What is an institutional repository? Definition, contents and how to start one

What is an institutional repository?

An institutional repository is a system that gathers, preserves and provides open access to the academic output of the institution itself: theses, articles, conference papers, research data and teaching materials. Unlike a catalog, which describes books and resources the library owns or subscribes to, the repository publishes documents that the institution produced. Depositing a version of a paper in a repository is also what the open-access movement calls the "green" route to open access, as opposed to publishing in an open-access journal.

The most widely used software for this purpose is DSpace: open source, governed by its community with LYRASIS as organizational home, and used by more than 3,000 organizations according to dspace.org. Our guide to what DSpace is covers its history and architecture.

Repositories come in a few flavors:

• Institutional: general in scope and organized by faculties or academic units, receiving all of the institution's own output. This is the most common model.
• Subject or thematic: centered on one area of knowledge, the choice of some specialized institutions such as research centers and observatories.
• Research data: specialized in datasets, with platforms such as Dataverse.

DSpace supports both the institutional and the subject model through its structure of communities and collections, which is defined during setup according to how the institution is organized, not according to a fixed template.

Institutional repository vs. library catalog

An academic library ends up needing both, not one instead of the other: the catalog manages what the library lends, the repository manages what the institution produces. When BiblioLabTech implements both at once, as at the Escuela Colombiana de Ingeniería Julio Garavito, with Koha for the catalog and DSpace for the repository, the goal is that users do not need to know in advance which of the two to search. For the catalog side, see our guide to the integrated library system.

What an institutional repository usually holds

• Theses and dissertations, which in many countries must be deposited under institutional or national rules.
• Journal articles, in the open-access version the publisher allows: preprint, accepted manuscript or published version, depending on each journal's policy. Services such as Sherpa Romeo summarize publisher policies on self-archiving.
• Conference papers and event materials produced by the institution.
• Research data, when the institution adopts an open data policy.
• Teaching materials, if the institution decides to publish them as open educational resources.

How deposit, review and access control work

A repository is not a shared folder: content enters through a controlled workflow. In DSpace, the workflow is defined per collection. Most often the author or the academic office submits the item, and the library reviews the metadata and the license before publication, with one or two approval steps. DSpace records who performed each action and lets reviewers return a submission with comments, so bibliographic quality control is documented rather than informal.

Access is just as granular. Permissions are set by community, collection, item and file, and assigned to user groups. A common practice is to publish the metadata record openly while keeping the full file restricted or under a temporary embargo, for example while a thesis is part of a patent process or awaiting journal publication. The record stays discoverable and citable, and the file becomes available once the embargo period ends.

Standards and identifiers that make content findable and citable

A serious repository does not just store files. It identifies them persistently, so they remain citable even if the URL changes, and it describes them in formats that other systems understand.

Exposing OAI-PMH is what puts a repository on the map. Aggregators such as OpenAIRE in Europe and LA Referencia in Latin America, and global services such as BASE and CORE, harvest repositories this way, and Google Scholar indexes repository content when item pages carry the metadata tags it expects. Indexing is not immediate: the first items usually appear within weeks, and full coverage takes a few months.

Open-access mandates: why repositories matter for compliance

Funders and institutions increasingly require that publicly funded research be openly available, and repositories are one of the main routes to compliance. Plan S, the initiative of the cOAlition S funders, sets explicit technical requirements for open-access repositories:

• Registration in OpenDOAR, the directory of open access repositories.
• Persistent identifiers for deposited versions, with DOI preferred and URN or Handle also accepted.
• High-quality article-level metadata in a standard, non-proprietary format, under a CC0 public domain dedication.
• Machine-readable information on open-access status and license.
• Uptime of at least 99.7%, excluding scheduled maintenance.

Most of these depend on configuration and hosting rather than on the software itself: identifiers, metadata exposure, license fields and uptime are decided when the repository is set up. Whether a given deposit complies still depends on the version and license the author is allowed to use.

When a simple repository is not enough: DSpace-CRIS

A classic institutional repository answers "where is this document?". Some institutions also need to answer "who is this researcher, which projects do they work on and what have they produced?", which is a different question and requires a current research information system (CRIS). DSpace-CRIS adds that layer on top of the same repository, with ORCID, ROR, OpenAlex and DataCite integrations. Since DSpace 10.0, released in May 2026, the first DSpace-CRIS capabilities have been merging into the DSpace core itself.

What your institution needs to get started

• Accumulated output. Theses, articles and other material that exists today as scattered files rather than as a published collection.
• A deposit policy. Who must deposit, in which format and under which open-access license.
• Infrastructure. Your own server or hosting managed by an external provider.
• A metadata model. Communities, collections and mandatory fields, defined before the first document is loaded.

In BiblioLabTech's projects, the most common risk in a new repository is the initial loading debt: if describing each accumulated document requires manual transcription, the repository launches with a handful of items and stays there. That is why we integrate MetaExtract API into the deposit workflow. It fills in descriptive metadata from the PDF, so library staff review instead of retyping; the approach is explained in our guide to AI metadata extraction.

Digital preservation: publishing is only half of the job

Publishing a document today does not guarantee that it will still be accessible in ten years if the file format becomes obsolete or the server it lives on disappears. DSpace stores the original file exactly as deposited and computes checksums to detect file corruption. On top of that come periodic backups with a copy off the main server, and real restore tests rather than copies nobody checks. Digital preservation is as much institutional policy, meaning which formats are accepted and how often fixity is verified, as it is technology.

Self-hosted or cloud-hosted

A repository can run on the institution's own servers or on infrastructure managed by a third party. Universidad Americana in Paraguay chose a DSpace repository hosted in the cloud and operated by BiblioLabTech, without needing servers or operations staff of its own (read the case study). Our DSpace hosting guide compares both options criterion by criterion.

What an institutional repository costs

DSpace is free software with no license fee; what you pay for is implementation, content migration and, where applicable, managed hosting. A new single-campus repository without DSpace-CRIS typically costs between USD 3,000 and 5,000 in its first year, as a reference rather than a fixed price. The full breakdown is in our guide to DSpace implementation cost.

Getting started, wherever you are

BiblioLabTech implements institutional repositories remotely for institutions in any country, always on DSpace, because it is the most widely adopted standard for this kind of project and guarantees that the institution never depends on a single provider to access its own data. A typical implementation takes three to four months, from defining the information model to opening the repository to the public; migrating tens of thousands of items, or source metadata that needs heavy normalization, can extend the project to five or six months. Through our DSpace implementation service we scope the repository with you: structure, policies, hosting model and the initial content load.