rethinkdb-index-tuning

Demonstrates different types of indexes in RethinkDB and how query performance is affected. Supplementary blog post.

Setup

• First install RethinkDB for your platform.

• Start the RethinkDB server by running this command in your terminal:

  rethinkdb
  

• Install dependencies, seed the database, and create the indexes by running these commands in another terminal:

  npm install
  npm run seed
  npm run setup
  

This will create a database named index_tuning containing a single table named posts.

Running

npm start

This will run setup to create the indexes. Then the benchmark suite will run, testing 4 different queries. All queries return documents in the posts table with a given author and tag. Each query returns the same result but uses different RethinkDB api calls and a different indexes.

1. filter with row commands

r.table('posts')
  .filter(
    r.row('author').eq(author).and(
    r.row('tags').contains(tag))
  ).run(conn);

2. filter with a predicate function

r.table('posts')
  .filter(post => r.and(post('author').eq(author),
    post('tags').contains(tag))
  ).run(conn);

3. getAll with simple index and chained filter

r.table('posts')
  .getAll(author, { index: 'author' })
  .filter(r.row('tags').contains(tag))
  .run(conn);

3. getAll with mapped multi index

r.table('posts')
  .getAll([author, tag], { index: 'authorTags' })
  .run(conn)

The benchmark will output some stats including ops/sec, mean run time, and will declare the fastest and slowest query of the bunch.

Cleanup

To drop the database:

npm run clean