TPC-DS performance test

更新时间:
复制 MD 格式

This topic describes how to run a performance test on StarRocks using the TPC-DS benchmark.

Prerequisites

The StarRocks instance is created. For more information, see Create an instance.

Background

TPC-DS is a decision support benchmark that models several common aspects of a decision support system, including queries and data maintenance, to measure the analytical performance of big data products.

TPC-DS models the sales and returns operations of a retail business across three sales channels: physical stores, the Internet, and catalogs. In addition to tables that model sales and returns, it includes a simple inventory system and a promotion system. This test case uses a benchmark that contains 99 complex queries and a dataset that contains 24 tables. The test uses two dataset sizes, 1 GB and 100 GB, as detailed in the following table.

Note
  • The TPC-DS implementation in this topic is based on the TPC-DS benchmark but is not fully compliant. Therefore, these test results are not comparable with officially published TPC-DS benchmark results.

  • Recommended instance specifications:

    • 1 GB: We recommend one frontend (FE) instance with 8 compute units (CUs) and three backend (BE) or compute node (CN) instances, each with 8 CUs.

    • 100 GB: We recommend one frontend (FE) instance with 8 compute units (CUs) and three backend (BE) or compute node (CN) instances, each with 16 CUs.

Data size

Description

Table names

100 GB

A TPC-DS performance test on a 100 GB dataset.

call_center

catalog_page

catalog_returns

catalog_sales

customer_address

customer_demographics

customer

date_dim

household_demographics

income_band

inventory

item

promotion

reason

ship_mode

store

store_returns

store_sales

time_dim

warehouse

web_page

web_returns

web_sales

web_site

1 GB

A TPC-DS performance test on a 1 GB dataset.

Step 1: Go to SQL Editor

  1. Go to the StarRocks Manager page.

    1. Log on to the E-MapReduce console.

    2. In the left navigation pane, choose EMR Serverless > StarRocks.

    3. In the top menu bar, select the required region.

    4. Click StarRocks Manager, or click Connect in the Actions column of your target instance.

      For more information about how to connect to a StarRocks instance, see Connect to a StarRocks instance by using StarRocks Manager.

  2. In the left-side navigation pane, click SQL Editor.

  3. The Queries page lists the TPC-DS performance test cases.

Step 2: Run the test

The following sections show you how to run the performance test on the 100 GB and 1 GB TPC-DS datasets.

TPC-DS 100 GB

  1. Initialize the database and tables.

    1. Click TPC-DS-100G - 01.Initialize database and tables to open the SQL script.

    2. Click Run. The script initializes the database and tables.

  2. Load the test data.

    1. Click TPC-DS-100G - 02.Load test data to open the SQL script. The script loads a 100 GB dataset.

    2. Click Run. The script loads the test data.

  3. Execute the test queries.

    1. Click TPC-DS-100G - 03.Execute test queries to open the SQL script. The queries run against a 100 GB dataset.

    2. Click Run. The results appear in the output.

TPC-DS 1 GB

  1. Initialize the database and tables.

    1. Click TPC-DS-1G - 01.Initialize database and tables to open the SQL script.

    2. Click Run. The script initializes the database and tables.

  2. Load the test data.

    1. Click TPC-DS-1G - 02.Load test data to open the SQL script. The script loads a 1 GB dataset.

    2. Click Run. The script loads the test data.

  3. Execute the test queries.

    1. Click TPC-DS-1G - 03.Execute test queries to open the SQL script. The queries run against a 1 GB dataset.

    2. Click Run. The results appear in the output.