Fivetran Managed ELT Cheat Sheet

Updated 2026-05-15

Next Topic: Google BigQuery for Data Engineering Cheat Sheet_v1_tables

🧠Study flashcards on this topic96 cards · spaced repetition→

Fivetran is a fully managed Extract, Load, Transform (ELT) platform that automates data movement from 700+ sources to cloud data warehouses, lakes, and analytics platforms with zero-maintenance pipelines. Unlike traditional ETL, Fivetran loads raw data first, then transforms it in the destination using tools like dbt, leveraging the compute power of modern warehouses. The platform's core value proposition is operational simplicity: automated schema migration, built-in connector maintenance, log-based change data capture, and consumption-based pricing measured in Monthly Active Rows (MAR)—rows that are inserted, updated, or deleted each month. Understanding Fivetran means recognizing the trade-off between managed convenience at a premium cost versus self-hosted flexibility, and knowing when its connector library, CDC capabilities, and enterprise compliance certifications justify the investment for your data engineering stack.

What This Cheat Sheet Covers

This topic spans 15 focused tables and 114 indexed concepts, 96 flashcards. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.

Table 1: Connector Types and CategoriesTable 2: Sync Frequency and SchedulingTable 3: Historical Sync and Initial LoadTable 4: Incremental Replication MethodsTable 5: Fivetran Transformations and dbt IntegrationTable 6: Normalized Schema Output and Table NamingTable 7: Pricing Plans and TiersTable 8: Connector Health Monitoring and AlertsTable 9: Fivetran REST API for AutomationTable 10: Destination Setup for Major PlatformsTable 11: Privacy, Security, and Data ProtectionTable 12: Hybrid Deployment and Private DeploymentTable 13: Census Acquisition and Reverse ETLTable 14: Advanced Features and OptimizationsTable 15: Common Use Cases and Anti-Patterns

Quick IndexSubscribe to unlock

A jump-to index of every table row in this cheat sheet.

Mind MapSubscribe to unlock

An interactive map of every table and concept in this topic.

Table 1: Connector Types and Categories

Connectors are the front door to Fivetran—each one knows how to extract from a specific kind of source, and the category tells you what extraction method and feature set to expect. Databases lean on log-based CDC, SaaS apps go through REST APIs, and the spectrum runs from fully-featured Standard connectors down to lightweight Lite ones and heavy-duty High-Volume Agents for enterprise databases.

Type	Example	Description
Database connectors	`PostgreSQL` `MySQL` `SQL Server` `Oracle`	• Replicate data from relational databases using log-based CDC or query-based methods • supports incremental sync, soft deletes, and schema drift handling
SaaS application connectors	`Salesforce` `HubSpot` `Zendesk` `Google Analytics`	• Extract data from business apps via REST APIs • handles rate limiting, pagination, and API version changes automatically
File connectors	`Amazon S3` `Google Cloud Storage` `Azure Blob Storage`	• Ingest CSV, JSON, Avro, Parquet from cloud storage • supports unstructured file replication with metadata tracking for files up to 5 GB
Event stream connectors	`Kafka` `AWS Kinesis` `Google Pub/Sub`	• Capture streaming data from message queues and event platforms • delivers near real-time ingestion for time-series and event-driven workloads
Function connectors	`Custom Python SDK` `REST API endpoint`	• Build custom connectors using Connector SDK with Python • Fivetran manages compute, scheduling, and error handling for your code
Standard connectors	`Snowflake source` `BigQuery source` `Postgres CDC`	• Professionally maintained, enterprise-grade connectors with full feature support • included in all pricing tiers with guaranteed SLAs
Lite connectors	`Slack Lite` `Okta Lite` `Chargebee Lite`	• Lightweight API-based connectors for niche SaaS apps • same fully managed experience but may have fewer features than Standard tier
High-Volume Agent (HVA) connectors	`SAP ECC on Oracle` `SAP on Db2 for i` `Oracle HVA`	• Enterprise-grade connectors requiring on-premises agent for high-volume databases • uses log-based CDC to minimize source impact and support massive scale

Table 2: Sync Frequency and Scheduling

How often a connector syncs is a direct trade-off between data freshness, API rate limits, and MAR cost. These options range from 1-minute near-real-time syncs on the top tiers down to once-daily batch runs, plus manual triggering for orchestrators and a pause/resume control for stopping a connector without losing its state.

Option	Example	Description
1-minute sync	`sync_frequency: 1 minute`	• Fastest available sync on Enterprise and Business Critical plans • near real-time data delivery for time-sensitive analytics

The rest of this section is part of the full cheat sheet

Table 3: Historical Sync and Initial Load

The first sync is special: Fivetran backfills all existing source data before switching to incremental mode, and crucially that historical load is free—MAR billing only starts on the incremental updates that follow. The settings here let you cap how far back to backfill, tune parallelism against source load, and re-sync a single table or the whole connector when data drifts.

Feature	Example	Description
Historical sync (initial load)	Full database replication on first sync	• Complete backfill of source data when a connector is first set up • Fivetran replicates all accessible tables and rows before switching to incremental mode

The rest of this section is part of the full cheat sheet

Table 4: Incremental Replication Methods

After the initial load, Fivetran only moves what changed—and the method it uses determines latency, source impact, and whether deletes get captured. Log-based CDC is the gold standard, reading transaction logs directly, while cursor-column and full-table approaches trade simplicity for missed deletes or higher cost. Knowing which method a source supports is key to predicting both freshness and your MAR bill.

Method	Example	Description
Log-based CDC (Change Data Capture)	`PostgreSQL: WAL` `MySQL: binlog` `SQL Server: CT/CDC`	• Reads database transaction logs directly for real-time, low-latency replication • captures inserts, updates, and deletes with minimal source impact • gold standard for CDC.

The rest of this section is part of the full cheat sheet

Table 1: Connector Types and Categories

Type	Example	Description
Database connectors	`PostgreSQL` `MySQL` `SQL Server` `Oracle`	• Replicate data from relational databases using log-based CDC or query-based methods • supports incremental sync, soft deletes, and schema drift handling
SaaS application connectors	`Salesforce` `HubSpot` `Zendesk` `Google Analytics`	• Extract data from business apps via REST APIs • handles rate limiting, pagination, and API version changes automatically
File connectors	`Amazon S3` `Google Cloud Storage` `Azure Blob Storage`	• Ingest CSV, JSON, Avro, Parquet from cloud storage • supports unstructured file replication with metadata tracking for files up to 5 GB
Event stream connectors	`Kafka` `AWS Kinesis` `Google Pub/Sub`	• Capture streaming data from message queues and event platforms • delivers near real-time ingestion for time-series and event-driven workloads
Function connectors	`Custom Python SDK` `REST API endpoint`	• Build custom connectors using Connector SDK with Python • Fivetran manages compute, scheduling, and error handling for your code
Standard connectors	`Snowflake source` `BigQuery source` `Postgres CDC`	• Professionally maintained, enterprise-grade connectors with full feature support • included in all pricing tiers with guaranteed SLAs
Lite connectors	`Slack Lite` `Okta Lite` `Chargebee Lite`	• Lightweight API-based connectors for niche SaaS apps • same fully managed experience but may have fewer features than Standard tier
High-Volume Agent (HVA) connectors	`SAP ECC on Oracle` `SAP on Db2 for i` `Oracle HVA`	• Enterprise-grade connectors requiring on-premises agent for high-volume databases • uses log-based CDC to minimize source impact and support massive scale

Table 2: Sync Frequency and Scheduling

Option	Example	Description
1-minute sync	`sync_frequency: 1 minute`	• Fastest available sync on Enterprise and Business Critical plans • near real-time data delivery for time-sensitive analytics

Table 3: Historical Sync and Initial Load

Feature	Example	Description
Historical sync (initial load)	Full database replication on first sync	• Complete backfill of source data when a connector is first set up • Fivetran replicates all accessible tables and rows before switching to incremental mode

Table 4: Incremental Replication Methods

Method	Example	Description
Log-based CDC (Change Data Capture)	`PostgreSQL: WAL` `MySQL: binlog` `SQL Server: CT/CDC`	• Reads database transaction logs directly for real-time, low-latency replication • captures inserts, updates, and deletes with minimal source impact • gold standard for CDC.