Emulators are a crucial building block in digital preservation strategies. They are supposed to bridge the widening gap between past and current digital environments. Like every other software they need to be updated from time to time if the actual environment they are running in changes. But, independent of those updates, the internal structures created for the original environments (of hard- and software) to be preserved have to be kept constant and reliable over time.
Not many emulators are digital preservation ready in the sense that they are programmed with long-term availability in mind. Dioscuri is one of the exceptions but is far from complete and not useable for the newer X86 operating systems like BSD, Linux, Windows 95 and above. QEMU fills a major gap here, as it is Open Source, has a large user community, and provides an X86 machine from the ISA beginnings to the modern variants. It can run every relevant X86 operating system and offers interfaces to be embedded into preservation frameworks for automated work flows.
In Grenoble, France, the first QEMU User Forum takes place March 18th and the OPF will be represented by Klaus Rechert of Freiburg University, who will give a talk on QEMU in digital preservation. We hope to raise awareness of the DP requirements among the technological-oriented user base of the tool and start a discussion how the DP community could contribute to the development efforts. It would be great if organizations like the OPF could help to support the QEMU development and quality ensurance. A joint effort of a number of memory institutions could result in a relevant contribution without overstraining the resources of single partners.
New communities for old
Several members of the KEEP team attended the QEMU workshop in support of the OPF paper mentioned above (http://www.keep-project.eu/ezpub2/index.php). Although Digital Preservation was not a mainstream topic at this workshop, the paper received an enthusiastic response, producing resonances in terms of both computer history and video games, each of whose communities are important to involve in DP work. For example, getting developers from the QEMU and the video game communities to automatically include metadata in their software would be a great help when specifying technical environments to emulate different digital objects, a subject that we hope to discuss in future on a dedicated QEMU users’ mailing list. Just how much information on emulation / technical environments etc. the computer history community holds is something KEEP is looking into.
At the workshop there were other synergies too, such as discussions around virtualization. The virtual machine is fundamental to KEEP and a key factor in future emulation strategies. In the JISC POCOS project (http://www.jisc.ac.uk/whatwedo/programmes/inf11/digpres/pocos.aspx) we are looking at more communities (e.g. software based artists) that will be using emulation in the future.
Does this mean that emulation has now come of age?
First Results of QEMU Testing
Achilles thesis on research into automated QEMU testing is available online from E-Lis. It demonstrates the feasibility of the VNCplay approach to do automated testing of emulation. At the momement the service requires the emulator to be directly installed to the testing framework. This should be extended to be a more generic service in the future where third party (the developers) could easily upload the binary and run tests on it. In general the idea stirred quite some interest at the QEMU workshop (thanks for the support there!):
This prototype is Java based and accessible via web frontend. In a future iteration (we hope to find some other master student for this) a proper more abstract web service should be provided which allows uploading of new QEMU versions and un-attended operation.