/

Fivetran

Filament vs. Fivetran

Side-by-side capabilities and cost.

Introduction

Fivetran is a managed ELT service. It is known for a wide catalog of connectors and for requiring very little setup. It is closed source, runs in Fivetran's cloud, and bills on the rows you move. Filament is an open-source replication engine that runs entirely on your own infrastructure with nothing metered.

HIGHLIGHTS

OPEN SOURCE

LICENSE

Filament

APACHE 2.0

Fivetran

NO

Filament is closed source and proprietary.

OPEN SOURCE

LICENSE

Filament

APACHE 2.0

Fivetran

NO

Filament is closed source and proprietary.

SELF-HOSTED

DEPLOYMENT

Filament

YOUR CLOUD

Fivetran

VENDOR CLOUD

The control pane stays with Fivetran.

SELF-HOSTED

DEPLOYMENT

Filament

YOUR CLOUD

Fivetran

VENDOR CLOUD

The control pane stays with Fivetran.

NO METERING

PRICING MODEL

Filament

COMPUTE ONLY

Fivetran

PER ACTIVE ROW

Fivetran meters monthly active rows.

NO METERING

PRICING MODEL

Filament

COMPUTE ONLY

Fivetran

PER ACTIVE ROW

Fivetran meters monthly active rows.

YOURS TO KEEP

IF CONTRACT ENDS

Filament

KEEPS RUNNING

Fivetran

SYNCS PAUSE

Fivetran stops working.

YOURS TO KEEP

IF CONTRACT ENDS

Filament

KEEPS RUNNING

Fivetran

SYNCS PAUSE

Fivetran stops working.

Galaxy product illustration

License and control

License

Apache 2.0, open source.

Proprietary, closed source. The Connector SDK package, the Terraform provider, and dbt Core from the dbt Labs merger are the open pieces.

License

Apache 2.0, open source.

Proprietary, closed source. The Connector SDK package, the Terraform provider, and dbt Core from the dbt Labs merger are the open pieces.

Self-hosted in full

Yes. One repository, no paid tier, nothing gated.

No. The control plane is Fivetran's. Hybrid Deployment (Enterprise and up) runs an agent in your VPC. Local Data Processing (Business Critical only) still registers with Fivetran, sends MAR consumption data hourly, and refreshes license keys daily.

Self-hosted in full

Yes. One repository, no paid tier, nothing gated.

No. The control plane is Fivetran's. Hybrid Deployment (Enterprise and up) runs an agent in your VPC. Local Data Processing (Business Critical only) still registers with Fivetran, sends MAR consumption data hourly, and refreshes license keys daily.

Adding a new source

A YAML manifest for REST APIs, loaded at runtime with no rebuild. A Go driver for databases and other driver-based sources.

Python Connector SDK, hosted and run by Fivetran, billed as MAR like any other connector.

Adding a new source

A YAML manifest for REST APIs, loaded at runtime with no rebuild. A Go driver for databases and other driver-based sources.

Python Connector SDK, hosted and run by Fivetran, billed as MAR like any other connector.

Connector maintenance

Every connector lives in the engine repository with a published maturity level (alpha, beta, stable).

700-plus connectors maintained by Fivetran. Lite connectors cover partial use cases. Partner-built connectors are maintained by the partner.

Connector maintenance

Every connector lives in the engine repository with a published maturity level (alpha, beta, stable).

700-plus connectors maintained by Fivetran. Lite connectors cover partial use cases. Partner-built connectors are maintained by the partner.

Services to operate

Server and control plane, with Postgres and NATS. Each run is its own Kubernetes Job. One binary for evaluation.

None for SaaS. Hybrid needs an x86-only agent host with 8 vCPU and 32 GB for 1 to 2 connections.

Services to operate

Server and control plane, with Postgres and NATS. Each run is its own Kubernetes Job. One binary for evaluation.

None for SaaS. Hybrid needs an x86-only agent host with 8 vCPU and 32 GB for 1 to 2 connections.

Runtime

Static Go binaries on distroless images. One runner code path wherever a run executes.

Fivetran's cloud (GCP by default) in one of seven regions. Cloud provider choice on Enterprise, specific cloud region on Business Critical.

Runtime

Static Go binaries on distroless images. One runner code path wherever a run executes.

Fivetran's cloud (GCP by default) in one of seven regions. Cloud provider choice on Enterprise, specific cloud region on Business Critical.

Scheduling

Built-in cron per pipeline. Safe to run multiple control planes.

Built-in intervals from 1 minute to 24 hours. 1-minute syncs need Enterprise or Business Critical. Standard is marketed at 15 minutes.

Scheduling

Built-in cron per pipeline. Safe to run multiple control planes.

Built-in intervals from 1 minute to 24 hours. 1-minute syncs need Enterprise or Business Critical. Standard is marketed at 15 minutes.

SaaS source execution

Manifests interpreted inside the engine, in the same process as the run. No per-connector containers.

Fivetran's cloud, or the hybrid agent for connectors marked hybrid capable.

SaaS source execution

Manifests interpreted inside the engine, in the same process as the run. No per-connector containers.

Fivetran's cloud, or the hybrid agent for connectors marked hybrid capable.

Embeddable

Yes. Go library (app.Run) with swappable event bus, state store, and secrets.

Powered by Fivetran. Embedding the managed service through its API, not the engine.

Embeddable

Yes. Go library (app.Run) with swappable event bus, state store, and secrets.

Powered by Fivetran. Embedding the managed service through its API, not the engine.

Interfaces

Web UI, ConnectRPC API, CLI.

Web dashboard, REST API, Terraform provider. CLI only for the Connector SDK.

Interfaces

Web UI, ConnectRPC API, CLI.

Web dashboard, REST API, Terraform provider. CLI only for the Connector SDK.

Correctness

Resume after failure

Mid-table, from per-shard checkpoints, for upsert, incremental, and CDC runs. A failed run lands in a partial state and continues on re-request.

Cursor checkpoints. Large tables copied in chunks so a dropped connection does not restart the table. Connections auto-pause after 14 days of failure.

Resume after failure

Mid-table, from per-shard checkpoints, for upsert, incremental, and CDC runs. A failed run lands in a partial state and continues on re-request.

Cursor checkpoints. Large tables copied in chunks so a dropped connection does not restart the table. Connections auto-pause after 14 days of failure.

Per-batch integrity check

CRC32-C computed independently on the read side and the write side of every batch. A mismatch fails the run.

None documented. Idempotent upserts by primary key and a manual re-sync when data drifts.

Per-batch integrity check

CRC32-C computed independently on the read side and the write side of every batch. A mismatch fails the run.

None documented. Idempotent upserts by primary key and a manual re-sync when data drifts.

Change data capture

Native. Postgres logical replication (pgoutput) with a consistent snapshot bootstrap. MySQL binlog with GTID cursors.

Log-based for Postgres, MySQL, SQL Server, Oracle. Query-based fallbacks may scan full tables and cannot see delete-then-reinsert.

Change data capture

Native. Postgres logical replication (pgoutput) with a consistent snapshot bootstrap. MySQL binlog with GTID cursors.

Log-based for Postgres, MySQL, SQL Server, Oracle. Query-based fallbacks may scan full tables and cannot see delete-then-reinsert.

Schema drift

New columns added to the destination automatically. Add-only, nothing dropped, nothing rewritten.

New columns added. Widening type changes applied, narrowing ignored. Deleted source objects left in place.

Schema drift

New columns added to the destination automatically. Add-only, nothing dropped, nothing rewritten.

New columns added. Widening type changes applied, narrowing ignored. Deleted source objects left in place.

Parallel reads within a table

Yes. Large tables split into up to 64 shards (keyset, bitmap, or ctid) read concurrently.

Multi-threaded queries and chunked reads are documented. Splitting one table across concurrent loaders is not.

Parallel reads within a table

Yes. Large tables split into up to 64 shards (keyset, bitmap, or ctid) read concurrently.

Multi-threaded queries and chunked reads are documented. Splitting one table across concurrent loaders is not.

Row lineage

Every replicated row carries the run id and start time. CDC rows also carry operation, source position, and sequence.

_fivetran_synced and _fivetran_deleted per row. _fivetran_id on tables without a primary key.

Row lineage

Every replicated row carries the run id and start time. CDC rows also carry operation, source position, and sequence.

_fivetran_synced and _fivetran_deleted per row. _fivetran_id on tables without a primary key.

Where your data travels

Source to sink inside your network.

Through Fivetran's cloud on SaaS. Metadata, row counts, and column names leave your network even on Hybrid.

Cost

What drives billing

Nothing. Compute you already run.

Monthly active rows, tiered per connection since March 2025. Deletes and history-mode changes count since January 2026. No public per-MAR rate card.

What drives billing

Nothing. Compute you already run.

Monthly active rows, tiered per connection since March 2025. Deletes and history-mode changes count since January 2026. No public per-MAR rate card.

Paid tiers

None.

Free up to 500,000 MAR. Standard, Enterprise, Business Critical. Hybrid needs Enterprise. Local Data Processing, private networking, and customer-managed keys need Business Critical.

Paid tiers

None.

Free up to 500,000 MAR. Standard, Enterprise, Business Critical. Hybrid needs Enterprise. Local Data Processing, private networking, and customer-managed keys need Business Critical.

Comparison

Full Filament vs. Fivetran benchmark report

Full Filament vs. Fivetran benchmark report

Full Filament vs. Fivetran benchmark report

Migrate to

Migrate to

Copyright © 2026 Galaxy. All rights reserved.