Some eye-opening costs for running IRs have emerged recently, highlighting
the dangers of setting specifications based on extended and possibly
untested functionality, that go well beyond what is needed for Open Access,
e.g.
https://arl.org/Lists/SPARC-OAForum/Message/2509.html
http://www.inasp.info/newslet/nov05.shtml#8
The former reveals less about IR software than the wide range of
specifications that are intended to pass as IRs.
The latter lists an extended range of functions for an IR, then admits
there are currently no turnkey solutions to support this list, only partial
solutions, and that this specification is costly: "three-year start-up
costs for hardware and software alone are over $300,000", without including
staff costs apparently.
Most other institutions just want to get an IR up and running quickly at
low cost.
A fully functioning IR to capture and make accessible all the research
outputs of an institution can be achieved at a fraction of highest cost found.
To prove it the newly launched EPrints Services team offers a range of
service packages from supporting local hosting to a fully hosted service
http://www.eprints.org/services/sales/
This approach also goes some way to reducing an institution's risk on costs.
To understand why IRs are needed and how to build one, Professor Arthur
Sale, a developer of software to measure usage of IRs, offers a practical
approach
http://www.eprints.org/news/features/worlds_best_practice.php
The higher costs in the examples above almost certainly include more
complex and costly DL-inspired functionality for IRs, which will arrive,
but these investigations can be left to others for now. As Arthur Sale puts
it: "Can people fly flags and ring bells over conquering Everest when it
turns out to be Highgate Hill?"
Steve Hitchcock
EPrints Community Manager
IAM Group, School of Electronics and Computer Science
University of Southampton, SO17 1BJ, UK
Email: sh94r_at_ecs.soton.ac.uk
Tel: +44 (0)23 8059 3256 Fax: +44 (0)23 8059 2865
http://www.eprints.org/community/
Received on Wed Nov 30 2005 - 20:43:35 GMT