[SPARK-27296] Efficient User Defined Aggregators

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.3.3, 2.4.0, 3.0.0
• Fix Version/s: 3.0.0
• Component/s: Spark Core, SQL, Structured Streaming
• Labels: performance usability
• Target Version/s: 3.0.0
• Flags: Important
• Assignee: Erik Erlandson
• Reporter: Erik Erlandson
• Created: 27/Mar/19 23:12
• Updated: 15/Jun/20 21:24
• Resolved: 12/Jan/20 07:38

Description

Spark's UDAFs appear to be serializing and de-serializing to/from the MutableAggregationBuffer for each row.  This gist shows a small reproducing UDAF and a spark shell session:

https://gist.github.com/erikerlandson/3c4d8c6345d1521d89e0d894a423046f

The UDAF and its compantion UDT are designed to count the number of times that ser/de is invoked for the aggregator.  The spark shell session demonstrates that it is executing ser/de on every row of the data frame.

Note, Spark's pre-defined aggregators do not have this problem, as they are based on an internal aggregating trait that does the correct thing and only calls ser/de at points such as partition boundaries, presenting final results, etc.

This is a major problem for UDAFs, as it means that every UDAF is doing a massive amount of unnecessary work per row, including but not limited to Row object allocations. For a more realistic UDAF having its own non trivial internal structure it is obviously that much worse.

Issue Links

• is related to: SPARK-30423 Deprecate UserDefinedAggregateFunction
• relates to: SPARK-30423 Deprecate UserDefinedAggregateFunction
• links to: GitHub Pull Request #25024