Lesson learned from the Australian IR experience, we should recognize that some institutions will need hosted software

I have been invited to speak at DFRIC 2009 in Tokyo in December, and I'm working on a paper with my CAIRSS colleague Kate Watson, to form the basis for my presentation, the working title of which is Suggestions for Asian/Australasian regional cooperation based on a critical evaluation of collaboration and standardization across Australian Institutional Repositories.

We want to look at the history of collaboration in the Australian IR space, and try to draw out some lessons for future efforts nationally, regionally and globally. With the benefit of hindsight, I think that one thing we could have done better in the main Australian Government funded collaboration projects, APSR, ARROW and RUBRIC was address the need for hosted repository solutions for small and/or resource-limited organizations.

It's not that we didn't collectively understand or anticipate the requirement. For example we looked seriously during RUBRIC at offering hosted solutions after prompting from some of our partners, and USQ (where I work, and the lead institution on RUBRIC) subsequently helped two small Higher Education institutions to set up Eprints repositories which are hosted by Southampton. UQ has looked at offering hosting in the past for their Fez software. And since ARROW finished I have been in discussions about hosting with people from more than one institution in the extended ARROW family, using the VITAL software from VTLS. These folks are considering the benefits of a hosted solution as they are finding that even a vendor-supported on-site repository is too much to manage, given the institutional context for repositories. We need to keep this in mind in future projects.

Of course there are some commercial hosting providers in the market, and that's good, but but we missed the opportunity while RUBRIC and ARROW and APSR were running to put serious effort into investigating how those solutions work with services like the NLA's Australian Research Online, or how easy it will be for institutions to migrate their data out of them when the time comes. Evaluating commercial services other than VITAL was out of scope for RUBRIC, something I would like to make sure we don't do in future similar projects.

I have a few thoughts on this:

1. For future collaborations, including forthcoming work on ANDS I think we should consider that some people will want a hosted solution and work with current and potential vendors to make sure that the interests of small institutions are represented. This might mean encouraging somebody to enter the market with an open-source based hosted solution.

2. It has crossed my mind that maybe one or more of the larger institutions might like to consider hosting services for others alongside their main repository; this could be quite a low-cost service particularly if the smaller institutions are willing to accept very limited options for customization.

3. Which leads to the idea that maybe we should be considering what AlisonDellit suggested here in a comment when I called for vendors to enter the market to support open source repository software. She suggested the 'one big repository' approach:

   Open Access, preservation and user-friendly workflows could be more easily facilitated through a single Australian repository than through dozens of separate repositories, I think. And there are institutions which could host such a thing, whatever governance arrangements it would have. If it could go along with a centralised system for identifying researchers and grouping their work in sensible ways, it would be awesome.

   I don't think we'd necessarily see a single repository for everyone, but what if a large or well resourced institution decided to run a multi-institution IR or data repository where they rent out services to smaller institutions? Not just another instance of the software – the idea would let other institutions put stuff into one joint system, you'd be able to see each institution's stuff on it's own of course as well as search the whole lot.

   A major advantage would be that the institutions could share metadata standards and so, to reduce some of the problems inherent in federating or aggregating content where local customization has reduced standardization, although as Alison points out, local customization and integration is still a requirement.

Both 2 and 3 above could help a large institution pay some of their repository management overhead, but could potentially still be cheaper for small institutions that having software installed and maintained on site. Curation workflows could either be done by each institution, or could be taken on by the host.

Kate and I will be putting something along these lines into our paper, but I am interested in feedback on these three points.

© Peter Sefton 2009. Licensed under Creative Commons Attribution-Share Alike 2.5 Australia. <http://creativecommons.org/licenses/by-sa/2.5/au/>