Amazon S3

Category

Files & storage

Runs in

The Phloem cloud — nothing to install

Lets you and Hamming query CSV, Parquet and JSON files in your S3 buckets with SQL.

#What Phloem reads

• CSV, Parquet (.parquet, .pqt) and JSON files in the buckets you pick
• Each table's columns, listed in Trace

Files are read in place when queried. Nothing is copied into Phloem first.

#What you need

An access key, or a role Phloem can assume, allowed to:

• list your buckets (s3:ListAllMyBuckets)
• list objects in the buckets you connect (s3:ListBucket)
• read those objects (s3:GetObject)

For IAM Role (STS Assume Role), the role's trust policy must let Phloem assume it, with your external ID if you set one.

#Connect

1. In the console, open Connectors, choose Add connector, then Amazon S3.
2. On AWS Credentials, choose an Authentication Method (Access Key & Secret or IAM Role (STS Assume Role)), fill in its fields, and pick an AWS Region. Session Token (Optional) is only for temporary credentials.
3. Phloem tests the credentials by listing your buckets.
4. On Select Buckets, pick the buckets to connect.
5. On Discovered Tables, review the tables found. Untick any you don't want, or edit a table's Table Name, File Pattern (raw/events/**/*.parquet) or Description (Optional).
6. On Save Connection, name the connection and save it.

#What Hamming can do

• Run SQL over the connected tables: query, sample rows, count rows, list columns

Nothing it does changes your buckets.

#Notes

• Discovery proposes one table per file. Set a File Pattern glob (raw/events/**/*.parquet) to query many files as one table.
• An assumed role's credentials last one hour and renew as needed.
• Credentials with a Session Token stop working when the token expires. Edit the connector with fresh ones.

#Connection fields

What the Add connector form asks for.

Updated November 6, 2026