Future of seismic data storage

Eurotech, ‘large companies have no long term storage strategy.’ Iron Mountain, ‘not all are ready to face storage challenges.’ Statoil as remastering poster child. IBM ‘cloudifies’ tape.

At a seminar on the future of seismic storage organized by the Norwegian Petroleum Directorate earlier this year in Stavanger, operators were warned that inaction is a high-risk option with regard to preserving legacy seismic data on tape. Egil Simones has worked with seismic related data since 1992 starting as tape monkey, then field engineer and latterly, as CEO of Eurotech Computer Services Norway, one of a few companies that still specialize in tape technology. Simones has experience with tape from many large companies and has ‘seen all the problems there are.’ Many large companies have no long term migration strategy. Only when somebody died does stuff get thrown away. Equipment and tapes are assumed to last perhaps thirty years. Information and systems are managed and planned for based on today’s viewpoint and values, not on what the future will bring. It now hard to access the equipment needed to read older tapes and it is not going to get any easier (or cheaper) in the future. Companies should have a data migration strategy. If this is done right, ‘you can actually get rid of the old tapes, not just add another set to manage!’

John Kjetil Pedersen (Iron Mountain) traced the history of media and technology for seismic and well data recording, from 7, 21 and 9 track tapes to modern 3592 media. There is now a general recognition that media and data formats change over time and that data deteriorates due to aging and storage conditions. The seismic industry is not alone here as film, broadcast and others face the same challenges. However, most data owners have no clear media strategy. Old media and formats are kept too long and meanwhile it is getting harder and more expensive to recover old data as hardware expires and operators retire. Unfortunately, ‘not everyone is willing to face these challenges properly.’

Judging by Sivert Kibsgaard’s presentation, Statoil should be a poster child for an orderly seismic remastering program. This has run in four phases from 2004 through to 2017 and has seen over half a million tapes upgraded to modern 3590/3592. Looking to the future, Kibsgaard weighed up the pros and cons of LTFS (see below). The latest media offers easy access to content. Even though it is tape, it is more like a USB stick or disk and there are no issues regarding block size and tape handling commands. On the other hand, its relationship with SEG-D is uncertain and its use may not be standard across vendors.

IBM’s Robert Haas described how IBM is ‘cloudifying’ tape storage with storage objects and by extending OpenStack Swift to high latency media (i.e. tape). IBM provides OpenLTFS as entry point to its Spectrum LTFS tape libraries. If there remains any doubt as to tape’s continuing importance in the modern IT/big data world, Haas pointed out that Google and others in the ‘Intel Super 7*’ offer various combinations of LTO and Jag tape drives in the cloud. Haas also provided a pointer to an old but interesting presentation by Google’s Raymond Blum on ‘How Google backs up the internet.’ Blum observed that a) backups are useless (in themselves), what is important is the restore and b) internet/Google scale mandates taking humans out of the loop and c) diversity is key, tape is great because it is not disk! More next month in our report from the SEG Technical Standards committee.

* The Intel Super 7 are the GAFA plus others which influence chip development.

Click here to comment on this article

Click here to view this article in context on a desktop

© Oil IT Journal - all rights reserved.