Goldberg Annual Best Paper Awarded to...

More than 110 papers in computer science, electrical engineering and mathematical sciences published in refereed conference proceedings and journals in 2011 were submitted by IBM Research authors worldwide for the annual Pat Goldberg Memorial Best Papers in computer science, electrical engineering and mathematics.

One of the winners are a team from IBM's Zurich lab in the area of tape storage. We spoke with a few members of the team to find out more.

Magnetic Tape wit h29.5 billion bits per square inch
The machine IBM used to demo a new record in magnetic tape data density of 29.5 billion bits per square inch

Q. Explain your achievement in the 29.5-Gb/in^2 Recording Areal Density on Barium Ferrite Tape paper.

Giovanni Cherubini: The paper describes a single-channel demonstration that investigated the future potential of tape from a recording physics, and track-follow-servo perspective. This includes a variety of technologies that we developed to demonstrate the feasibility of operating at an areal density of 29.5 Gb/in^2 using prototype perpendicular Barium Ferrite (BaFe) media.

This density was more than a factor of 30 higher than the areal density of IBM's Linear Tape Open Generation 5 and the IBM 3592 (Jaguar 3) tape drives -- which were state of the art at the time of the demo. The work demonstrated the potential for the continued scaling of tape, based on low cost particulate media technology for at least the next 10 years. This gives organizations and enterprises confidence to continue investing in tape infrastructure to meet their growing archiving and data protection needs

Q. Why did you select Barium Ferrite for the demonstration?

Mark LantzThe state-of-the-art media technology at the time of the demo, which is based on metal particle (MP) technology, is running out of continued scaling potential. We've previously investigated a variety of new media technologies with the potential to replace MP.  However, BaFe was found to be the best candidate that is still based on low cost particulate coating technologies.

Q. What is next for this research?

Angeliki Pantazi: The single-channel demonstration described in our paper showed the potential for future scaling, but that did not take into account several of the challenges that arise from parallel channel tape recording, and from using production level hardware (rather than specialized experimental tape transport systems).

The next step we took was to transfer the technologies developed for the single channel into production-level hardware and perform a parallel in-drive cartridge demo -- which we completed at the beginning of this year. We are now planning to continue this work, and have started working on a new single channel demo with a target areal density of 100 Gb/in2.

4. Will we ever see Barium Ferrite tape products in the market and when?

ML: BaFe media was introduced with IBM’s latest enterprise tape drive: TS1140 (Jaguar 4), and will also be introduced for use in LTO6 (the next generation mid range product), along with MP media, but these are less advanced versions of BaFe media than was used in the demo.

5. Why do we still hear that tape is no longer viable, particularly in this era of Big Data?

GC: Some have claimed that tape has been dead for more than 20 years!

Tape suffers from being a back-end solution no longer visible to the average consumer, who likely has many disk drives in products ranging from laptops to digital video recorders. Unfortunately, this lack of visibility makes such claims about tape more believable.

Despite this lack of visibility, tape still plays an extremely important role in the storage hierarchy. In fact, total tape media capacity shipped each year exceeds the total capacity of external disk systems -- which highlights how much of enterprise data actually resides on tape. One of the goals of this work was to increase the visibility of tape and clearly demonstrate that not only is tape not dead, but that it has a bright future as a scalable, low cost technology for archiving and data protection.

Labels: , , , ,