Kelman – new European pitch and Gulf Resources Canada DM deal (June 1997)

Kelman Technologies out of Calgary Alberta have just announced that they have signed with Gulf Resources Canada for the remastering and management of Gulf's legacy seismic dataset.

The contract involves the remastering of Gulf's 160,000 legacy tapes (corresponding to over a million km of seismics) to an STK/Magstar tape store running under IBM's Hierarchical File Management system, ADSM. Simultaneously, Kelman made their first European conference pitch at the Geneva EAGE demonstrating their data management technology. Kelman is first and foremost a seismic processing company, and as such knows a thing or two about managing seismic data. With their Kelman Archives division, they have packaged their remastering and data management know how into an archive solution with what they term the "Virtual Tape Device" as its centerpiece. Other solutions to the remastering of legacy data involve either the transcription of legacy formats to SEG-Y, or the encapsulation, using one of the emerging RODE encapsulation schemes, of the old format onto a modern high-density media. Kelman is circumspect about the merits of RODE, in view of its multiple implementations and complexity (see the editorial in this issue), and have come up with their own solution to the problem.

This involves the encapsulation of the legacy tape format onto a hierarchical storage management system such as IBM's Adstar Distributed Storage Manager (ADSM). Rather than re-formatting legacy data formats, the Kelman process captures everything that was on the original tape, including end of file marks, checksum information and tape status information. The resulting encapsulated data is stored on the ADSM on either tape or disk – which physical media used is actually irrelevant using this technology. When the data is required, decoding software returns the data to the calling application just as if the original tape drive was spinning the tape. Kelman claim to have "broken the boundary between tape and disk", and a large seismic dataset can be processed without a single tape actually being mounted.

Kelman's archival technology is wrapped into an online Data Management and Storage System (DMASS) with three main components -data management from the desktop, -on-line data retrieval and automated mirroring of the archive to a separate geographical location.

This last element should be of particular interest to companies operating in a parent company/subsidiary environment, where the security of a remote mirrored site could provide a spin-off in the form of a duplicate data set at head office, although I know of more than one local manager who would rather die than let this happen! Kelman are understandably reserved about the techniques they use for encapsulating and resituating original data, but they appreciate the importance of this technology and have not ruled out putting it into the public domain. This after all need not be a commercial loss to the developer as initiatives such as Netscape and Java have shown. But to convince the industry that their data is really safe, Kelman will have to show that their DMASS works on a wide range of hardware and software platforms, otherwise purchasers may find themselves locked in to elaborate combinations of ASDM and Magstar 3494 robotics. These do not come cheap, and there are other much more economical alternatives whose perceived reliability problems may become less important in the context of an automated mirroring environment.