Quick map (AWS service → Snowflake / Databricks)
- Athena
- Snowflake: Direct via the Athena Snowflake connector (federated query over JDBC; registered in Glue/Lake Formation).
- Databricks: Indirect—Athena can read Delta Lake tables in S3 (so it can query Databricks-written Delta data) when tables are in Glue Catalog.
- AWS Glue
- Snowflake: Direct—native Glue → Snowflake connectivity GA (read/write; Glue Studio source/target).
- Databricks: Works with Glue catalog patterns (historic); Databricks now recommends Unity Catalog but still documents federation to Glue.
- QuickSight
- Snowflake: Direct—native QuickSight connector to Snowflake.
- Databricks: Direct—native QuickSight connector to Databricks SQL. Amazon Web Services, Inc.
- Kinesis / Firehose
- Snowflake: Direct via Firehose → Snowpipe Streaming (seconds-level ingest); classic Snowpipe via S3 also works.
- Databricks: Direct—Structured Streaming connector for Kinesis (source/sink).
- S3 + SQS/SNS (events)
- Snowflake: Direct—Snowpipe auto-ingest uses S3 events → SQS; can fan-out via SNS; error notifications to SNS supported.
- Databricks: Common—uses S3 natively; SQS/SNS typically via AWS SDKs or event fan-out into Kinesis/Firehose (no special first-party SQS listener).
- DynamoDB
- Snowflake: No native direct; typical paths are Glue ETL or 3rd-party CDC/ETL to S3/Snowpipe.
- Databricks: Via Spark connectors/partners (for example, DynamoDB Spark/EMR connector; partner tools).
- Redshift
- Snowflake / Databricks: No native direct bridge. Use S3 interchange (UNLOAD/COPY), Athena federation where applicable, or partner ETL. (AWS and 3P references illustrate S3/ETL patterns.)
- EMR
- Snowflake: Indirect via S3/Glue pipelines.
- Databricks: Parallel platforms; “integration” is usually shared S3 data + Glue Catalog (EMR uses Glue as Hive metastore).
- Timestream
- Snowflake / Databricks: No first-party direct integration documented; common approaches are export to S3 (then Snowpipe/Delta), Athena/SDK/JDBC middle tiers. (No official Snowflake/Databricks connector pages.)
- SNS/SQS (messaging)
- Snowflake: see Snowpipe auto-ingest + SNS error notifications.
- Databricks: handled via AWS SDKs/streams (or route through Kinesis/Firehose).
How to choose (fast decisioning)
- Query Snowflake from AWS analytics? Use Athena Snowflake connector (governed via Glue/Lake Formation).
- ETL between Snowflake and AWS data lake/services? Prefer Glue’s native Snowflake connectivity (Studio visual jobs; SQL source; upsert/merge).
- Real-time into Snowflake? Firehose → Snowpipe Streaming (or S3→Snowpipe with SQS/SNS).
- Real-time with Databricks? Kinesis Structured Streaming (or MSK).
- BI on Snowflake/Databricks? QuickSight native connectors for both.
- Athena on Databricks-written data? Store as Delta Lake in S3 with Glue catalog; Athena reads it natively.