ADMIRAL planning phase 2
From ImageWeb
ADMIRAL project phase 2 plan
Outline plan for the period Jan-2011 to Apr 2011
Workpackage summary (from proposal)
| WP | Activity |
|---|---|
| WP0 | Project management |
| WP1 | Data storage/LSDS |
| WP2 | Annotation |
| WP3 | Dataset and metadata packaging |
| WP4 | Repository submission of selected datasets |
| WP5 | Links to external web services |
| WP6 | Formalization of annotations |
| WP7 | DOIs for datasets |
| WP8 | Licences for published datasets |
| WP9 | Evaluation |
Our priority will be to consolidate and extend the ADMIRAL deployments with research groups, and to ensure that we can continue to support them and thereby gain greater understanding of local-level data management concerns through other projects to be conducted over the coming years. In particular, this will require continued work on WP1, WP2 and WP4 in consultation with our research group users.
Following delays in phase 1 of the project, WP5 and WP6 must be considered "at risk", and will be addressed to the extent permitted by other project commitments, and also to the extent that they may be used to support our research users' requirements.
WP7 issuance of DOIs is being handled in concert by the university library service and British Library, and we see no obstacles to its successful implementation.
WP8 scope needs to be fully defined, and is not expected to be a great effort. The main goals are to provide a suitable interface for a user to specify a desired licence, and to include some form of machine processable licence statement for each published dataset accessible via the databank service. We will aim to identify a set of preferred licences, and a default one to be suggested where no alternative has been specified.
WP9 will involve the recording and analysis of information gleaned from ongoing interactions with our research users, including a data audit "exit poll".
For the remainder of the project, we don't have a set roadmap of activities, but intend to address specific points as they arise, selected to address the broad priorities indicated above.
Activity analysis
At the time of preparing this plan, the following areas of activity have been identified (with indications of priority in the range 1-3, where priorty 1 is the highest.
Databank deployment with real datasets submitted (priority 1)
To prove the ADMIRAL concept, we aim to have researchers submit real datasets for preservation and publication in the library Databank service. This depends critically on getting a stable, public deployment of the Datbank service that we have ben testing to date. See also http://imageweb.zoo.ox.ac.uk/wiki/index.php/ADMIRAL_20101216_project_partners_meeting for relevant partner discussions.
Secondly, we need to make improvements to the basic dataset submission interface so that researchers are able to use it without guidance to curate their datasets for preservation and publication.
Finally, we need to work with researchers to encourage them to select and submit datasets, and to gather feedback about the functionality and usability of the ADMIRAL submission, resubmission, browse and query interfaces.
ADMIRAL deployment issues (priority 1)
As we deploy ADMIRAL to research groups, we are starting to face maintenance problems. In particular, the current framework for deploying updated ADMIRAL software to existing research users is not well defined. This needs to be improved so that the upgrades can be deployed easily and safely without disrupting researchers' day-to-day work.
(Note that the supporting tools used for dataset submission and viewing are already easily updated in situ - it is the base file storage and management system upgrade mechanisms that need to be improved.)
ADMIRAL features (priority 2-3)
A number of improvements to existing ADMIRAL features have been noted, and new ones doubtless will be raised, with varying importance. These will be prioritized and addressed according to the needs expressed by our users.
Integration with departmental storage facility (priority 1)
This activity was not envisaged in the original plan, but arises as our researchers are increasingly generating high volumes of data that cannot be accommodated by the ADMIRAL servers alone. In response to these needs, our departmental IT support has installed a shared storage service that will be made available to research groups on a deployment cost recovery basis. This facility provides raw storage only, with no associated data management capabilities, and is currently estimated to cost about £400 per formatted terabyte of storage. This storage is accessed using iSCSI protocols, which the ADMIRAL server hardware is capable of supporting.
We aim to confirm that the ADMIRAL servers can be configured to use this facility, and then make it available to our research group users to meet increasing needs for data storage that are currently hampering their use of the ADMIRAL services (with the cost of such additional capacity being met by the users).
We also aim to negotiate with IT support to document a clarified cost model that takes into account the 10 year data maintenance requirement of BBSRC research proposals (http://www.bbsrc.ac.uk/web/FILES/Policies/data_sharing_policy.pdf), a widley used source of research funding in our department.
Deploy ADMIRAL services to the remaining research groups (priority 2)
We next intend to create an ADMIRAL deployment for the Behaviour group, and are in discussion about their requirements.
We are also in discussion with the Elephant group who have some very different requirements due to the remote field-based nature of their data acquisition and analysis activities. If we can successfully deploy ADMIRAL to meet their needs, this will prove that the basic ADMIRAL data management model is sufficiently flexible to meet a great variety of research needs. However, this may need to address some requirements that lie well outside the original ADMIRAL scope and, as such, field deployment may require some additional funding.
Technical debt and bugs (priority 1-3)
As the ADMIRAL software is deployed, we will inevitable encounter a range of technical problems and needed improvements that do not necessarily impact users directly, but that do impact the sustainability of the ADMIRAL system. These will be prioritized taking acccount of their impact on sustainability, as well as any impact they may have on our users.
Dissemination (priority 3)
In this final phase of the project, we need to prepare to disseminate what we have learned in developing and deploying ADMIRAL, using the normal channels.
One particular goal for us is to create a stand-alone demonstration version of the ADMIRAL environment (probably using VMWare) which can be used to show the full front-to-back process for creation of datasets, submission and publication, showing how this may lead to subsequent use by new researchers.
We shall be seeking to use lessons learned from ADMIRAL as input to other projects with which we are engaged, with a particular involvement in Research Objects (http://eprints.ecs.soton.ac.uk/21587/) and data management workflow preservation.
Sustainability planning
We are also developing plans to incorporate ADMIRAL and our relationships with research groups to provide ongoing support and evaluation of data management activities.
