Partition data for efficient joining for Spark dataframe/dataset

Rainfield picture Rainfield · Jan 9, 2018 · Viewed 19.7k times · Source

I need to join many DataFrames together based on some shared key columns. For a key-value RDD, one can specify a partitioner so that data points with same key are shuffled to same executor so joining is more efficient (if one has shuffle related operations before the join). Can the same thing can be done on Spark DataFrames or DataSets?

Answer

Silvio picture Silvio · Jan 9, 2018

You can repartition a DataFrame after loading it if you know you'll be joining it multiple times

val users = spark.read.load("/path/to/users").repartition('userId)

val joined1 = users.join(addresses, "userId")
joined1.show() // <-- 1st shuffle for repartition

val joined2 = users.join(salary, "userId")
joined2.show() // <-- skips shuffle for users since it's already been repartitioned

So it'll shuffle the data once and then reuse the shuffle files when joining subsequent times.

However, if you know you'll be repeatedly shuffling data on certain keys, your best bet would be to save the data as bucketed tables. This will write the data out already pre-hash partitioned, so when you read the tables in and join them you avoid the shuffle. You can do so as follows:

// you need to pick a number of buckets that makes sense for your data
users.bucketBy(50, "userId").saveAsTable("users")
addresses.bucketBy(50, "userId").saveAsTable("addresses")

val users = spark.read.table("users")
val addresses = spark.read.table("addresses")

val joined = users.join(addresses, "userId")
joined.show() // <-- no shuffle since tables are co-partitioned

In order to avoid a shuffle, the tables have to use the same bucketing (e.g. same number of buckets and joining on the bucket columns).