> ## Documentation Index
> Fetch the complete documentation index at: https://docs.blockdb.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Amazon S3 Archive Delivery

> BlockDB writes EVM blockchain dataset exports directly to your S3 bucket — scheduled, partitioned, and lifecycle-controlled.

## Why S3

* Keep ownership of the data inside your AWS account.
* Apply lifecycle, encryption, or replication policies that match your governance model.
* Land both full-history archives and incremental refreshes in the same layout you use internally.

## Delivery Model

| Item             | Details                                                                                       |
| ---------------- | --------------------------------------------------------------------------------------------- |
| **Format**       | Parquet (default) or CSV; schema aligned with `/BlockDb.Postgres.Tables.Public`.              |
| **Partitioning** | `dataset_id=/0101/chain_id=1/date=2024-01-01/part-*.parquet`.                                 |
| **Access**       | BlockDB assumes an IAM role (`sts:AssumeRole`) or uses access keys scoped to a folder prefix. |
| **Integrity**    | Each object has an accompanying `.manifest.json` listing row counts and SHA-256 hashes.       |

## Setup Steps

1. Create (or reuse) an S3 bucket in the region closest to your workloads; enable versioning if you want change history.
2. Provision an IAM role with `s3:PutObject`, `s3:AbortMultipartUpload`, and `s3:ListBucket` on the target prefix. Share the ARN with BlockDB.
3. Send dataset IDs, chain coverage, and initial backfill ranges to [support@blockdb.io](mailto:support@blockdb.io).
4. Optionally enable cross-region replication or Glacier policies—BlockDB's writes respect your bucket policy.

## Consuming the Data

* Use AWS Glue or dbt to catalog the partitions and hydrate warehouse tables created from the BlockDB SQL scripts.
* Track ingestion by reading the manifest files and comparing against `_tracing_id` ranges.
* If you need mutable tables or streaming CDC, pair S3 archives with the [Real Time Channels](/user-guide/real-time-delivery/introduction) strategy.

<Callout icon="lightbulb" color="#3B82F6" iconType="regular">
  Coordinate bucket naming and encryption (SSE-KMS, SSE-S3, or client-side) before the first drop. KMS keys must grant BlockDB's IAM role encrypt/decrypt permissions.
</Callout>