Question 1

What are the write modes and how do they differ?

Accepted Answer

Replicate writes the complete dataset to one stable object and replaces it on each run, which suits a consumer that always reads the same path. Snapshot writes the complete dataset to a new timestamped object per run, which suits archival and audit. There is also an incremental option on replicate that writes only changed records into new timestamped files. That option cannot be combined with snapshot mode, since the two describe different intentions.

Question 2

Which file formats can Snowflake data be written as?

Accepted Answer

CSV, JSON Lines, JSON array, and Parquet. Parquet is usually the right default for anything that will be queried later, since it carries types rather than forcing everything through text and reads efficiently by column. CSV remains useful when a downstream consumer expects it. There is no user facing compression setting, and Parquet is written with the library defaults rather than exposing codec or row group controls.

Question 3

How do I control where files land?

Accepted Answer

The output path is the bucket and prefix on the connection, then an optional subfolder, then the output name, which is configurable per source schema. Both the subfolder and the output name are evaluated as expressions, so a datetime expression can create date based folders automatically. That is how most teams organise exports. It is worth being clear that this is path templating rather than field based partitioning, which is not offered.

Question 4

What happens to deleted records in an incremental export?

Accepted Answer

They can appear as tombstone records carrying just the key and a deletion timestamp, rather than being removed from previously written files, which would not be possible once a file is closed. That means a consumer reading the full set of incremental files needs to apply those tombstones to arrive at current state. If that logic is unwelcome downstream, a scheduled replicate or snapshot run gives you a complete picture instead.

Connect Snowflake with Amazon S3

Connect Snowflake and Amazon S3 with Polytomic

Replicate or snapshot

Changed records only

Expression driven paths

Trusted by businesses worldwide

Step-by-step integration guide

Connect Snowflake in Polytomic

Connect Amazon S3 in Polytomic

Create a sync between Snowflake and Amazon S3

Map fields and enable the sync

Frequently asked questions

The sync and ETL platform that does it all

Powerful data sync

Advanced analytics

Enterprise security

Easy setup

Ready to connect Snowflake with Amazon S3?