic disk industry. If you miss a market window by three or four months [in the hard disk industry], what's the point of getting into that market, because it's already approaching the peak?

MARK: What about virtual tape? Where is it going?

BOB: There are a lot of undertows with esoteric things like virtual tape, the impact of the SAN, and concepts like tape sharing. All these things have features in common--all are shaping or influencing the direction of tape. Tape is moving away from being a component to being a solution all by itself.

When you look at a SAN, you tend to look at a large storage network, a network of storage devices, tape, or tape libraries. But the SAN really needs these other things I've identified in order to play effectively for the next set of applications here.

For the enterprise, this is where tape is moving: it needs virtual tape in order to be cost efficient. It needs tape sharing in order to function well in a system, otherwise multiple users can't get individual tape devices, whether it's a library or a tape drive that's on the SAN. It needs Fibre Channel or some form of superconnectivity--it NEEDS that kind of performance level and feature rich things that go along with it in order to operate effectively. These will all be working together, buoying all the technologies, in order to be an important part of storage. The confluence of all these trends will enable tape to move into a new kind of a role in storage and this is a very powerful concept.

MARK: What about RAIT (Redundant Arrays of Independent Tape)?

BOB: There's ultimately a place for it and clearly a place where RAID is a very good solution. Supercomputer applications require screaming performance speeds. You can't get it with any standalone applications just put together in a RAIT array, just as you can't get it with just a hard drive. On an array, you can let them zing and they do. [RAIT] is a small, limited market, not a niche market because it crosses a lot of disciplines into a lot of vertical markets. Ultimately, I don't think there's a big market there because tape is a limiting factor in most systems.

When you look at SuperDLTs, the LTOs--Ultrium and Accelis with 10 or more MB/sec performance for small low cost products, you're looking at very capable technology at a very affordable price. In a library environment, as opposed to RAIT, cascading multiple drives for backup, you can run multiple drives together, not striping, run concurrently, you can get very good transfer rates. The small market for RAIT will grow at its own pace. I'm not bullish about it, but it's not at end of its life, either. It will always be there at some level of market acceptance.