Mirror vs. Turbo Pipelines

Both Mirror and Turbo stream onchain data to your infrastructure, but they differ in configuration, supported features, and operational behavior. Use this page to understand the differences when choosing between them or migrating from one to the other.

Top-level configuration

Feature	Mirror	Turbo
`apiVersion` field	Required (`apiVersion: 3`)	Not used
Resource sizes	`s`, `m`, `l`, `xl`, `xxl`	`xs`, `s`, `m`, `l`, `xl`, `xxl`
Job mode	Not supported	`job: true` for one-time batch runs
CLI namespace	`goldsky pipeline <cmd>`	`goldsky turbo <cmd>`
Live inspect	Not supported	`goldsky turbo inspect`
Delivery semantics	At-least-once	At-least-once with checkpointing
Startup time	Slower	Under 5 seconds

Sources

Source type	Mirror	Turbo	Notes
Dataset (direct indexing)	Supported	Supported	Both support curated datasets
Subgraph entity	Supported	Not supported	Mirror-only source type
EVM chains	130+ chains	Supported
Edge EVM (RPC-direct)	Not supported	Supported	Any EVM chain via a Goldsky Edge endpoint or your own RPC
Solana	Not supported	Supported	Full historical data from genesis
Bitcoin	Not supported	Supported
Stellar	Not supported	Supported
NEAR	Not supported	Supported
`start_block` / `end_block`	Not supported	Supported	Block range filtering
`start_at` (latest/earliest)	Supported on dataset sources	Supported	Starting position control
`in_order` mode	Not supported	Supported (Solana only)	Ensures blocks are emitted in sequential slot order

Transforms

Transform type	Mirror	Turbo	Notes
SQL	Full SQL	Limited SQL	Turbo uses Apache DataFusion; no joins, aggregations, or window functions in streaming mode
WebAssembly / TypeScript	Not supported	Supported	Custom logic compiled to WASM
External HTTP handler	Supported	Supported
Dynamic tables	Not supported	Supported	Real-time updatable lookup tables
Handler `schema_override`	Supported	Supported	Add, change, or drop columns
Handler `payload_columns`	Supported	Not supported	Bandwidth optimization for Mirror handlers
Handler `batch_size`	Supported (default 100)	Supported
Handler `batch_flush_interval`	Supported (default 1s)	Supported
Handler `one_row_per_request`	Supported	Supported	When `true`, sends one HTTP request per row instead of batched arrays
Handler `payload_version`	Supported	Supported	`0` (flat JSON, default) or `1` (envelope with `metadata.op` + `data`)
Handler `secret_name`	Supported	Supported	Reference a stored secret instead of inline `headers`
`_gs_op` column tracking	Not documented	Supported	Tracks insert/update/delete operations

Sinks

Sink type	Mirror	Turbo	Notes
PostgreSQL	Supported	Supported	Turbo adds `on_conflict`, `update_where`, and `parallelism` options
PostgreSQL aggregation	Not supported	Supported	Trigger-based real-time aggregations
ClickHouse	Supported	Supported	See sink tuning section below
MySQL	Supported	Supported
Elasticsearch	Supported	Not supported
Timescale	Supported	Not supported
Kafka	Supported	Supported	Turbo adds `data_format`, `topic_partitions`, `message_max_bytes`
Object storage / File	Supported	Not supported
S3	Via object storage sink	Native `s3_sink`
SQS	Supported (`type: sqs`)	Supported (`type: sqs_sink`)	Different type names
DynamoDB	Supported	Not supported
Snowflake	Via S3	Not supported	Not a native Mirror sink; load periodically from S3
Webhook	Supported	Supported	See webhook options and secrets sections below
Google Cloud Pub/Sub	Not supported	Supported	Turbo-only
S2	Not supported	Supported
Blackhole	Not supported	Supported	Testing sink

Sink tuning parameters

Mirror and Turbo expose tuning parameters at different layers. Mirror applies buffering at the pipeline level, while Turbo configures batching and concurrency per sink.

Batching and throughput

Capability	Mirror	Turbo
Batch size	`sink_buffer_max_rows` (pipeline-level)	`batch_size` (per sink: PostgreSQL, ClickHouse, MySQL, Kafka, Webhook)
Batch flush interval	`sink_buffer_interval` (pipeline-level)	`batch_flush_interval` (per sink: PostgreSQL, ClickHouse, MySQL, Kafka, Webhook)
Parallelism	Not configurable	`parallelism` (PostgreSQL, ClickHouse, Kafka)
Kafka message size limit	Not configurable	`message_max_bytes`

PostgreSQL

Parameter	Mirror	Turbo
`on_conflict`	Not supported	Supported (default: `update`)
`update_where`	Not supported	Supported

ClickHouse

Parameter	Mirror	Turbo
`append_only_mode`	Supported	Supported
`schema_override`	Supported (`schemaOverride`)	Supported
`versionColumnName`	Supported	Not supported

Webhook and handler options

Both webhooks (sinks) and external HTTP handlers (transforms) share several configuration options:

Option	Mirror	Turbo	Notes
`one_row_per_request`	Supported	Supported	When `false` (default), rows are sent as a JSON array. When `true`, one HTTP request per row
`payload_version`	Supported	Supported	`0` (default, flat JSON) or `1` (envelope format with `metadata.op` + `data`)
`primary_key`	Not supported	Supported	Turbo webhook sinks only
`secret_name`	Supported	Supported	Reference a stored secret instead of inline `headers`

Secrets and authentication

Feature	Mirror	Turbo	Notes
Database sink secrets	`secret_name` with JSON object format (`type`, `host`, `port`, etc.)	`secret_name` with connection string format (`postgres://user:pass@host/db`)	Different formats
Webhook secrets	`secret_name` with `httpauth` type	`secret_name` or inline `headers`	See webhook options section above
SQS secrets	`secret_name` only	`secret_name` or direct `access_key_id`/`secret_access_key` inline	Turbo allows inline credentials
S3 secrets	N/A	`secret_name` or direct credentials

For more details on managing secrets, see Mirror secrets.

Choosing between Mirror and Turbo

Use Mirror when you need:

Subgraph entity sources
Elasticsearch, Timescale, or DynamoDB sinks
Sink buffer tuning parameters

Use Turbo when you need:

Non-EVM chains (Solana, Bitcoin, Stellar, NEAR)
TypeScript/WASM transforms
Dynamic tables
Live pipeline inspection
Job mode for batch processing
PostgreSQL aggregations
Block range filtering (start_block / end_block)

Overview

Solutions

More information

Mirror vs. Turbo pipelines

Top-level configuration

Sources

Transforms

Sinks

Sink tuning parameters

Batching and throughput

PostgreSQL

ClickHouse

Webhook and handler options

Secrets and authentication

Choosing between Mirror and Turbo

​Top-level configuration

​Sources

​Transforms

​Sinks

​Sink tuning parameters

​Batching and throughput

​PostgreSQL

​ClickHouse

​Webhook and handler options

​Secrets and authentication

​Choosing between Mirror and Turbo

Top-level configuration

Sources

Transforms

Sinks

Sink tuning parameters

Batching and throughput

PostgreSQL

ClickHouse

Webhook and handler options

Secrets and authentication

Choosing between Mirror and Turbo