Skip to content
A SQLite driver for S3 and Amazon Athena 😳
Python Shell Dockerfile
Branch: master
Clone or download
Type Name Latest commit message Commit time
Failed to load latest commit information.
lambda-layer Initial revision 😳 Dec 11, 2019
sample-data Cleaning up the code in the main lambda function and removing hardcod… Dec 15, 2019
.gitignore Add serverless repo template Dec 16, 2019
LICENSE Add serverless repo template Dec 16, 2019
athena-sqlite.yaml Add serverless repo template Dec 16, 2019

Athena SQLite Driver

Using Athena's new Query Federation functionality, read SQLite databases from S3.

Install it from the Serverless Application Repository: AthenaSQLiteConnector.


I occasionally like to put together fun side projects over Thanksgiving and Christmas holidays.

I'd always joked it would a crazy idea to be able to read SQLite using Athena, we are!


The PyArrow library unfortunately weighs in over 250MB, so we have to use a custom compilation step to build a Lambda Layer.


Drop SQLite databases in a single prefix in S3, and Athena will list each file as a database and automatically detect tables and schemas.

Currently, all data types are strings. I'll fix this eventually. All good things in time.


This project is under active development and very much in it's infancy.

Many things are hard-coded or broken into various pieces as I experiment and figure out how everything works.


The documentation for this is a work in progress. It's currently in between me creating the resources manually and building the assets for the AWS SAR, and most of the docs will be automated away.


  • Docker
  • Python 3.7

Lambda layer

First you need to build Lambda layer. There are two Dockerfiles and build scripts in the lambda-layer/ directory.

We'll execute each of the build scripts and copy the results to the target directory. This is referenced by the SAR template, athena-sqlite.yaml.

cd lambda-layer
cp -R layer/ ../target/

Upload sample data

For the purpose of this test, we just have a sample sqlite database you can upload.

aws s3 cp sample-data/sample_data.sqlite s3://<TARGET_BUCKET>/<TARGET_PREFIX>/

Feel free to upload your own SQLite databases as well!

Lambda function

There are three components to the Lambda code:

  • - A SQLite Virtual File System implementation for S3
  • - The actual Lambda function that handles Athena metadata/data requests
  • - Helper functions for access SQLite databases on S3

Create a function with the code in lambda-function/ that uses the previously created layer. The handler will be s3qlite.lambda_handler Also include the and files in your Lambda function

Configure two environment variables for your lambda function:

  • TARGET_BUCKET - The name of your S3 bucket where SQLite files live
  • TARGET_PREFIX - The prefix (e.g. data/sqlite) that you uploaded the sample sqlite database to

Note that the IAM role you associate the function with will also need s3:GetObject and s3:ListBucket access to wherever your lovely SQLite databases are stored.

Configure Athena

Follow the Athena documentation for Connecting to a data source. The primary thing to note here is that you need to create a workgroup named AmazonAthenaPreviewFunctionality and use that for your testing. Some functionality will work in the primary workgroup, but you'll get weird errors when you try to query data.

I named my function s3qlite :)

Run queries!

Here's a couple basic queries that should work:

SELECT * FROM "s3qlite"."sample_data"."records" limit 10;

SELECT COUNT(*) FROM "s3qlite"."sample_data"."records";

If you deploy the SAR app, the data catalog isn't registered automatically, but you can still run queries by using the special lambda: schema:

SELECT * FROM "lambda:s3qlite".sample_data.records LIMIT 10;

Where s3qlite is the value you provided for the AthenaCatalogName parameter.


  • Move these into issues :)
  • Move into it's own module
    • Maybe add write support to it someday 😱
  • Publish to SAR
  • Add tests...always tests
  • struct types, probably
  • Don't read the entire file every time :)
  • Escape column names with invalid characters
  • Implement recursive listing

Serverless App Repo

These are mostly notes I made while figuring out how to get SAR working.

Need to grant SAR access to the bucket

aws s3api put-bucket-policy --bucket <BUCKET> --region us-east-1 --policy '{
  "Version": "2012-10-17",
  "Statement": [
      "Effect": "Allow",
      "Principal": {
        "Service":  ""
      "Action": "s3:GetObject",
      "Resource": "arn:aws:s3:::<BUCKET>/*"

For publishing to the SAR, we just execute two commands

sam package --template-file athena-sqlite.yaml --s3-bucket <BUCKET> --output-template-file target/out.yaml
sam publish --template target/out.yaml --region us-east-1

If you want to deploy using CloudFormation, use this command:

sam deploy --template-file ./target/out.yaml --stack-name athena-sqlite --capabilities CAPABILITY_IAM --parameter-overrides 'DataBucket=<BUCKET> DataPrefix=tmp/sqlite' --region us-east-1
You can’t perform that action at this time.