Blog Post: Hardware Abstraction as a Preservation Methodology


Michael Nolan, Intel

"An on-going curation function will be needed to ensure that the re-deployment process always has a container available to it that can host our preserved environments and execute them on current generation hardware."




The digital universe continues to expand rapidly. But what is driving this? As we approach the middle of 2013, the percentage of people worldwide who use the internet stands at 34% according to with emerging markets seeing the largest growth in new users. But that is just part of the story. The ways in which we interact with the internet are changing too. New devices compete for their place in our lives. Just think of how much we love our smartphones, tablets and e-readers. The appeal of this new generation ‘hedonistic’, rather than utilitarian compute devices, has brought technology into the lives of more people than ever before. This evolving relationship with internet enabled devices of all kinds has seen demand for richer content grow rapidly. Our reliance on digital information is therefore also growing and the need to access our data in decades to come is slowly dawning on larger numbers of individuals and organisations.

TIMBUS is an FP7 research project funded by the EU whose motivations are set against that backdrop. The consortium brings together 10 partners whose breadth of expertise spans the areas software quality management, legal, civil engineering and traditional memory institutions. Digital preservation is not in fact anything new. It has existed for a long time, but in a domain where the focus has been on the preservation of static digital artefacts. TIMBUS introduces a novel approach which seeks to leverage this expertise pool to implement solutions capable of preserving more complex digital objects than previously possible. The big picture, therefore, is that the TIMBUS research is forming part of the story of how our society is going to deal with this problem of long-term access to our digital assets.

The digital objects which the TIMBUS project will preserve are entire business processes (or any single or distributed ‘software environments’, to put it in more generic terms). TIMBUS is tasked with describing these processes and capturing their interdependencies and constituent components in order to re-deploy them in the future. The intent is to do this in a way which allows future interaction with them. While the rapid pace of IT evolution makes it difficult to predict with any certainty what types of technology changes we can expect to see, the past at least can be an indicator of what we might expect. Hardware will change, but the use of game console emulators has shown that doesn’t necessarily mean all our old digital artefacts become unusable. TIMBUS therefore adopts the idea of hardware abstraction in two ways:

  • Firstly, abstraction is employed in the re-deployment of the software environments which constitute a business process. Long after the hardware has ceased to be available, emulation can be used to re-run legacy operating systems and their supporting suite of applications provided the original components, along with sufficient metadata, has been preserved.
  • In the second instance, the preservation system itself is a software environment. Its back end storage should not be dependent on any particular medium (tape, disk), vendor or technology. In this regard TIMBUS abstracts from back-end storage through the use of POSIX interfaces, meaning that our architecture supports the replacement of the back-end without major re-development of the preservation solution itself. The TIMBUS prototypes are being developed on cloud based block storage at present. Block storage allows us to address many issues relating to data fixity while at the same time utilising API's which can run cross-vendor so that we are not tied into a particular block storage solution.

Even this approach does not guarantee success. Preservation in such a manner will not be the usual “file and forget” approach taken to traditional file archival in many industries. Instead, success can only be ensured from an on-going curation function which will be needed to ensure that the re-deployment process always has a container available to it that can host our preserved environments and execute them on current generation hardware. Without this function, the preserved environments remain susceptible to obsolesce, though less so than if traditional archival methods had been employed. However, for now this abstraction approach holds promise and the development of prototypes to investigate and demonstrate the suitability of hardware abstraction to the preservation and re-deployment of software environments is currently underway within TIMBUS.  As our research advances, this blog will relate more details of the technologies we are using and point to any online demonstrators which we publish.

Please register in order to comment

You have no rights to post comments