An Empirical Evaluation of Columnar Storage Formats
Эмпирическая оценка колоночных форматов хранения данных
2023-10-01
SCID: 54.1/g6fa5nf8
Discuss with AI
Apache ORCApache Parquetbenchmark evaluationcolumnar storage formatsdictionary encoding
Figures from the paper
Abstract (AI)
Columnar storage is a core component of a modern data analytics system. Although many database management systems (DBMSs) have proprietary storage formats, most provide extensive support to open-source storage formats such as Parquet and ORC to facilitate cross-platform data sharing. But these formats were developed over a decade ago, in the early 2010s, for the Hadoop ecosystem. Since then, both the hardware and workload landscapes have changed. In this paper, we revisit the most widely adopted open-source columnar storage formats (Parquet and ORC) with a deep dive into their internals. We designed a benchmark to stress-test the formats' performance and space efficiency under different workload configurations. From our comprehensive evaluation of Parquet and ORC, we identify design decisions advantageous with modern hardware and real-world data distributions. These include using dictionary encoding by default, favoring decoding speed over compression ratio for integer encoding algorithms, making block compression optional, and embedding finer-grained auxiliary data structures. We also point out the inefficiencies in the format designs when handling common machine learning workloads and using GPUs for decoding. Our analysis identified important considerations that may guide future formats to better fit modern technology trends.
Key Findings
1
Making block compression optional and embedding finer-grained auxiliary data structures are identified as beneficial design choices for contemporary workloads.
2
Parquet and ORC exhibit inefficiencies for common machine learning workloads and GPU-based decoding.
3
The benchmark identifies modern-hardware advantages from default dictionary encoding and integer encodings that prioritize decoding speed over compression ratio.
4
The findings provide design considerations for future columnar formats aligned with modern hardware, data distributions, and workload trends.
5
The study empirically evaluates Parquet and ORC, widely adopted open-source columnar storage formats, under varied workload configurations.
Research Object
Open-source columnar storage formats Parquet and ORC
Research Subject
Performance and space efficiency of Parquet and ORC under modern hardware, real-world data distributions, machine learning workloads, and GPU decoding, including the impact of their internal design decisions
Publication Details
Publication Date
2023-10-01
Journal
Publisher
ISSN
Cited by
47
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai4
Good to the Last Bit: Data-Driven Encoding with CodecDB2021
BtrBlocks: Efficient Columnar Compression for Data Lakes2023
The FastLanes Compression Layout: Decoding > 100 Billion Integers per Second with Scalar Code2023
The impact of columnar file formats on SQL‐on‐hadoop engine performance: A study on ORC and Parquet2019