Skip to content

Scan large DynamoDB tables faster with parallelism

License

Notifications You must be signed in to change notification settings

shelfio/dynamodb-parallel-scan

Repository files navigation

dynamodb-parallel-scan CircleCI npm (scoped)

Scan DynamoDB table concurrently (up to 1,000,000 segments), recursively read all items from every segment

A blog post going into details about this library.

Install

$ yarn add @shelf/dynamodb-parallel-scan

This library has 2 peer dependencies:

  • @aws-sdk/client-dynamodb
  • @aws-sdk/lib-dynamodb

Make sure to install them alongside this library.

Why this is better than a regular scan

Easily parallelize scan requests to fetch all items from a table at once. This is useful when you need to scan a large table to find a small number of items that will fit the node.js memory.

Scan huge tables using async generator or stream. And yes, it supports streams backpressure! Useful when you need to process a large number of items while you scan them. It allows receiving chunks of scanned items, wait until you process them, and then resume scanning when you're ready.

Usage

Fetch everything at once

const {parallelScan} = require('@shelf/dynamodb-parallel-scan');

(async () => {
  const items = await parallelScan(
    {
      TableName: 'files',
      FilterExpression: 'attribute_exists(#fileSize)',
      ExpressionAttributeNames: {
        '#fileSize': 'fileSize',
      },
      ProjectionExpression: 'fileSize',
    },
    {concurrency: 1000}
  );

  console.log(items);
})();

Use as async generator (or streams)

Note: highWaterMark determines items count threshold, so Parallel Scan can fetch concurrency * 1MB more data even after highWaterMark was reached.

const {parallelScanAsStream} = require('@shelf/dynamodb-parallel-scan');

(async () => {
  const stream = await parallelScanAsStream(
    {
      TableName: 'files',
      FilterExpression: 'attribute_exists(#fileSize)',
      ExpressionAttributeNames: {
        '#fileSize': 'fileSize',
      },
      ProjectionExpression: 'fileSize',
    },
    {concurrency: 1000, chunkSize: 10000, highWaterMark: 10000}
  );

  for await (const items of stream) {
    console.log(items); // 10k items here
  }
})();

Read

Publish

$ git checkout master
$ yarn version
$ yarn publish
$ git push origin master --tags

License

MIT © Shelf