With revenues estimated at $6bn in 1997, IBM’s Storage Systems Division (SSD) would be a giant in the computing landscape even if it was not part of the world’s largest computer company. But big has not always meant successful, and many of the storage unit’s grandest schemes have gone awry lately. Its next- generation storage technology, Serial Storage Architecture (SSA), was originally intended to replace SCSI as the de facto standard for attaching storage subsystems to servers, but the design has failed to win industry acceptance and has been sidelined by the fibre channel, an alternative technology. That comes after the company stumbled in its initial efforts to bring a RAID disk array line to market and had to turn to rival Storagetek for design and manufacturing help. Now IBM’s plans to introduce an intelligent storage server, the centerpiece of its universal data storage strategy, have fallen behind schedule by as much as a year. Unfortunately, says Mike Casey, an analyst with the Gartner Group, SSD’s product missteps bring to mind the IBM PCdivision’s less than successful efforts with its Microchannel architecture and OS/2 operating system, earlier instances of over-engineered IBM solutions. In some areas it’s not clear what the strategy is and in other cases it hasn’t executed on time, which gives the other vendors time to go out and bring stuff to market. At least SSD can boast one success story. Its disk drive component design and manufacturing business is winning customers among leading drive vendors especially for its innovative magneto-resistive disk drive heads, which increase the potential storage density on a hard disk. That has not come without huge investment. IBM has plowed $680m into manufacturing plants for the heads since 1996, but in February the company celebrated the shipment of the 10 millionth unit, only six months after full-scale production began. According to estimates by industry analyst Annex Research, component sales to other manufacturers contributed $1.9bn, of SSD’s total revenues in 1997, but it is the subsystem business that is the core activity, representing $3.4bn. Despite the failure to sell the concept of SSA to other vendors, IBM has been successful in selling SSA-based subsystems into its own customers’ datacenters, aided by SSA’s current price advantage over fibre channel. According to Gartner’s Casey, it is also selling 15% of its SSA shipments for attachment to other manufacturers’ systems, something of a feat given customer loyalty to the incumbent systems vendor. But IBM SSD is somewhat crestfallen, having had to accept the victory of fibre channel in the wider market, a victory which will ultimately consign SSA to a role within IBM’s own subsystems. Another source of dissatisfaction for engineering staff was the 1996 decision to hand over design and manufacture of RAID systems, the Ramac Virtual Array subsystems, to former rival Storagetek.
Seascape and Tarpon
The IBM team’s morale is set for a boost with the launch of a new intelligent storage server, reputedly code named Tarpon, which was originally due to ship at the end of last year. But by the end of the first quarter there was still no sign of Tarpon. Rather, IBM has most recently been unveiling assorted additions to its universal enterprise storage architecture, Seascape. Seascape is a framework and a set of building blocks for building flexible storage solutions, that promises to lay the foundation for enabling universal data access. A February announcement included the Cross Platform Extension, which allows Unix and Windows NT servers to share Ramac arrays, and a number of enhancements both to IBM’s SnapShot data replication software and to its flagship storage management product, Adstar Distributed Storage Manager. But until IBM can deliver Tarpon, the attempt to make a coherent whole out of Seascape’s many fragmented parts is missing its essential, unifying centerpiece. Tarpon, which is based on IBM RS/6000 Unix server technology, provides the high- powered intelligence that is a vital ingredient in any universal storage system. As the number two vendor in both the enterprise and Unix storage markets, IBM has a great deal of share to defend – 37.5% of the mainframe disk storage market and 15% of the Unix market, according to IDC’s figures for 1997. Jim Vanderslice, SSD chief, made components his first priority on taking charge of the division. Now that the components business is ticking over nicely, he has shifted his focus to subsystems and software. Yet despite a $2.4bn investtment in SSD over the last 18 months, outsiders still ask whether IBM has left it too late to claw back much of its lost share from competitors such as EMC, Hitachi Data Systems and Sun Microsystems, all of whom are targeting its mainframe and Unix accounts with their own multiplatform offerings. By the time IBM has assembled all the ingredients of its Seascape product portfolio, it may find more customers have already set sail for other shores.
Computer Business Review.
