Skip to content

Latest commit

 

History

History
52 lines (32 loc) · 1.92 KB

README.md

File metadata and controls

52 lines (32 loc) · 1.92 KB
Raw

TPC-DS tools for Apache Impala

The official and latest TPC-DS tools and specification can be found at tpc.org

The query templates and sample queries provided in this repo are compliant with the standards set out by the TPC-DS benchmark specification and include only minor query modifications (MQMs) as set out by section 4.2.3 of the specification. The modification list can be found in query-templates/README.md.

If you use this repo for any results publication, please see Fair Use of TPC Benchmarks.

Step 0: Environment Setup

Install Java JDK and Maven if need be:

sudo yum -y install java-1.8.0-openjdk-devel maven

Install the necessary development tools:

sudo yum -y install git gcc make flex bison byacc curl unzip patch

Step 1: Generate Data

Data generation is done via a MapReduce wrapper around TPC-DS dsdgen. See tpcds-gen/README.md for more details on the commands to generate the flat files.

Step 2: Load Data

Adjust the source/text and target/Parquet schema names and flat file paths in the sql files found in the scripts/ directory. See the comments at the top of each.

Create external text file tables:

impala-shell -f impala-external.sql

Create Parquet tables:

impala-shell -f impala-parquet.sql

Load Parquet tables and compute stats:

impala-shell -f impala-insert.sql

Step 3: Run Queries

Sample queries from the 10TB scale factor can be found in the queries/ directory. The query-templates/ directory contains the Apache Impala TPC-DS query templates which can be used with dsqgen (found in the official TPC-DS tools) to generate queries for other scale factors or to generate more queries with different substitution variables.