Autoloaders help overcome DDS capacity shortcomings: now, even price-sensitive businesses are looking at automation. (DDS Replacement).

DDS--now in its fourth generation--has served ever since 1989 as the backbone of DAS (direct-attached storage) technology. It's still a viable technology with healthy sales--according to Gartner Dataquest, DDS shipments totaled more than 1.2 million units in 2001. However, DDS is at the end of its development cycle and its former champions have turned to other tape technologies with more capacity and higher performance. Successive generations of DDS have taken longer to develop and delivered smaller gains in performance. DDS-3, released in 1995, was the third generation drive. It delivered three times the capacity of DDS-2, and it took two years to develop. DDS-4, however, was released in 1999, four years after DDS-3, and it did not even double capacity. Its specifications are still adequate for the SOHO (small office/home office) world, but mid-tier companies and enterprises found that its DDS drives lacked the capacity and performance their business increasingly needed. Steve Whitner, director of marketing a t ADIC, said, "We saw five years ago that DDS was penetrating the low-end enterprise environment, and going beyond what it could really do effectively. The 4mm ended up getting squeezed out in a low-end enterprise by DLT." Capacity and performance were issues, and so was the robustness of DDS automation--or the lack thereof. Kelly Beavers, VP of marketing at Exabyte, said, "When you make a 3.5-inch design, you sacrifice a good deal of robustness. DDS products really couldn't hold up under that kind of usage. Failure rates became higher and higher."

By 2000, it became clear that there was never going to be a DDS-5. The announcement was a non-event: companies simply shifted their development dollars to other tape technologies. They hoped that these higher capacity, higher performance tapes would capture the OEM and entry level/mid-tier markets. Another pressure on DDS development is that the backup market seemed to be turning away from the DDS automation of choice-autoloaders--and moving towards libraries and distributed backup products. Due to its strong OEM base and low price point, DDS is still shipping strongly to its large user base of 9 million. But the technology has reached the end of its lifecycle, and DDS is making a slow march towards the sunset.

However, it's hardly out the door yet. DDS is still the top choice for server-attached storage devices, and server OEMs and manufacturers often include them as backup devices with their servers. DDS remains at a low price point largely due to its giant economies of scale. Also, unlike other tape technologies where the body of the cartridge is precision made to accurately align the tape with the heads, most of the precision alignment is done by the mechanism itself. More expensive tape technologies such as DLT, AIT and LTO have sought to gain server market share, but price point and drive height are serious issues here. Putting in more expensive tape drives would up server prices in the price-sensitive small business market, and might require factory retooling. Drive width in this space is not a particular problem. Although DDS has a smaller width than LTO, DLT and VXA (3.5-inch vs. 5.25-inch), most servers are manufactured with a 5.25-inch slot that can easily accept 3.5-inch. But height is an issue, and LTO and DLT are twice as tall as VXA and Benchmark's DLT1. Thus far, VXA has won out over Benchmark in the DDS replacement space because of a lower price point. (Beavers believes a successful DDS replacement drive must offer improved capacity and performance over DDS-4 at a drive price of under $1,000.) VXA has landed OEM agreements with Compaq, Fujitsu, and Apple, and has recently added IBM's P-series server.

DDS's capacity and performance lag behind newer tape entries--DDS-4 delivers 20GB native capacity with a native transfer rate of up to 3MB/second--but the technology is still sound. Digital Data Storage (DDS) is a helical scan format where data is written diagonally across the width of the tape. This allows for a capacity advantage by allowing subsequent tracks to have different writing angles and to overlap. No guard bands are necessary. The data format also allows for fast location of tracks and data. With a native transfer rate of 3MB/sec, a DDS-4 drive with DDS-4 media could still back up a typical mid-range server in 74 minutes (assuming 1.5 compression ratio), For many smaller companies this is perfectly acceptable and will be for some time to come. DDS autoloaders, which load multiple tapes in succession help close capacity gap though not the performance issue. However, companies are no longer developing DDS and its drives are not particularly robust in automated environments. What's a DDS user to do?

DDS and Automation

Autoloaders can help overcome DDS's capacity shortcomings. But the drives do not last as long in automated devices as they do standalone, and the DDS autoloaders do not display the same price/performance sensitivity, and are considerably more expensive than the drives. There are more than 300,000 DDS autoloaders in North America and Europe, operating at a cost of $26 per GB of native capacity. This is not cheap. Jim Watson, president and CEO of Peak Storage Solutions, believes that when considering replacement automation approaches, companies should look at volume efficiencies of read/write technologies. Helical scan and linear are two popular approaches to storing data on magnetic tape. (DDS is a helical scan technology.) Helical scan enables the tape drive to stripe data diagonally across the tape and read it in one pass. DDS, VXA, Mammoth and AIT are helical scan technologies. Linear tape technology writes data on tracks and passes the tape back and forth across the heads. DLT and LTO are linear. And in ta pe, size does matter. Mammoth, AIT and VXA are 8mm, DDS is 4mm, and LTO and DLT are half-inch. According to Watson, helical scan drives take up less space in an automation device, which means that mean time to data is faster with a helical scan than with a linear tape solution. Linear half-inch drives do have many advantages for larger systems and are the usual drive of choice in larger libraries, but for more economical autoloaders, helical scan, 8mm media can best fit the bill. Another impact on price is that the linear motors are larger than that of helical scan, making power issues more costly. Helical scan drives use a smaller motor with its tape heads doing most of the movement.

Autoloaders to Libraries

Bob Covey, VP of marketing at Qualstar, noted that many autoloader installations are changing over to libraries. As the amount of data grows ever larger, adding tape drives and sequential autoloaders to handle the backup merely makes the process longer. Capacity becomes less important than a different technical approach to recording data. Multi-drive automated libraries help solve this problem by allowing random access to multiple tape cartridges and using parallel drives. Covey defines the difference between automated tape libraries and autoloaders:

* Autoloaders have a single tape drive, and must load and unload individual tapes in a sequential operation.

* Automated tape libraries contain multiple tape drives, which 'can concurrently read and write to tapes. Adding additional drives improves throughput, and libraries exchange tapes faster than autoloaders.

Bob Rogers, BMC's chief storage technologist, agrees that many environments are moving away from autoloaders. and towards networked libraries as part of a distributed backup strategy. He points out that a small business might use an autoloader if they're just backing up the. server in the office, but more companies are building storage area networks and installing libraries. This is a movement away from the direct-attached model, and allows IT to protect and centralize data throughout the enterprise.

Rogers said, "A direct attached autoloader in some situations makes sense. But it doesn't make sense in that it localizes your backup solution. Many customers could take advantage of a SAN attachment, locate their library a moderate distance away, not necessarily adjacent to the server. If there were a catastrophic event, then the probability of being able to restore the data would be much higher."

Not everyone agrees that autoloaders will--or should--give way to networked libraries. Exabyte's Beavers pointed out that large segments of the business market do not have the resources to sink into expensive options such as distributed backup and storage networking. "When DDS came out, the users always had a choice of spending more money to buy a drive with higher capacity and greater data rates. They've always had the choice. The market has always existed where they spend Less money to get a drive. If you charge too much, they don't even think of you as a solution."

However, even price-sensitive businesses are looking more at automation products. Many confirmed DDS users never adopted the expensive autoloaders, but that may change if automation price points come in line with less expensive DDS replacement drives. The value of automation is just as important to smaller business as it is to larger companies, since automation replaces manual tape swapping. The enterprise may be a magnitude larger, but the relative value remains the same.

The success of DDS replacement technologies will depend at healthy capacity and performance, low price points for drives, media and automation, and sensitivity to manufacturing processes. With this type of tape technology, price-sensitive companies can keep their backups running and their businesses sound.

www.adic.com

www.exabyte.com

www.peakstor.com

www.qualstar.com

Autoloader Manufacturers

DAT      8mm            DLT            LTO
HP       Exabyte        ADIC           ADIC
Seagate  Maxoptix/Peak  Benchmark      HP
Sony     Sony           Exabyte        Exabyte
                        HP             Overland
                        Overland Data  Seagate
                        Quantum