Mailing List CNI-ANNOUNCE@cni.org Message #113427
From: Clifford Lynch <cliff@cni.org>
Sender: <cgplmgr@cni.org>
Subject: Materials from Storage Architectures for DIgital Preservation, Sept 22-23, 2009
Date: Wed, 07 Oct 2009 20:06:29 -0400
To: <CNI-ANNOUNCE>
Materials from Storage Architectures for DIgital Preservat
On September 22-23, 2009 I was fortunate to be able to attend a Workshop on Storage Architectures for Digital Preservation organized by the Library of Congress NDIIPP program. This workshop was one of a series that LC has convened in recent years to bring together people from the Federal Government, storage technology providers, NDIIPP partners and other specialists in digital preservation to focus on the interplay between the development of commercial large scale storage systems and the evolution of digital preservation systems. The presentations, along with a meeting summary prepared by the Library of Congress staff, are now available at:

http://www.digitalpreservation.gov/news/events/other_meetings/storage09/index.html

I found many of these talks very valuable; the set of presentations on Data Integrity were particularly helpful. Each segment of the workshop included extensive discussion, and one point mentioned only in passing in the LC notes that I want to highlight for the CNI community was a developing initiative within the US Federal Government and the High Performance Computing community addressing Resilient Computing -- the design of systems that continue to function in useful ways even in the face of extensive component failures. Thusfar, much of the thinking in this area has focused on computational systems rather than storage systems (see, for example, the frightening report "Towards Exascale Resilience" at http://jointlab.ncsa.illinois.edu/pubs/Toward_Exascale_Resilience.pdf and additional materials hosted at http://institutes.lanl.gov/resilience/). I believe that these ideas are going to be very important in future thinking about how to design digital preservation systems that minimize and constrain loss, rather than pursuing perfectly lossless systems, which are likely to be both technologically and economically unachievable at very large scale, as some of the presentations at the LC Symposium suggest.  I've expanded a little bit on these ideas in the October 6, 2009 CNI Conversation (the audio file of this session will be available very shortly at http://www.cni.org/cni_conversations/).

My thanks to the LIbrary of Congess for convening these meetings; they are a great service to the community.


Clifford Lynch
Director, CNI

Subscribe (FEED) Subscribe (DIGEST) Subscribe (INDEX) Unsubscribe Mail to Listmaster