Technology Watch Report

Institutional Repositories in the context of

Digital Preservation

Paul Wheatley
University of Leeds

DPC Technology Watch Series Report 04-02

March 2004
Digital Preservation Coalition 2004

1.0 Scope
This report will focus on the requirements, functions and use of digital preservation in an
institutional repository context. It will also provide an overview of existing institutional
repository software as well as details of working systems and their core aims and
purpose. Software that could be installed and used to perform the role of an institutional
repository will be covered by this report.
A number of existing publications provide comparisons between existing institutional
repository software. A specification document for the DARE Project contains a
comparative discussion of DSpace, ARNO and NCP [1]. The Open Society Institute has,
at the time of writing this report, released a systematic comparison of repository software
entitled "A Guide to Institutional Repository Software" [2]. This OSI guide provides a
detailed checklist of features present in open source institutional repository software.
These documents do not discuss digital preservation in any detail (if at all). A publication
by the DAEDALUS Project [3] describes the experiences at Glasgow University in
implementing, running, configuring and building on both the ePrints software and the
DSpace software.
Rather than duplicating this work, this DPC report will concentrate on issues of crucial
importance in achieving long term digital preservation in the context of institutional
repositories. The reports described above only briefly touch on digital preservation
related features.

2.0 Institutional repository - definition

Clifford Lynch defines his view of an institutional repository as "a university-based
institutional repository is a set of services that a university offers to the members of its
community for the management and dissemination of digital materials created by the
institution and its community members. It is most essentially an organizational
commitment to the stewardship of these digital materials, including long-term
preservation where appropriate, as well as organization and access or distribution. [4].
The crucial word Lynch uses here, is community. The term institutional repository
implies a community based service although this is interpreted by repository developers
in different ways. Some embody a cross-subject, cross-department service which requires
flexibility to meet the requirements of many different types of users. Some focus more
specifically on a particular subject and possibly type of material to be archived, while still
delivering an institution-wide service.
Although, as Lynch implies, the term institutional repository suggests a higher education
context, this is not always the case. Institutions with requirements to store, preserve and
provide access to digital materials may require their own repositories, and this is likely to
become more common over time.

In the context of this report, the term institutional repository will be used simply to refer
to an actual instance of an institutional repository or the software that enables an
institutional repository.

3.0 Technology Watch

The development of institutional repositories is very much in its infancy. The specific
requirements for repositories are still being defined and existing software is only just
beginning to attempt to fulfill these needs. With respect to the application of digital
preservation in institutional repositories, current implementations are at an early stage,
but understanding of the problem and the solutions to address it are advancing rapidly.
This makes the subject of Institutional Repositories in the context of digital preservation
an important area for this current series of DPC Technology Watch reports.
The subject has relevance to the DPC membership with regard to:
Understanding the key issues relating to digital preservation and longevity
Where the contribution of requirements, expertise and best practice can be made to
institutional repository development
Awareness of purpose and suitability of current products and technologies
Readers should also be aware of a related DPC report The Open Archival Information
System Reference Model: Introductory Guide [5].

4.0 Repository purpose of use

Institutional repository software has been developed with a range of aims and purposes in
mind, and there is not a consensus across the available products of what those roles are.
The contention between improving access and dissemination now, versus protecting that
access for posterity (preservation) is discussed in detail by Pinfield and James [6]. In
practical terms there is an apparent difference in focus of institutional repository
implementations, where some prioritize digital preservation as an aim and others do not
(for example DSpace and ePrints respectively). Experiences of the DAEDALUS Project
[3] suggest that repository software like DSpace and ePrints are not really in direct
competition. In fact, these differing implementations have quite distinct aims but similar
attributes.
Differences can also be found in repository software with regard to types of materials the
software is primarily designed to hold. Some repository software concentrates (primarily
if not exclusively) on document type materials and others (perhaps more embodying the
true sense of the term institutional repository) are designed from the outset to attempt to
realistically meet the requirements of holding any type of digital object.
The overall purpose and aim of an institutional repository will have a significant bearing
on the selection policy and practice for that repository. Materials may be selected for

archiving by those responsible for administering the central repository, by devolved

community based management or even by individual users (or creators of digital
materials). Institutional repositories may not even have an overall selection policy, but
instead simply provide the mechanisms and flexibility for the development of community
or department level policy.
The Cedars project suggested that issues of digital preservation (eg. data format,
dependencies, etc) should not affect initial selection decisions [7], but repository
implementations that limit the range of acceptable submission formats are common. It
remains to be seen whether digital preservation understanding and support will advance
to a sufficient level that choices in selection and submission really can be format
independent.

5.0 Current best practice?

A consensus in best practice in performing digital preservation is yet to be reached.
While recent research and testing is now getting closer to practical implementation and
real world usage, the definition of best practice is some distance away. OAIS (Open
Archival Information System [8] provides a useful framework of terminology and
understanding, describing the key elements of a digital repository design. The
RLG/OCLC work on Trusted Digital Repositories [9] builds on the OAIS structures,
detailing the requirements for effective repository based digital preservation.
For now, the foundations provided by OAIS, the RLG/OCLC work and the practical
experiences of those who have developed digital repositories and preservation systems
must guide archival and preservation work where possible. Some of these key
developments are described in more detail below.

6.0 Requirements and aims for effective digital preservation

6.1 Understanding the problem
Where the long term accessibility of archived objects is a key aim of a particular digital
repository, a number of requirements must be met in order to effectively meet this goal.
In order to understand the process necessary to achieve the long term digital preservation
of objects placed within a repository, it is first useful to break down what we understand
as effective preservation itself. The key functional goals can be summarised as follows:
1. Data can be maintained in the repository without being damaged, lost or maliciously
altered.
2. Data can be found, extracted from the archive and served to a user.
3. Data can be interpreted and understood by the user.
4. Goals 1, 2 and 3 can be achieved in the long term.

Goal 1 is easily recognized as a key requirement for any form of digital repository and
has been well fulfilled by fundamental techniques of computer science for many years. In
recent times, developments in areas such as security, authenticity, verification and storage
have made significant advances and it is clear that current repository software and actual
repository implementations address this goal well. As technology advances, procedure,
software, and technology will also need to progress.
Complications arise when digital objects are tied to the media upon which they are
stored. But as Holdsworth points out, The media is not the message [10]. By mapping
representations of digital objects to simple bytestreams, while ensuring that all the
required significant properties are retained, media independent preservation can be
achieved.
The second goal has received much attention in the last few years as the "information
revolution" has sought to improve access, finding and integration. The process of Access
and Dissemination (in OAIS terms) integrates these wider search and retrieval issues with
the user authentication and the extraction and delivery of the data to the user.
Repositories will need to support a number of methods of searching and retrieval
dependent upon their subject and user base. These methods are likely to change over time
as standards and widely accepted techniques go in and out of fashion. For example, the
Open Archives Initiative Protocol for Metadata Harvesting is currently an essential
access route to support and is seeing wide uptake across a number of communities.
An issue which remains unclear is the underlying requirement for the unique and
persistent identification of each digital object in a repository. There must be a system for
resolving these identifiers and directing queries to the physical location of stored digital
objects. In most cases this will be the underlying technology on which searching and
finding aids depend. RLG/OCLC noted in 2001 [9] that a standard which supported the
requirements of digital preservation was yet to emerge and recommended that efforts be
made in this area. Since this report was produced, standardisation in this area has still not
been achieved. Unique identifiers will be addressed in more detail below.
The third goal can be seen as the core digital preservation aspect of the repository
question. Although the first two goals must certainly be achieved in order to provide real
long term digital preservation, the third goal ensures the work involved in the first two is
still useful in the long term. Ingest, archiving and then providing access to a digital object
in the space of for example, 2 years, is unlikely to tax the success of meeting of goal 3.
Add another 10 years or more between ingest and access and a user will struggle to make
sense of the digital object they have been provided with as technology change makes the
original hardware and software obsolete.
OAIS loosely terms the process required to achieve this goal as "Preservation planning",
but rather than effectively forming a separate function within the archival model, it plays
an important integrated role within most of the key archival processes which OAIS also
describes (eg. ingest, administration, dissemination). This integration aspect must be

recognised and understood. Repositories must provide the flexibility to allow digital
preservation functions to be incorporated as they are developed.
This is the least understood of these 4 goals, and continues to see a trickle of research into
techniques and consequently best practice.
The fourth and final goal suggests that some thought and effort needs to be given to
ensuring that the first 3 goals can still be achieved successfully in the future. This implies
a degree of continuity which ideally should be attained without undue expenditure. While
goal 3 implies consideration of the long term perspective, the first two goals do not and
thought must be given to sustaining them over time. This is particularly important in an
institutional repository context.
6.2 Preservation processes
Ultimately, achieving the third goal requires a process that adds preserves, interprets and
adds meaning to accessed data. However, as suggested above this is not a stand alone
process. It can only be achieved with input throughout what Beagrie and Jones term the
"lifecycle" of a digital resource [11]. In repository terms this requires specific
preservation type processes from ingest, through to storage, administration and finally
access. These functions must capture, store and enable use of various types of metadata.
In particular, Representation Information, which describes how to gain access to the
intellectual content encoded within a digital object (see glossary).
A summary of the functions and infrastructure required might include:
1. A process of ingest that creates or extracts the metadata necessary to ensure
preservation.
2. A framework within which the required Representation Information can be stored,
managed and utilized (a Representation System).
3. A process of "technology watch" which monitors technology dependencies and the
recorded Representation Information, and takes action to ensure continued
preservation where technology obsolescence occurs.
4. A process of rendering (displaying or making sense of) retrieved digital objects.
5. A process and related framework for recording change metadata
Beginning with the ingest of a digital object to an archive, the ingest process must capture
an appropriate amount of Representation Information. This metadata must be stored in an
appropriate way to facilitate both its maintenance (a process of keeping it current) via a
preservation watch function and its use in a representation and rendering capacity (see
below). The first of these functions will monitor the representation information and
technology it depends upon, to ensure it is still current. The second of these functions will
provide a user with the appropriate information to render the digital object, perhaps
starting a further distinct rendering process (for example Migration on Request [12] or
Emulation [13]). The rendering process itself may require additional thought and
resources if the repository in question is responsible for maintaining a relevant rendering

method. This could mean maintaining a current tool or replacing it with a new one. If
format migration is chosen as the preservation strategy, rather than changing or updating
a tool when technology obsolescence occurs, a migration from format to format of all
objects of that type in the repository may have to be made. A related process is to record
change metadata describing any changes made to objects in the repository.
As will be discussed later in this report, not all of these functions need to be undertaken
by a specific repository, but at the very least support will have to be provided for the
integration with external services. In the case of ingest in particular, this is not trivial.
Consequently these issues need to be considered in the design of an institutional
repository.
As well as enabling these functions to provide for long term preservation, the repository
design must also ensure that the repository itself can survive in the long term, again
another important design consideration.

7.0 Preservation functions

The report will now examine the design considerations for institutional repositories that
will provide for long term digital preservation.

7.1 Unique, persistent identification

Purpose - mechanism for managing digital objects, and for finding them even when their
physical location changes.
Method - name allocation, name resolution
As mentioned above, institutional repository developers have recognised the need for
unique identification of archived objects and incorporated them into their designs but a
range of identifier standards have been used. As is often the case with standards there are
several to choose from! Requirements for the design of the identifiers and related systems
will vary from institution to institution but there is consensus on the obvious benefits of
interoperation in finding aids which a single all encompassing standard would provide.
While the design of the identifiers itself can be made sufficiently simple and flexible to
satisfy most requirements, agreement on the resolving services may be harder to reach.
Leadership from a consortium of institutions (perhaps the Digital Preservation Coalition
[14] or Digital Curation Centre [15] in the UK) may be able to champion an appropriate
standard and make progress in this area. Repositories may be able to integrate their
existing identifiers with a flexible but perhaps different leading standard that emerges at a
later date. Prefixing existing local identifiers with an institutional label may facilitate this.
The various standards for unique identification (eg. URN, DOI, ARK, etc) are discussed
in detail on the PADI website [16].

7.2 Ingest
Purpose - An ingest process fulfils a range of aims, but this report will concentrate on the
capture of Representation Information during ingest
Method - Modular tools for identification and verification of file formats and also for the
automated extraction of metadata
Ingest processes which aim to capture metadata are recognised as a crucial area to
develop and automate to reduce this potentially high effort, high cost function. For most
repositories, it is unrealistic to gather and or extract sufficient metadata to enable
preservation. A Range of institutional repository ingest functions will need to be
developed. These include:

Automated extraction of metadata

Automatic identification of file formats
Verification of an objects compliance to a relevant file format specification

The highly specialised nature of these functions will necessitate modular solutions which
can be plugged into institutional repository systems as required. Integration with other
key preservation systems such as those that address the storage and use of Representation
Information will be crucial. Research, development and evaluation work in these areas is
currently being undertaken by MIT, the University of Pennsylvania and the UK National
Archives.

7.3 Representation Systems

Purpose - Representation Systems provide a mechanism for storing and utilising
Representation Information. Representation Information provides the knowledge of how
to gain access to the intellectual content encoded in the digital object (see glossary).
Method - There are two emerging and related methods for addressing semantic
Representation Information - file format registries and OAIS Representation Networks.
There are a range of techniques for addressing structural Representation Information,
from the use of existing technologies to newly defined standards like METS [17].
While much importance has been placed on the definition of preservation metadata
schemas, the hardest part of the metadata problem is certainly the technical metadata or
Representation Information (see also Recording Change Metadata below).
Examination of existing preservation metadata schemas suggests that the nonRepresentation Information component of the problem is well understood. Achieving
standardisation for the non-Representation Information component of preservation
metadata will improve and simplify interoperability and searching. This is now primarily

an organizational problem and will depend on co-operation and agreement between the
major players in the field, who are able to develop and promote an appropriate standard.
In the case of Representation Information, a set of basic fields within a metadata schema
will not be sufficient. Where this has been attempted (for example in the recent NLNZ
Preservation Metadata Schema [18]) the Representation Information fields have been
defined very weakly. Even where technical dependencies can be listed in fields of this
type (eg, format, rendering application, system the application runs in, etc) the task of
maintaining and updating this information over time as it becomes obsolete is colossal!
Changes will have to be applied to the metadata of every applicable object in a
repository. Moving this to a referenced external system where only one entry
(representing possibly thousands of objects) has clear advantages.
A range of approaches have been suggested for describing structural Representation
Information. The Cedars Project [19] took a pragmatic view, and utilised existing
technologies to describe simple file structures and relationships. The use of the TAR file
structure and associated tools for unpacking TAR to a usable file system (details of which
were described in a Representation Network (see below)), provided an effective way to
address objects composed of multiple files. The more recent development of the METS
standard [17] shows promise for describing more detailed structural Representation
Information and is being explored in detail by various institutions and communities,
including DSpace.
There is a growing consensus that an external system or repository of referenced
Representation Information will provide a more manageable and effective solution than
raw repository based metadata fields when dealing with semantic Representation
Information. For the purposes of this report these repositories of semantic Representation
Information have been termed "Representation Systems". These systems store the
technical metadata independently from the digital objects in a repository, allowing
several objects of the same format to point to the same single piece of metadata.
Monitoring and updating the metadata then becomes a much simpler and more
manageable task.
Representation Systems will play a crucial role in achieving long term digital
preservation and data curation. So far little work has been devoted to their development
and only one system is currently in operational use, PRONOM at the UK National
Archives [20]. Both PRONOM and the proposed Global Digital Format Registry [21]
broadly follow a "file format registry" approach, which is based around a simple database
of file formats. A defined categorisation of file formats (ideally far more specific than for
example, MIME) is used to structure the recorded Representation Information. In its most
simple form, the file format registry will be held in a database at a single location. It is
likely that access would be provided to remote sites sharing the registry via the internet
(this facility will be present in the next version of PRONOM). The simplicity of this
approach is its strong point, but it is unclear if sufficient format detail can be maintained
without creating lengthy and unusable categories of file formats (what is classed as a
format?). The next release of PRONOM and expected external take up will provide an

invaluable practical evaluation of these issues. In particular whether the PRONOM

format categories have reached a suitable compromise between practical use and detail
specifics.
The second approach was suggested in the OAIS framework as a network of technical
metadata, termed a Representation Network. Much of the information about a particular
file format will form some overlap with other file formats. So the Representation
Network approach utilises a network of nodes of metadata and a system for pointing from
one node to another. For example, nodes for the Word format and the PDF format will
both point to nodes describing ASCII and UNICODE.
Initially a Representation Network for a current file format may be relatively simple. For
example, a node describing the Word format will have links to documentation of the file
structure, a link to a node describing Microsoft (which in turn will point to the Microsoft
website) and a link to a node describing a method of rendering; perhaps using Word
viewer. In ten years time when Word viewer becomes obsolete, the representation
network will be updated with new nodes to describe new methods of rendering, in this
case perhaps migration or emulation tools. So Representation Networks will grow over
time as technologies are monitored and action is taken to ensure access to data formats
over time.
The CEDARS Project and subsequent Representation and Rendering Project developed
this concept into a working demonstrator [19]. Cedars argued that by allocating unique
identifiers to each node of Representation Information and providing a form of name
resolution service, efficient networks of Representation Information could be constructed.
Preserved digital objects reference appropriate Representation Information using the
identifier and resolving system. A key advantage is the distributed nature of the Cedars
Representation Network, where different institutions could provide Representation
Information in their areas of format expertise. These would then be available to all other
institutions using the same system. A community of digital preservationists could then be
created who share, maintain and test each others Representation Information. The
network structure facilitates efficient technology watch functions (see below) by
highlighting the key dependencies where technology obsolescence will occur [22].
Representation Networks have not been developed to an operational level but were tested
using functional demonstrators as part of the Cedars project.
Discussions on the Global Digital Format Registry suggest that even with a working
format registry, some materials will be too unique to be adequately described by the
Representation Information contained in the registry and so perhaps inevitably
institutionally controlled Representation Networks may also be needed. The flexibility
and distributed nature of the Representation Network approach could well prove
important.
Populating Representation Systems is a major challenge in itself. Gaining access to
accurate proprietary file format information can be very difficult, and even open file
format documentation can be incomplete or inaccurate [23]. The scale of the problem will

certainly require co-operation and Representation Information sharing across the digital
preservation community.
Whichever of these two approaches is widely adopted (and possibly both could work in
tandem) Institutional Repository software must be designed with the flexibility to
incorporate linking to Representation Systems as they become available. Given that most
of the physical metadata is referenced, the main issues will reside in the integration of the
ingest and dissemination processes of the repository software. Integration and
interoperability with repositories will require open, flexible designs and some degree of
standardisation.
7.4 Technology Watch
Purpose - A Technology Watch function monitors Representation Information and related
rendering capabilities and provides alerts when the Representation Information is no
longer current due to technology obsolescence.
Method - Unclear at the current time, but will likely involve a range of techniques from
automated processes to manual surveys and evaluations.
Technology Watch is a frequent function that must be performed to ensure
Representation Information is maintained in a current state. Representation Systems are
likely to incorporate integrated Technology Watch functions and also rely on external
Technology Watch operations like that of the DPC. As well as the primary role of
maintaining Representation Information, technology watch must also be provided for the
software and hardware on which repositories themselves depend (see Overall Repository
Structure, below). Although the shape and form of adequate technology watch functions
is yet to be fully understood it seems clear that as with Representation Systems, cooperation and community integration will be important, where sharing of results and
expertise will be required.
7.5 Rendering
Purpose - To turn a bytestream into meaningful information or to gain access to the
intellectual content encapsulated in the raw data.
Method - Many rendering strategies have been proposed, including migration and
emulation.
Rendering will not be addressed in detail in this report as it has been discussed in detail
elsewhere [24] [25], and the implications for the design and integration with repositories
are effectively covered under Representation Systems and Recording Change Metadata.
7.6 Overall Repository Structure
Purpose - To ensure a repository survives technological change.
Method - Layered design and the choice of stable technologies in the construction of the
repository.

Just as Representation Information will need to be monitored and updated as technology

obsolescence impacts upon it, repositories themselves will change and develop over time.
Change may occur to ensure continued function of the archive in the face of dependent
technologies becoming obsolete, or alternatively may be instigated to provide new or
different services at the point of use. If long term survival of a repository is required, it
must therefore be accepted that significant parts of the repository will change. For
example, in the case of the LEEDS long term file store [10], migrations across 3 different
storage technologies were required in the space of just 12 years.

Figure 1 : repository abstractions

Figure 1 shows a simple break down of abstracted repository layers. By careful design of
the interfaces between these layers, the providing technologies for the layers themselves
(primarily the top and bottom layers) can be changed without major impact to the
repository as a whole. As the technical, functional and user paradigms of modern
computing change over time (and go in and out of favour) we have to accept that the
applications which depend on them will also change. Clearly, the current front end
implementations of the repositories described at the start of this report will not survive in
their present form for five years, let alone one hundred. Choosing a sensible high level
design can simplify this inevitable change and hopefully prevent any data loss in the
process. Browsing through the many institutional repository implementations on the web
reveals several with warning labels about adding content to systems that may close down
without hope of migrating data to a new replacement system. The dangers of not
addressing this issue are all too apparent.
7.7 Recording Change metadata
Purpose - To record changes made to digital objects in a repository in order to assist in
answering questions of authenticity and to inform future maintenance and preservation
actions.
Method - Requires a repository function to record and update metadata.

Digital objects in a repository and their respective metadata can require changes to be
made to them for a number of reasons. Changes to a digital object may occur following
maintenance or preservation action (eg. format migration), and new versions of a
particular digital object may be created through redaction or revision. These changes
must be recorded in the metadata record as a form of history or change metadata. Note
that recording an audit trail to changes in the metadata as well as the digital object itself
is a sensible course of action [26].
The process of recording change metadata will by necessity have to be quite an integrated
repository function given the range of processes that may alter or update objects or
metadata. Mechanisms for recording change history exist for many other purposes and
are relatively straight forward, but a question remains as to the quantity and detail of
change metadata required in order to fulfill the purpose.

8.0 Institutional repository software

This sections describes open source institutional repository software, listed in
alphabetical order.
8.1 ARNO
Project home : http://www.uba.uva.nl/arno
Example installation : http://dare.uva.nl
Core purpose : document storage and dissemination
ARNO is a university archive server based repository system that has seen significant
take up amongst Dutch HE institutions. With a strong focus on repository integration,
ARNOs focus is on the stewardship and dissemination of scholarly publication and
supports the OAI protocol. ARNO is written in PERL and is structured around a
document store and metadata database. Long term digital preservation is not cited as a
key aim of this development. ARNO is open source.

8.2 CDSware (CERN Document Server Software)

Project home : http://cdsware.cern.ch/
Example installation : http://cds.cern.ch/
Core purpose : document storage and dissemination
CDSware was developed to store and disseminate scholarly publishing from CERN as
well as providing a unified, portal like interface to existing repositories and information
sources. The software is a Python/PHP implementation incorporating a MySQL database
server and provides support for OAI metadata harvesting. Long term digital preservation
is not cited as a key aim of this development. CDSware is open source.

8.3 DSpace

Project home : http://dspace.org/index.html

Example installation : https://hpds1.mit.edu/index.jsp
Core purpose : storage, preservation and access to a wide range of institutionally derived
materials.
DSpace was developed by MIT and Hewlett Packard to address the preservation and
dissemination needs of MIT. DSpace has at its core a model which attempts to address
the real needs of a flexible institutional repository. This Communities and Collections
structure aims to allow devolved control and administration from a central service to
individual departments. DSpace is written in JAVA and utilises a PostGreSQL database
layer. Long term digital preservation is a key aim of the DSpace system and MIT (and
others) are investing considerable research in enhancing this aspect of the respository
software. Version 2 of DSpace is expected to offer a more modular design and extended
support for digital preservation functions. DSpace is open source.
8.4 ePrints
Project home : http://software.eprints.org/
Example installation : http://libeprints.open.ac.uk/
Core purpose : document storage and dissemination with flexible support for different
format types
ePrints was developed by the University of Southampton with the aim of enhancing open
access to scholarly materials. The software was developed in PERL and manages a
MySQL database layer. It has now reached version 2. Uptake of the ePrints software has
been considerable. Long term digital preservation is not cited as a key aim of this
development. ePrints is open source.

8.5 FEDORA (Flexible and Extensible Digital Object and Repository Architecture)
Project home : http://www.fedora.info/
Core purpose : storage and dissemination with flexible support for different uses
FEDORA is a comprehensive repository and digital library system developed from the
FEDORA architecture at Cornell University and the University of Virginia. FEDORA is
currently being tested by a variety of institutions across the US and UK including the
Library of Congress. The software is implemented in JAVA and the system relies on a
range of developing standards including SOAP and METS. Long term digital
preservation is not cited as an initial aim of this development but technology watch
functions have been mentioned as development goals for new versions of the software. A
related project PRISM is investigating digital preservation and is utilising the
FEDORA architecture. FEDORA is open source.

8.6 MyCoRe
Project home : http://www.mycore.de/engl/index.html
Core purpose : storage and dissemination with flexible support for different uses

MyCore is a flexible repository system designed by a group of German universities

working in collaboration. MyCore is written in JAVA and incorporates a flexible
database layer allowing different database backends to be used with it (see Overall
Repository Structure, below). Long term digital preservation is not cited as a key aim of
this development. MyCore is open source.

9.0 Future and Related Developments

A range of institutions are involved in the development and maintenance of digital
repositories which have a relevance to this discussion. Two key current developments are
described below.
9.1 UK National Archives
Project home : http://www.pro.gov.uk/about/preservation/digital/archive/default.htm
The UK National Archives has recently developed a digital repository system, the Digital
Archive, which is planned for an open source release. Digital preservation is a key aim of
this development. Further details, including links to articles about the Digital Archive and
the other work of the TNA Digital Preservation Department, can be found at the TNA
website above.

9.2 DAITSS (Dark Archive In The Sunshine State)

Project home : http://www.fcla.edu/
DAITSS is an open source digital repository system designed and implemented by the
Florida Centre for Library Automation. The Centre is developing the software to support
over 50 public and university libraries in Florida. DAITSS is written in Java and
incorporates a DB2 database layer. Digital preservation is a key aim of the system and an
initial release is expected in the near future.

10.0 Current support for digital preservation in institutional

repository software
The OSI guide to institutional repository software [2] lists only DSpace and CDSware as
having a "defined digital preservation strategy". The CDSware strategy is reliant on
migration on ingest to an archival format, in this case Adobe's PDF format. DSpace goes
much further in defining categories of "Known", "Supported" and "Unsupported" formats
[27]. Instances of DSpace can give different levels of commitment to preserving formats
depending on the category within which they fall. Currently research and development
work in a number of areas (in collaboration with Cambridge University and others) is
being conducted with the aim of meeting these preservation commitments [28]. The
current version of DSpace supports only very basic Representation Information including

minimal information like the deposited object's MIME type. While this is certainly a
starting point, MIT are quick to acknowledge that this is not adequate for the purposes of
long term digital preservation. Again, MIT is concerned about this issue and with
Harvard University and the Digital Library Federation is leading the Global File Format
Registry initiative to address the issue of Representation Information.
Digital preservation is not currently addressed as a key aim of the other repository
software listed above. Clearly the provision of support for digital preservation in
institutional repository software is at a very early stage. The key for current repository
software is to provide flexible and extensible designs that can adapt to take advantage of
digital preservation developments as they become available. As long as the main digital
preservation issues described above are understood, this should be possible.
Addressing all aspects of digital preservation at the repository level is unlikely to be
achievable due to the scale of the task at hand, and a degree of cooperation, sharing and
external support will be required. In recognising this need, the JISC and eSCP are at the
time of writing engaged in the establishment of a Digital Curation Centre that aims to
provide support to existing digital repositories [15].

11.0 Recommendations
The key recommendations from this report are for the continued development of specific
requirements for trusted digital repositories, and also for the creation of independent
certification services for digital repositories that will evaluate how repositories meet these
requirements. A clearer picture can then be presented as to how well institutional
repository software, as well as specific digital repositories, can deliver effective digital
preservation.
The report also makes the following recommendations:
Preservation functions require integration with institutional repository design and
must be considered from the outset both in the development of repository software
and in the establishment of a given repository.
Digital preservation developments are at an early stage in many areas so where
possible, developments in institutional repository software should be made as
modular, flexible and extensible as possible to allow integration with digital
preservation developments as they become available. If an element of fore thought is
given to the demands of digital preservation as described above, this process can be
considerably simplified.
Careful consideration must be given to the preservation needs of materials to be
archived within an institutional repository. Very good reasons must be identified for
not addressing digital preservation.
Community wide efforts must be invested in developing the solutions to identified
requirements for digital preservation in a repository context. The following areas are
considered to be crucial:
Ingest

Representation Systems
Rendering
Where possible concentrate development on distributed preservation functions which
offer community wide sharing, and community based ownership and maintenance.
Continue to build on the OAIS model (particularly with respect to Representation
Networks, the value of which has been ignored in many sectors).

12.0 Glossary
Technology obsolescence : Where current hardware and software is superseded by new
technology, which may not be compatible with older systems. This can lead to the loss of
the ability to make sense of or render (see below) data.
Media obsolescence : Where storage media is superseded by newer media. Note that
although much emphasis is often placed on the readable lifetime of digital media, it is
almost always the obsolescence of the hardware that reads the media that prevents access
to the data (see above). For example, the videodiscs upon which the BBC Domesday
Project data are stored were designed to last a hundred years, but the special LVROM
readers which read the discs have not been manufactured or supported for over a decade
(and surviving units are now prone to breaking down). Long lived media does not equal
long lived preservation.
Digital preservation : An organised series of actions taken to ensure continued use of
digital objects is possible over time. The key elements of the solution include ensuring
digital objects are : never lost or damaged, can always be found, and can always be
understood. For example, placing a digital object into a repository where it will be backed
up to prevent loss, where it will be given a unique identifier so it can always be found and
where it will be linked to Representation Information which will describe how it can be
rendered.
Representation Information : Metadata which describes how the bytestream of a digital
object can be turned from a meaningless series of numbers into a human readable
representation. This could include a simple textual description of the type of data in
question, a detailed breakdown of a specific file format or a description of the tools which
render that format.
Rendering : The process of displaying a digital object in a human readable way. For
example, using WordViewer to display a Microsoft Word file, or running a BBC Micro
emulator to render the BBC Domesday Project software.
Technology Watch : The monitoring of software and hardware dependencies to ensure
that when technology obsolescence occurs, appropriate action is taken to update the
relevant Representation Information and associated Rendering process

12.0 References
[1] The Case for Institutional Repositories: A SPARC Position Paper, Crow, R,
http://www.arl.org/sparc/IR/IR_Final_Release_102.pdf
[2] A Guide to Institutional Repository Software, Open Society Institute,
http://www.soros.org/openaccess/software/
[3] DAEDALUS: Initial experiences with EPrints and DSpace at the University of
Glasgow, Nixon, W, Ariadne, issue 37, http://www.ariadne.ac.uk/issue37/nixon/
[4] Institutional Repositories: Essential Infrastructure for Scholarship in the Digital
Age, Lynch, C, A, http://www.arl.org/newsltr/226/ir.html
[5] "The Open Archival Information System Reference Model: Introductory Guide",
Lavoie, B, http://www.dpconline.org/graphics/reports/index.html#intoais
[6] The Digital Preservation of e-Prints, Pinfield, S, James, H,
http://www.dlib.org/dlib/september03/pinfield/09pinfield.html
[7] Cedars : collection management, http://www.leeds.ac.uk/cedars/colman/colman.html
[8] OAIS, http://ssdoo.gsfc.nasa.gov/nost/isoas/
[9] Trusted Digital Repositories: Attributes and Responsibilities, RLG/OCLC ,
http://www.rlg.org/longterm/repositories.pdf and the follow up work undertaken by the
Task Force on Digital Repository Certification
http://www.rlg.ac.uk/longterm/certification.html
[10] The Medium is not the message, Holdsworth, D,
http://www.personal.leeds.ac.uk/~ecldh/paper2.html
[11] The Handbook, Beagrie, N, Jones, M, J,
http://www.dpconline.org/graphics/handbook/
[12] Migration on Request, University of Leeds,
http://www.leeds.ac.uk/reprend/migreq/migreq.html
[13] Emulation, http://www.nla.gov.au/padi/topics/19.html
[14] Digital Preservation Coalition, http://www.dpconline.org/graphics/index.html
[15] Digital Curation Centre, http://www.ucs.ed.ac.uk/bits/2004/february_2004/
[16] Persistent identifiers, http://www.nla.gov.au/padi/topics/36.html

[17] METS, http://www.loc.gov/standards/mets/

[18] Metadata Standards Framework : Preservation Metadata,
http://www.natlib.govt.nz/files/4initiatives_metaschema_revised.pdf
[19] Cedars (technical), http://www.leeds.ac.uk/cedars/archive/archive.html
[20] PRONOM, UK National Archives, http://www.records.pro.gov.uk/pronom/
[21] Global Digital Format Registry (GDFR),
http://www.erpanet.org/www/products/Rome/Dale_registry_pres.pdf
[22] A Blueprint for Representation Information in the OAIS Model", Holdsworth, D,
Sergeant, D, http://esdis-it.gsfc.nasa.gov/MSST/conf2000/PAPERS/D02PA.PDF
[23] Survey and assessment of sources of information on file formats and software
documentation, Wheatley, P,
http://www.jisc.ac.uk/uploaded_documents/FileFormatsreport.pdf
[24] CAMiLEON, http://www.si.umich.edu/CAMILEON/
[25] Testbed Digitatle Bewaring,
http://www.digitaleduurzaamheid.nl/index.cfm?paginakeuze=181&categorie=2
[26] Preservation Metadata : Pragmatic First Steps at the National Library of New
Zealand, Searle, S Thopmson, D,
http://www.dlib.org/dlib/april03/thompson/04thompson.html
[27] DSpace supported formats, https://dspace.mit.edu/help/formats.html
[28] A way forward for developments in the digital preservation functions of DSpace :
options, issues and recommendations, Wheatley, P,
http://dspace.org/news/articles/DpAndDSpace.pdf