01 / Empty staging
Nothing to replicate
A new pipeline in a test environment has no rows and no changes. CDC, mappings and targets go live untested, or against a copy of production nobody should have.
Gluesync Module · Whisperer · Ships with Gluesync 2.3.1
Synthetics and seeding
Generate realistic mock data on the fly, keep seeding the tables your pipelines replicate, and export ORM-ready stub data without touching a database. Whisperer is the Gluesync module behind it, built into Core Hub from Gluesync 2.3.1.
Synthetics and seeding
Synthetics and seeding
| Bound table | Mix | Rate | Status | |
|---|---|---|---|---|
| sales.ordersP-PG-MYSQL Orders · PostgreSQL source | 70 / 20 / 10ins / upd / del |
5 × 1 sUnlimited · no time limit | Idle | |
| sales.customersP-PG-MYSQL Orders · PostgreSQL source | 100 / 0 / 0ins / upd / del |
100 × 0.1 sUnlimited · no time limit | Idle | |
| acme.stockP-MYSQL-PG Inventory · MySQL source | 33 / 33 / 34ins / upd / del |
10 × 2 s2,000 ops · 10 m 0 s | Idle |
- ONE_SHOTsales.customers500 rows · 2.1 sCOMPLETED
- CONTINUOUSacme.stock10 m 0 s · 2,000 ops · duration limitCOMPLETED
- 01 Bind a table on a pipeline you already run
- 02 Seed it once, or keep a pillow fight going
- 03 Export fixtures with no database at all
Why Synthetics and seeding
Give the pipeline traffic before production does
02 / Scripts and spreadsheets
Hand-rolled load
Ad-hoc INSERT loops and CSVs ignore keys and types, flood query history, and stop whenever the laptop running them does.
03 / The platform outcome
One module, three outputs
Whisperer binds to the agents Gluesync already knows, writes type- and key-aware rows at the rate you set, and exports the same generated data as fixtures.
How it works
Built into Core Hub, run through Query Studio
- 01
Bind
Pick a table on an agent
Whisperer binds to pipelines and agents Gluesync already manages. There is no second connection store and no extra credentials to protect.
- 02
Generate
Type- and key-aware rows
Generators are inferred from column names and types. Primary keys stay unique, foreign keys are sampled from the parent table, and nullable columns get the occasional NULL.
- 03
Run
Once, or continuously
A fixed batch of rows, or an unbounded pillow fight with an exact INSERT / UPDATE / DELETE split, a delay between batches, and operation or duration limits.
Capabilities
Three ways to produce data
The same generator sits behind every output: a table you create for the purpose, a continuous workload on a table you already replicate, or a file with no database involved. Whisperer started as a standalone open-source tool; from Gluesync 2.3.1 it runs inside Core Hub under the Control Plane's roles.
Generate mock data on the fly
Realistic rows from your schema
Column-aware generators, per-column overrides, and tables created for the purpose
Point Whisperer at an existing table, or let it create one from a single CREATE TABLE statement or by copying another table's DDL on the same dialect. The preview shows the exact statement and the generator each column will get before anything runs.
- Names, emails, phones, addresses, companies, job titles and descriptions from column names; booleans, integers, decimals, dates and timestamps from column types
- Per-column overrides: a faker hint, exclusion from writes, or forced NULL on or off
- Create a table from SQL or copy an existing one; preview the DDL without executing it
- Non-blocking warnings, such as a missing primary key that would make UPDATE and DELETE no-ops
Synthetics and seeding / Create table
Create table and bind it
CREATE TABLE sales.orders_synthetic (order_id BIGINT PRIMARY KEY,customer_email VARCHAR(120),customer_name VARCHAR(80),status VARCHAR(12),amount NUMERIC(12,2),created_at TIMESTAMP NOT NULL);
Parsed sales.orders_synthetic · 6 columns · primary key order_id
| Column | Generator |
|---|---|
order_id | sequence |
customer_email | internet.email |
customer_name | name.fullName |
status | lorem.word |
amount | decimal 1–1000 |
created_at | date.past 3y |
Generators are inferred from column names and types. Override any column with a faker hint, exclude it from writes, or force NULLs.
Continuously seed data to database tables
Pillow fights
Continuous INSERT, UPDATE and DELETE load with an exact operation mix
A pillow fight keeps writing to a bound table until you stop it or a limit is reached. Set the split per 100 operations, the batch size and the delay, then watch the counters move while CDC picks every change up.
- Operation mix as an exact split, for example 70 / 20 / 10, or a single operation type
- Batch size, delay between batches, maximum operations and maximum duration; limits end the run as COMPLETED
- Run once with a fixed row count, or stop a continuous run and keep it recorded as STOPPED
- Cascade delete for foreign-key constraints; a batch stuck on a dead connection is cancelled after 30 seconds so Stop always works
- Every statement goes through Query Studio's SafetyGate without filling its history or audit; failures reach the Core Hub notification bell
Synthetics and seeding
Synthetics and seeding
| Bound table | Mix | Rate | Status | |
|---|---|---|---|---|
| sales.ordersP-PG-MYSQL Orders · PostgreSQL source | 70 / 20 / 10ins / upd / del |
5 × 1 sUnlimited · no time limit | Idle | |
| sales.customersP-PG-MYSQL Orders · PostgreSQL source | 100 / 0 / 0ins / upd / del |
100 × 0.1 sUnlimited · no time limit | Idle | |
| acme.stockP-MYSQL-PG Inventory · MySQL source | 33 / 33 / 34ins / upd / del |
10 × 2 s2,000 ops · 10 m 0 s | Idle |
- ONE_SHOTsales.customers500 rows · 2.1 sCOMPLETED
- CONTINUOUSacme.stock10 m 0 s · 2,000 ops · duration limitCOMPLETED
Sized for the pipeline you are about to run
Validate CDC behaviour, mappings and target throughput with a repeatable workload you control, then run Validator to prove the target still matches the source.
ORM-oriented stub data generation
Export without a database
Describe the columns, download the rows
Give Whisperer a column spec, or the columns of a bound table, and stream generated rows as CSV or JSON with the matching CREATE TABLE. No agent is queried, so it works for fixtures, ORM seeds and CI databases that do not exist yet.
- CSV and JSON exports streamed from a column spec: name, data type, faker hint, nullability
- ANSI CREATE TABLE preview for the same spec, to create the table your ORM expects
- Reuse a bound table's columns as the spec, so fixtures match the schema you replicate
- Same type- and hint-aware generator as a pillow fight, so fixtures and live load agree
Synthetics and seeding / Export
Export without a database
| Column | Type | Generator | Nullable |
|---|---|---|---|
id |
integer |
id | |
full_name |
varchar(80) |
name | |
email |
varchar(120) |
||
company |
varchar(80) |
company | |
country |
varchar(40) |
country | |
signed_up |
date |
date | |
notes |
text |
description |
No agent involved: rows come straight from this column spec, with the same type- and hint-aware generator a pillow fight uses. Drop the output into ORM seeds, test fixtures and CI databases.
id,full_name,email,company,country,signed_up,notes 1001,"Ada Lovelace",ada.lovelace@example.com,Initech,Spain,2025-11-02,"Prefers invoices by email" 1002,"Linus Carver",linus.carver@example.com,Umbrella,Brazil,2024-07-29, 1003,"Mia Okafor",mia.okafor@example.com,Hooli,Italy,2026-01-15,"Key account since 2024" 1004,"Noah Bergström",noah.bergstrm@example.com,Vandelay,Germany,2025-05-06,
Explore the platform
Continue exploring Gluesync
Synthetics and seeding feeds the rest of the platform: replicate what it writes, validate the result, and schedule the runs.
Validator
Seed a table, let the pipeline replicate it, then prove the target still matches the source.
Explore Validator →
Query Studio
Every statement Whisperer generates runs through Query Studio, with its SafetyGate and cancellation.
Explore Query Studio →
Schedules and events
Chain a seeding run before a snapshot, or start one from a Chronos schedule or platform event.
Explore Schedules and events →
Gluesync
Real-time CDC replication. Synthetics and seeding gives it traffic to prove itself on before go-live.
Explore Gluesync →Resources
Whisperer documentation and resources
Documentation
Technical guides, API reference, deployment tutorials.
Read documentation ↗
Public Roadmap
What we're building next. Submit feature requests.
View roadmap ↗
Support
Access technical support and operational assistance.
Get support →
Service Status
Real-time platform availability and incident history.
Check status ↗Seed it. Replicate it. Prove it.
Request a free trial or talk with the MOLO17 team about using Synthetics and seeding to rehearse your pipelines before production traffic arrives.