Secure Data Enclave

A controlled-access computing environment in which approved researchers analyze restricted or sensitive data in place, without the ability to download, copy, or otherwise export the raw underlying records. Access is typically granted only after training, credentialing, and a signed data use agreement, computation happens on infrastructure the data steward controls (an on-site terminal room, a remote-access session, or an isolated cloud workspace), and only aggregate or disclosure-reviewed outputs — tables, statistics, model results — are allowed to leave the environment. The defining feature is not where the data sits but that raw data never crosses the boundary; only vetted outputs do.

ByCASRAI Editorial Board

· Last updated 18 Jul 2026

Worked examples

Is an instance

A researcher accesses Census Bureau microdata through a Federal Statistical Research Data Center's Secure Standalone Enclave, running all analysis on-site with no ability to copy files out; only disclosure-reviewed output tables are released.

Is an instance

A university-hosted virtual data enclave (VDE) lets an approved research team analyze a licensed, restricted-use administrative dataset via a remote virtual desktop that blocks file downloads and clipboard export.

Looks similar, but isn't

Not an instance

Downloading a de-identified data extract to a personal laptop under a signed data use agreement is not a secure data enclave, since the raw/extracted records leave the steward's controlled environment.

Not an instance

An open-access institutional repository that allows anyone to download the full dataset file is a repository with open download, not an enclave.

A secure data enclave (also called a virtual data enclave, VDE, or secure computing enclave) is a controlled-access computing environment built to let approved researchers analyze restricted or sensitive data without ever removing the raw data from that environment. Instead of shipping a dataset to a researcher’s own laptop, the researcher comes to the data: they log into a monitored workspace, run their analysis there, and take away only the outputs that a disclosure review has cleared for release — typically aggregate statistics, tables, figures, or trained model parameters, never row-level records.

Enclaves exist because some research data cannot be shared as an open or even a standard restricted-access download without an unacceptable re-identification or confidentiality risk — government administrative records, detailed health records, or licensed data governed by a strict data use agreement are common cases. Rather than refusing access outright, the data steward keeps custody of the raw file and instead grants supervised, in-place computation.

What makes an environment a secure data enclave

No raw data export. Researchers cannot copy, download, email, or print the underlying microdata. Some enclaves are physically air-gapped terminal rooms with disabled USB ports and no internet access; others are remote-access or cloud-hosted sessions where export is blocked at the platform level.
Credentialed, individual access. Users typically must complete data-security or human-subjects training, be named on an approved project, and sign a data use agreement or equivalent before an account is issued. Access is usually project- and person-specific, not transferable.
Output disclosure review. Before any result leaves the enclave — a regression table, a chart, a summary statistic — it passes a review step (automated, human, or both) checking that it doesn’t reveal information about an individual record, such as a cell count too small to protect confidentiality.
Steward-controlled infrastructure. The computing environment itself — hardware, storage, and often the analytical software available — is provisioned and monitored by the organization that holds the data, not by the individual researcher.

Real examples

Several well-established programs illustrate the model:

The U.S. Census Bureau’s Federal Statistical Research Data Centers (FSRDCs) give qualified researchers access to restricted federal statistical microdata through a Secure Standalone Enclave (a closed, non-networked room with no printing or transfer capability) or a remote Regulated Research Environment. All research results are subject to a federal disclosure review before release to the public.
NIH’s National Center for Advancing Translational Sciences hosts the N3C Enclave for the National COVID Cohort Collaborative: a FedRAMP-certified, cloud-based environment where approved investigators analyze linked clinical data entirely within the platform — no data may be downloaded, and NIH administrators control access down to the row or column level.
NIH’s National Institute on Aging similarly runs the LINKAGE Program Enclave, a FISMA-certified cloud statistical environment for analyzing linked survey and administrative datasets in place.
NORC at the University of Chicago operates the NORC Data Enclave, a multi-tenant secure virtual research environment launched in 2006 in partnership with the National Institute of Standards and Technology (NIST) — one of the first platforms of its kind. Rather than serving a single research program, it hosts restricted-use data on behalf of many different federal and state agencies under one shared infrastructure, including the Bureau of Labor Statistics Virtual Data Enclave, the USDA Economic Research Service Secure Data Enclave, the NSF Secure Data Access Facility, and the California Cradle-to-Career Data Enclave. More than 1,000 provisioned users across 100+ public and private entities access approved datasets there under role-based, credentialed accounts, with all outputs subject to the same no-raw-export and disclosure-review model described above.
Many universities operate their own institutionally hosted virtual data enclaves (VDEs) so researchers can work with restricted-use datasets — licensed commercial data, sensitive administrative records, or data covered by a funder’s or provider’s data use agreement — under the institution’s own access controls rather than a laptop copy.

These are distinct implementations of the same underlying pattern, and the specific technical architecture (physical terminal room vs. remote virtual desktop vs. cloud workspace) varies by steward and by how sensitive the data is judged to be.

Secure data enclave vs. adjacent concepts

A secure data enclave is the infrastructure that enforces restricted access; it is not the same thing as the legal agreement that authorizes it or the repository that may host the data outside an enclave context. A data use agreement is the contract a researcher signs to be granted enclave access in the first place — the enclave is how that access is technically enforced once the agreement is in place. A sensitive-data repository is the broader category of a repository that holds restricted or sensitive datasets; some sensitive-data repositories offer enclave-based access as one access tier alongside others (such as fully open or credentialed-download tiers), while a plain trusted digital repository certification (for example CoreTrustSeal) speaks to a repository’s overall preservation and governance practices and doesn’t by itself imply enclave-style access controls exist. For the contractual side of restricted-data transfers more broadly, see CASRAI’s guide on data sharing agreements between collaborators and institutions.

Counter-example

Downloading a de-identified extract of a licensed dataset onto a personal laptop, even under a signed data use agreement, is not a secure data enclave — the moment raw or extracted records leave the steward’s controlled environment and reach the researcher’s own device, the defining no-export property is gone. Likewise, a standard open-access institutional repository that lets anyone download a full dataset file is not an enclave; it’s an open- or restricted-download repository, a different access model entirely.

References

U.S. Census Bureau, ‘Federal Statistical Research Data Centers,’ census.gov/about/adrm/fsrdc.html.
National Center for Advancing Translational Sciences, ‘N3C Enclave Data Overview,’ ncats.nih.gov.
National Institute on Aging, ‘LINKAGE Program Enclave Features and Security,’ nia.nih.gov.
NORC at the University of Chicago, ‘NORC Data Enclave,’ norc.org/services-solutions/data-enclave.html.

Use in your systems

JATS XML <role> element

xml

<role vocab="credit"
      vocab-identifier="https://casrai.org/dictionary/"
      vocab-term="Secure Data Enclave"
      vocab-term-identifier="https://casrai.org/dictionary/term/secure-data-enclave" />

Schema.org DefinedTerm (JSON-LD)

json

{
  "@context": "https://schema.org",
  "@type": "DefinedTerm",
  "@id": "https://casrai.org/dictionary/term/secure-data-enclave",
  "name": "Secure Data Enclave",
  "identifier": "https://casrai.org/dictionary/term/secure-data-enclave",
  "description": "A controlled-access computing environment in which approved researchers analyze restricted or sensitive data in place, without the ability to download, copy, or otherwise export the raw underlying records. Access is typically granted only after training, credentialing, and a signed data use agreement, computation happens on infrastructure the data steward controls (an on-site terminal room, a remote-access session, or an isolated cloud workspace), and only aggregate or disclosure-reviewed outputs — tables, statistics, model results — are allowed to leave the environment. The defining feature is not where the data sits but that raw data never crosses the boundary; only vetted outputs do.",
  "inDefinedTermSet": "https://casrai.org/dictionary/domain/data-infrastructure#set",
  "url": "https://casrai.org/dictionary/term/secure-data-enclave",
  "sameAs": [],
  "license": "https://creativecommons.org/licenses/by/4.0/",
  "publisher": {
    "@id": "https://casrai.org/#organization"
  },
  "dateModified": "2026-07-18T06:30:50",
  "inLanguage": "en"
}