Security by architecture, then by controls

The hosted service never needs your rows, so it never receives them.

The strongest control is the one that makes a leak structurally impossible. DataNivra is designed so the hosted service never needs your rows.

The agent connects out; nothing connects inInside your network the DataNivra agent reads sources read-only and resolves credentials from your own secret store. It opens outbound HTTPS connections to DataNivra Cloud, which receives secret references, never values. There are no inbound connections into your network.Your networkSecret storekeys and credentialsSourcesread-onlyDataNivra agentfails closedoutbound HTTPS onlyDataNivra Cloudreferences onlyno inbound connections

Design principles

  • Customer-resident processing. Discovery, profiling, subsetting, masking, synthesis, validation, certification and provisioning run in your environment. The control-plane code base does not even contain the row-processing engine.
  • Zero raw-production-data egress. Every field that crosses the boundary is classified as control metadata, an aggregate metric, an evidence reference or a secret reference. Messages containing anything else are rejected on both sides.
  • Outbound-only connectivity. The agent initiates every connection over HTTPS. No inbound ports, no VPN peering, no database exposure.
  • Secret references, not secrets. Source credentials and masking keys stay in your secret store; the control plane only ever sees names such as a vault path.
  • Fail closed. Uncertain privacy or integrity state stops the job. Failed or revoked datasets are never provisioned.
  • Tamper-evident audit. Privileged actions produce hash-chained audit events; certification evidence is kept customer-side with checksums.
  • Tenant isolation. Every API request is scoped to a tenant; one tenant cannot enumerate or infer another.

Your data centre or cloud account is where rows are read, transformed and stored. That does not change when you use DataNivra.

How raw data is kept from leavingInside your environment, engine output stays local; a report builder produces metadata; the EgressGuard checks every outbound message against allow-listed schemas and content detectors. Allowed out: control metadata, aggregate metrics, evidence references and secret references. Blocked: rows, samples and credentials. In the cloud, ingress validation rejects anything non-conforming before it is stored.Your environmentEngine output (rowsstay local)Report builder(metadata only)EgressGuard: schema +content checksDataNivra CloudIngress validationMetadata store & auditAllowed: metadata, aggregates, evidence & secret referencesBlocked: rows, samples, credentials
How raw data is kept from leaving. Inside your environment, engine output stays local; a report builder produces metadata; the EgressGuard checks every outbound message against allow-listed schemas and content detectors. Allowed out: control metadata, aggregate metrics, evidence references and secret references. Blocked: rows, samples and credentials. In the cloud, ingress validation rejects anything non-conforming before it is stored.
Text description
  1. Your environment: Engine output (rows stay local) → Report builder (metadata only) → EgressGuard: schema + content checks

    Connection: Allowed: metadata, aggregates, evidence & secret references — Blocked: rows, samples, credentials

  2. DataNivra Cloud: Ingress validation → Metadata store & audit

What we do not claim

DataNivra helps you run privacy and governance programmes — for example work related to HIPAA, GDPR or PCI DSS — by keeping production data in your environment and producing evidence. Using a tool does not make an organisation compliant, and masking alone is not proof of anonymization. We do not display certification badges we have not earned.

SOC 2 readiness

DataNivra is designed with SOC 2 Type II readiness in mind. The controls below map to the Trust Services Criteria most relevant to a test-data management platform:

  • CC6.1 — Logical access. Tenant isolation is enforced server-side by row-level security; every API request is scoped to one tenant. Users are provisioned with named roles; workload identity replaces service-account passwords.
  • CC6.3 — Least privilege. The read-only source account principle is enforced during discovery and job runs. The agent uid (10001) has no shell and no write access to anything outside its designated directories.
  • CC7.2 — Anomaly detection. Every privileged state change produces a hash-chained audit event. ComplianceGuard scans run continuously for sensitive-field drift and policy violations, with configurable Slack / webhook alerts.
  • CC9.2 — Vendor risk. Customer data rows never reach the hosted service (see Architecture); the principal risk of a control-plane breach is metadata exposure, not data exposure.
  • A1.2 — Availability. The agent polls the control plane; the control plane going offline pauses new jobs but does not revoke provisioned datasets or affect running applications.

A formal SOC 2 Type II audit report will be published here when completed. Contact the enterprise team to request the current evidence package or to discuss timeline.

Operational controls

  • Encryption in transit. All connections between the agent and the control plane use TLS 1.2 or later; the agent verifies the certificate against the system trust store.
  • Encryption at rest. Control-plane state is stored in an encrypted database (AES-256). Secret references (the only secret-shaped values the control plane holds) are encrypted as individual columns.
  • Privileged-access controls. Production environments require MFA. Administrative access is logged to a tamper-evident audit chain viewable by the tenant.
  • Vulnerability disclosure. We publish a /.well-known/security.txt and welcome responsible disclosure.

Compliance readiness

DataNivra is designed to support your HIPAA, GDPR and PCI DSS programmes by keeping production data in your environment and producing evidence. Using DataNivra does not make an organisation compliant; that requires your own controls and audits.

We are working toward SOC 2 Type II and ISO 27001 certification. These are in progress and have not yet been issued. Prospective customers who need a current security posture review before those certifications complete should contact us for our security review package (architecture diagrams, threat model, pen-test scope and results, privacy boundary register and evidence samples).

Report a vulnerability

Email security@datanivra.com (also listed in our /.well-known/security.txt). Please do not include customer data, and give us a reasonable time to fix an issue before disclosing it.

Want to check our claims yourself? See trust and verification. For what the controls do not cover, see known limitations and residual risks.