For developers
A realistic database on your laptop, without a production dump
Hand-written fixtures cover the happy path; production dumps cover everything and leak it. DataNivra gives you relational, realistic synthetic data you can load locally and in CI today, and the same workflow on masked data from your own systems when you need it.
The problem
Most bugs that reach staging were invisible on a developer database with ten tidy rows.
- Fixtures drift from the schema and never contain the awkward cases: nulls, long names, cancelled orders, overdue invoices.
- Loading realistic data fails on the first foreign key when tables go in alphabetical order.
- Asking for "a copy of prod" takes a ticket, a week and an exception nobody wants to sign.
Live sample — synthetic, runnable now
These rows come from the published General Enterprise bundle: invented records, generated from a fixed seed, with every key intact.
customer_id | first_name | last_name | city | created_on |
|---|---|---|---|---|
| CUS00000001 | Tamara | Wrenton | Jorvport | 2019-07-08 |
| CUS00000002 | Celis | Ashvane | Aldrenvale | 2020-09-26 |
Masked by the real engine with a fixed public sample key: CUS00000001 becomes CUS18551589 in both crm.customers.customer_id and commerce.orders.customer_ref, so the join still holds after masking.
Run it yourself
Fetch the CRM customers of the General Enterprise bundle (public, synthetic) and look at the first rows. No account needed.
BASE=https://www.datanivra.com/downloads/packs/general-enterprise/2.0.0
curl -fsSLO "$BASE/synthetic/crm.customers.csv"
cut -d, -f1-3 crm.customers.csv | head -4Expected output:
customer_id,first_name,last_name
CUS00000001,Tamara,Wrenton
CUS00000002,Celis,Ashvane
CUS00000003,Pelo,QuarrshawFull tested example: A synthetic PostgreSQL test database in foreign-key order. Or open the synthetic playground for this pack — no account.
The outcome
- A multi-schema database (orders, customers, invoices, employees, suppliers) that loads with every declared foreign key intact.
- The same seed in every checkout, so a failing test reproduces on your machine and in CI.
- Named scenarios such as backorders, late shipments and overdue receivables instead of happy-path rows.
- When synthetic data is not enough, request a masked subset of your own systems through the same workflow.
Security: production data stays home
Runs in your environment Row-level work happens only in your environment.
- Synthetic bundles contain invented people and companies only; every row carries a SYNTHETIC provenance marker.
- On your own data, discovery, subsetting and masking run in the agent inside your network; DataNivra Cloud receives metadata and evidence, never rows.
- Masking keys stay in your secret store; DataNivra holds only a reference to them.
Customer-resident architecture · Verify it yourself · Security model
Your first action
- Open the synthetic playground and preview the General Enterprise pack — no account.
- Run the PostgreSQL scenario to build a load script for your local database.
- Start free when you want named datasets per branch and the console workflow.
Other teams: QA automation · Data engineer · Platform engineering · Healthcare · Finance · Insurance · Manufacturing · All teams