ddbimp 0.2

Easily load data from CSV to test out your DynamoDB table design

dynamodb-dev-importer

Easily load data from CSV to test out your DynamoDB table design.

When working with DynamoDB, it is common practice to minimise the number of tables used, ideally down to just one.

Techniques such as sparse indexes and GSI overloading allow a lot of flexibility and efficiency.

Designing a good schema that supports your query patterns can be challenging. Often it is nice to try things out with a small amount of data. I personally find it convenient to enter data into a spreadsheet and play around with it there.

When ready to try out an approach with DynamoDB, it's a hassle to then create a items in a table through the AWS Console or CLI, so this script:

• reads a CSV file (exported from your spreadsheet) and imports it into a DynamoDB table
• columns 0 and 1 are used for the key: partition key pk: S and sort key sk: S - your target table needs these keys defined
• column 2, if not an empty string, is set to data: S
• all other columns are added as non-key attributes

Your CSV should contain columns for:

• pk
• sk
• data (optional)
• anything after those three can contain arbitrary attributes of form attribute_name: value i.e. city: Edinburgh

Example row:

PERSON-1,sales-Q1-2019,Alex,jan: 12012,feb: 1927

Will yield an item like this:

{
    pk: 'PERSON-1',
    sk: 'sales-Q1-2019',
    data: 'Alex',
    jan: 12012,
    feb 1927
}

Usage

Assuming DynamoDB table example(pk, sk) is setup and you're in a virtual environment. If you already have boto3 installed, you don't need to install any packages.

$ pip install ddbimp
$ ddbimp --table example --skip 1 example.csv

Key ideas

• Table consists of partition key pk: S and sort key sk: S - their meaning varies depending on the item
• A secondary index swaps the sort and partition keys, so the partition key is sk: S and sort key pk: S
• A final secondary index uses sk: S and data: S where data is an arbitrary value you might want to search for, the meaning of data depends on the item it is part of
• Group items through a shared partition key, store sub items with a sort key e.g.
  • e.g. pk:PERSON-1, sk:sales-Q1-2019, jan:12012, feb:1927

See example.csv for an example input file.

AWS recently released a preview build of a tool called NoSQL Workbench. It builds on the above ideas. I've tried it out and it seems pretty good, but I am a luddite and am faster working in a spreadsheet right now. I'd certainly recommend giving it a try.

Useful resources

• https://docs.aws.amazon.com/amazondynamodb/latest/developerguide/bp-indexes.html
• https://docs.aws.amazon.com/amazondynamodb/latest/developerguide/workbench.html
• https://www.youtube.com/watch?v=6yqfmXiZTlM

Caveats, TODO

• Uses your default AWS profile
• Region needs to be set
• Make work directly with a Google Sheets via sheets API