[PARQUET-131] Vectorized Reader In Parquet

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: parquet-mr
• Assignee: Zhenxiao Luo
• Reporter: Zhenxiao Luo
• Created: 11/Nov/14 00:25
• Updated: 23/Jun/24 03:27

Description

Vectorized Query Execution could have big performance improvement for SQL engines like Hive, Drill, and Presto. Instead of processing one row at a time, Vectorized Query Execution could streamline operations by processing a batch of rows at a time. Within one batch, each column is represented as a vector of a primitive data type. SQL engines could apply predicates very efficiently on these vectors, avoiding a single row going through all the operators before the next row can be processed.
As an efficient columnar data representation, it would be nice if Parquet could support Vectorized APIs, so that all SQL engines could read vectors from Parquet files, and do vectorized execution for Parquet File Format.

Detail proposal:
https://gist.github.com/zhenxiao/2728ce4fe0a7be2d3b30

Issue Links

• is related to: HIVE-8128 Improve Parquet Vectorization

Sub-Tasks

• [Vectorized Reader] Support Complex Types (Map, Array, Struct) in Parquet Vectorized Reader: In Progress
• [Vectorized Reader] ColumnVector length should be in terms of rows, not DataPages: In Progress
• [Vectorized Reader] Make sure all encodings work in Parquet Vectorized Reader: In Progress
• [Vectorized Reader] Lazy Load in Vectorized Reader: Open
• [Vectorized Reader] Lazy Decoding in Vectorized Reader: Open
• [Vectorized Reader] Add Testcases/Benchmarks for ParquetVectorizedReader: In Progress
• [Vectorized Reader] Add attributes in ColumnVector and RowBatch: Open