Skip to main content
The GoldRush Pipeline API supports six destination types for delivering processed blockchain data to your infrastructure. Each destination operates in one of two pipeline modes:
  • Structured - Data passes through normalization, optional ABI decoding, and optional transforms before delivery.
  • Raw - Bytes are forwarded directly from the source with no processing.

​
Destination Routing Table

Destination TypePipeline ModeDelivery Guarantee
ClickHouseStructuredAt-least-once
PostgresStructuredAt-least-once
KafkaRawExactly-once
Object Storage (S3/GCS/R2)StructuredAt-least-once
AWS SQSStructuredAt-least-once
WebhookStructuredAt-least-once

​
Table Naming Convention

For database destinations (ClickHouse and Postgres), tables are written to {schema}.{table} where:
  • schema is derived from the project name. Hyphens in the project name are converted to underscores (e.g., project analytics-prod writes to schema analytics_prod).
  • table is the normalizer output table name.
Tables don’t need to pre-exist in the target database. The pipeline auto-creates tables.

​
Available Destinations

ClickHouse

Batch inserts into ClickHouse with configurable buffer size and flush intervals.

Postgres

Batch inserts into PostgreSQL with automatic column quoting for reserved words.

Kafka

Raw byte passthrough to a Kafka topic with exactly-once transactional delivery.

Object Storage

Write JSON or Parquet files to S3, GCS, or R2 with time-based partitioning.

AWS SQS

Deliver records to standard or FIFO SQS queues with built-in deduplication.

Webhook

HTTP POST delivery with retry logic, backoff, and idempotency keys.