Glossary

An alphabetical list of terms used in the DBConvert Streams documentation.

A

Alias

A short identifier assigned to a connection in federated query mode. Aliases are auto-generated based on connection type (e.g., pg1, my1, aws) and used to qualify table references across multiple sources.

API Key

A unique authentication token required to access DBConvert Streams services. See Licensing, Seats, and Telemetry.

B

Binary Log (Binlog)

A collection of log files generated by a MySQL server that records data modifications (INSERT, UPDATE, DELETE). DBConvert Streams reads binlog events in CDC mode to replicate changes to the target.

C

CDC Mode

A stream mode where DBConvert Streams captures changes from database transaction logs — binlog for MySQL, WAL for PostgreSQL. Supports INSERT, UPDATE, and DELETE operations. See What is Change Data Capture.

Change Data Capture

A technology that extracts row-level changes from database transaction logs. DBConvert Streams reads MySQL binlog and PostgreSQL WAL to capture data. See CDC Mode.

Connection

A configured link to a database, file system, or S3-compatible storage provider. Connections store host, credentials, and access settings. See Connection Management.

Load Mode

A stream mode where DBConvert Streams reads data directly from source tables in bulk, without using transaction logs. Supports INSERT operations only. In API payloads, this mode is represented as "mode": "load". See Load Mode.

CSV

Comma-separated values file format. Supported as both a file source (via DuckDB read_csv_auto()) and a target output format.

D

Data Bundle Size

The number of rows grouped into a single event before publishing to NATS. Default: 100 rows. Configurable per stream in the stream configuration.

Data Explorer

The built-in interface for browsing database schemas, viewing and editing data, running SQL queries, generating diagrams, and executing federated queries. See Data Explorer documentation.

DDL

Data Definition Language — the subset of SQL statements that modify schema structure: CREATE, ALTER, DROP for databases, tables, views, and other objects.

Desktop App

A native application built with Wails that bundles the API server, source reader, target writer, and NATS into a single process. Available for Windows, macOS, and Linux. See Desktop Apps.

DuckDB

An embedded analytical SQL engine used by DBConvert Streams for querying local files and S3 objects, powering federated queries, and providing SQL diagnostics in the SQL Console.

E

Event

A unit of data representing a row-level change (INSERT, UPDATE, or DELETE) captured from a source for replication to a target.

Event Stream

A continuous sequence of events flowing from a source through NATS JetStream to one or more target writers.

F

Federated Query

A SQL query that spans multiple sources — databases, local files, and S3 storage — in a single statement. Powered by DuckDB with alias-qualified table references. See Federated Queries.

H

HashiCorp Vault

An optional secrets management backend for Docker deployments. Stores database passwords, SSL certificates, and API keys. The desktop app uses a local encrypted file instead. See Credential Management.

I

Ingestion

The process of reading data from a source — either by direct table reads (Load mode) or from transaction logs (CDC mode).

J

JSONL

JSON Lines format — one JSON object per line. Supported as a target output format and queryable via DuckDB read_json_auto().

I

Initial Load + CDC

A stream configuration where DBConvert Streams copies existing source rows first, then continues CDC from the matching source log position. Enabled with source.options.initialSnapshot.enabled=true in CDC mode. See Initial Load + CDC.

L

Load Progress

The durable per-table copy cursor stored in the NATS JetStream KV bucket under <configID>.loadprogress. Tracks the primary-key range boundary of each completed chunk so a resumable load can continue from the next incomplete chunk after a restart. Only populated while the chunked path is active.

Logical Replication

A PostgreSQL replication method that streams row-level changes via the Write Ahead Log at the logical level (PostgreSQL 12+). See PostgreSQL CDC Reader.

M

Metrics

Measurements reported during stream execution: event counts, data sizes, transfer rates, and elapsed time. Visible in the Monitor tab.

N

NATS

A messaging system used by DBConvert Streams for communication between services. NATS JetStream provides persistent, ordered event delivery between source readers and target writers.

P

Parquet

A columnar file format optimized for analytical workloads. Supported as a target output format and queryable via DuckDB read_parquet().

R

Reader

See Source Reader.

Reporting Intervals

Configurable time intervals (in seconds) that determine how frequently progress updates are reported from source and target components during stream operation.

Resumable Load

A load mechanism that splits an eligible table into primary-key range chunks and persists the boundary of each completed chunk. If the stream is interrupted, the next start continues from the next incomplete chunk instead of re-reading rows that already reached the target. Routing is automatic per table — there is no public flag to enable or disable it. See Resumable Load.

S

S3-Compatible Storage

Object storage accessible via the S3 protocol. Includes Amazon S3, DigitalOcean Spaces, MinIO, and other compatible providers. Used as both a file source and a target destination. See S3-Compatible Storage.

Snowflake

A cloud data warehouse supported as a target destination. Data is loaded via staged Parquet files. See Snowflake Target.

Source

A database, file system, or S3 storage from which DBConvert Streams reads data. Supported sources: MySQL 5.7+, PostgreSQL 12+, local files (CSV, JSON, Parquet), and S3-compatible storage.

Source Reader

The service (stream-reader) that reads data from a source, converts it into events, and publishes them to NATS JetStream for consumption by target writers.

SQL Console

The query editor inside Data Explorer. Supports direct database queries, DuckDB-powered file queries, and federated queries with autocomplete, diagnostics, and templates. See SQL Console.

SSE (Server-Sent Events)

A protocol used to stream real-time logs and metrics from the API server to the browser. The endpoint /logs/stream delivers log events without requiring WebSocket connections.

Stream Limits

Optional constraints on a stream:

• numberOfEvents — maximum number of events to process
• elapsedTime — maximum duration in seconds

Stream States

The operational states a stream can be in:

• READY — initial state after creation
• RUNNING — active data transfer
• PAUSED — temporarily suspended
• STOPPED — terminated (cannot be resumed)
• FINISHED — successful completion
• FAILED — error state
• TIME_LIMIT_REACHED — stopped after reaching time limit
• EVENT_LIMIT_REACHED — stopped after reaching event limit

T

Target

A database, file system, or S3 storage to which data is replicated. Supported targets: MySQL, PostgreSQL, Snowflake, CSV, JSONL, and Parquet.

Target Writer

The service (stream-writer) that consumes events from NATS JetStream and writes them to the target destination. Multiple writers can run in parallel for horizontal scaling.

Template

A ready-made SQL snippet in the SQL Console that adapts to the current query context — database-specific queries for direct connections, DuckDB read_* functions for file/S3 connections, and cross-source patterns for federated queries.

U

Upsert

An operation that inserts a row if it does not exist or updates it if it does, based on primary key matching. Used by the target writer in both Load and CDC modes.

W

Write Ahead Log (WAL)

A PostgreSQL mechanism that records all changes before they are applied to the database. DBConvert Streams reads WAL entries in CDC mode to capture row-level changes. See Logical Replication.