This project includes the data set for Redshift benchmark.

• SlideShare
  • amazon redshift is 10x faster and cheaper than hadoop hive (English)
  • amazon redshift is 10x faster and cheaper than hadoop hive (Japanese)

• Test Data
  • s3://hapyrus-examples/redshift-benchmark/ad-network-examples/case-01/ad_campaigns
  • s3://hapyrus-examples/redshift-benchmark/ad-network-examples/case-01/advertisers
  • s3://hapyrus-examples/redshift-benchmark/ad-network-examples/case-01/publishers
  • s3://hapyrus-examples/redshift-benchmark/ad-network-examples/case-01/imp_logs
  • s3://hapyrus-examples/redshift-benchmark/ad-network-examples/case-01/click_logs

• Redshift cluster
  • required to a launch a Redshift cluster
  • minimum instance type is enough
    • dw.hs1.xlarge single node
  • see Amazon Redshift Getting Started Guide
• Local environment
  • postgresql client

• Create tables
  • run sql/create_tables_redshift.sql on your Redshift cluster
• Copy the test data set on our s3 to your Redshift cluster(it took over 17 hours in our case)
  • edit sql/copy_all_[data-size].sql and write your own [aws-access-key-id] and [aws-secret-access-key].
  • run sql/copy_all_[data-size].sql on your Redshift cluster
• Run the test sql
  • run sql/test-query.sql on your Redshift cluster

• See sql/create_tables_hadoop_hive.sql to create Hive tables.
• Run the test sql
  • see sql/test-query.sql

This project is liscensed under the Apache License, Version 2.0 and powered by Hapyrus Inc.