Toward the integration of emulation and virtualization strategies into the conservation of software-based art collections
Part of PERICLES core mission is finding ways to manage the challenges posed by the changing technical environments of complex digital content. For software-based artworks, obsolescence of the original technical platform and the resulting need to migrate the work to a new one, can result in inaccurate performance of the work, or at worst, render it completely inaccessible.
Emulation and virtualization techniques offer potential preservation strategies for such works by removing the artwork’s dependencies on ephemeral physical computer hardware. By simulating the presence of physical components as virtual hardware, it is possible to maintain a more portable environment that can run the original software independently of the original hardware. Despite the topics prevalence in digital preservation literature, questions remain over how to effectively apply emulation and virtualization to real world collections. A new tract of research for the PERICLES project attempted to address this gap. The work was led by Klaus Rechert at the University of Freiburg, who guided the project with his deep technical expertise in emulation and virtualization strategies for long-term digital preservation. The research was undertaken in collaboration with Patricia Falcão and Tom Ensom at Tate, who provided real world context based on experience working with collections of software-based artworks. This meeting of perspectives has ensured that the resulting framework provides a clear means of exchange and shared understanding between both parties, despite differences in depth of technical specialisation.
The foundation of our approach is a conceptual model to guide the analysis and documentation of the hardware and software elements of software-based artworks and their environments, particularly dependencies and risks. Technical analysis of the digital components of software-based artwork (from here on referred to as the ‘digital artefact’) and its environment should be based on three conceptual layers:
- Digital artefact description and configuration
- Hardware runtime environment and configuration
- Hardware environment
Through the separate analysis and documentation of these layers, preservation requirements can be assessed, and risks associated with interdependencies between the layers identified. This differentiated view on technological risk and potential mitigation options helps to make a changing technological environment more manageable. The higher the degree to which an artefact can be abstracted from its technical environment, the more options for preservation actions are available. An artefact's software environment is of particular concern for long-term preservation, as it usually connects the digital artefact with hardware dependencies. If the software environment is known in its composition and configuration, it can be manipulated in order to mitigate the effects of change. The environment can be rebuilt in order to mitigate risk (e.g. substituting problematic dependencies) or the artefact can be adapted for different technical environments (e.g. external dependencies may be removed).
We suggest that a successful emulation or virtualization based preservation strategy should focus on the acquisition and maintenance of software environments as virtual disk images (captures of the content of a hard drive or other storage device) and the ongoing analysis and monitoring of their technical interfaces. The information gathered during the technical analysis phase can be used to select the most appropriate options when creating disk images, and to make decisions about which emulation or virtualization platform is more appropriate in supporting these disk images. The process of disk image creation and emulation/virtualization provides a consistent process to build knowledge about a works functionality and guide the selection of preservation strategies. If a software environment has been successfully rebuilt through emulation or virtualization, it likely that all dependencies are well understood, and a complete, verified and potentially machine readable description of the software and hardware environments can be generated during the process.
With the decoupling of digital artefact and environment, emulation and virtualization work can often be standardized and shared as virtual disk images. While an emulation or virtualization-based preservation strategy for a software-based artwork still requires a work-specific preservation plan to ensure both conceptual and technical properties are maintained, the software and hardware environment layers are not work-specific. In many cases software environments can be shared among several artworks, pointing to a need for coordinated (and therefore cost effective) efforts to create, manage and consolidate these images. The characterisation of common components among many artefacts of a similar type makes it possible to focus monitoring and technical migration work. If a particular technical interface breaks, a migration path is only developed once and applied to all suitable artefacts.
A workshop held at Tate as part of this project brought together time-based media conservation, research and curatorial staff at Tate with the technical expertise of Klaus Rechert and Dragan Espenschied (digital conservator at Rhizome). The group critically examined the application of emulation and virtualization strategies to a series of software-based artwork case studies from the Tate collection. This work further informed the development of the model, and particularly served to highlight that while emulation and virtualization can support preservation goals, it is no ‘silver bullet’, and much supporting work is required to fully understand the work and its properties.
In developing our approach, we identified a need for systematic archiving of common software components that may form software environments. Primarily, a dedicated software archive (including management of licenses) is required to help ensure that a given software environment can be rebuilt. Additionally, it is useful to maintain a testbed of typical environments and common technical interfaces to be tested with newly released emulation and virtualization platforms. In contrast to testing an artwork’s work-specific significant properties, these elements of technical infrastructure can be shared and re-used not only for sets of similar artworks, but also among different institutions and software types.
The approach we have developed focuses on the digital artefact’s technical dependencies, instead of looking at work-specific properties. If emulators were able to perfectly reproduce out-dated hardware components, this technical perspective would be sufficient – at least in accounting for the preservation of any computational aspects of the artwork. In practice however, emulators are far from perfect in this respect and as such, verification of an emulated result is essential. This is a complex challenge as properties of a work may be subtle and hard to automatically verify. This is an ongoing area of research in emulation and virtualization, and more work in this area should be considered a priority moving forward.
A brief summary of the project has been presented here, but several forthcoming publications provide more detail. The first is a paper which will be presented at iPRES 2016 in Bern this October. The second is a report to be released shortly via PERICLES, which includes in-depth discussion of issues touched on above, recommendations and a draft workflow for applying emulation-based preservation strategies.
Tom Ensom, August 2016
Tom is a PhD student hosted by King's College London, working collaboratively with Tate on, 'Technical Narratives: Method, Purpose, Use and Value in the Technical Description and Analysis of Software-Based Art' and is involved in the PERICLES project through this work.
Image: Discussion in progress at the workshop, with an emulated test version of Rafael Lozano-Hemmer’s Subtitled Public (2005) in the background. (Photo credit: Alistair Ashe)