All Products
Search

This Product

    E-MapReduce:SSB performance test

    Document Center

    E-MapReduce:SSB performance test

    Last Updated:Aug 20, 2026

    Star Schema Benchmark (SSB) is an industry-standard tool for evaluating OLAP query performance in data warehouses. This topic walks you through running an SSB performance test on EMR Serverless StarRocks.

    Note

    The SSB implementation in this topic is based on the SSB benchmark. The results are not comparable to published SSB benchmark results.

    Reference results

    The following results were measured on an 8-CU instance (8 cores, 32 GB memory). Each value is the query response time in milliseconds. Lower is better.

    Test environment

    Component Specification
    Compute units (CUs) 8 CUs (8 cores, 32 GB memory)
    Scale factor 1 (1 GB of raw data)

    SSB wide table query (ssb-flat mode)

    Query Time (ms)
    Q1.1 26
    Q1.2 13
    Q1.3 20
    Q2.1 46
    Q2.2 40
    Q2.3 26
    Q3.1 40
    Q3.2 33
    Q3.3 30
    Q3.4 23
    Q4.1 26
    Q4.2 30
    Q4.3 20
    Total 373

    SSB standard table query (ssb mode)

    Query Time (ms)
    Q1.1 26
    Q1.2 20
    Q1.3 23
    Q2.1 103
    Q2.2 53
    Q2.3 50
    Q3.1 86
    Q3.2 46
    Q3.3 46
    Q3.4 36
    Q4.1 83
    Q4.2 63
    Q4.3 53
    Total 688

    SSB overview

    SSB simulates a typical star schema data model with five tables: Date, Customer, Supplier, Part, and Lineorder. It evaluates the following performance metrics of a data warehouse or OLAP system:

    • Query response time

    • Concurrent processing

    • Data loading

    • Query optimization

    SSB also supports customizing the data model — for example, flattening the star schema into a wide table to reduce join overhead.

    The SSB generation tool uses a scale factor (SF) to control data volume: 1 SF = 1 GB of raw data. Index storage is not included, so reserve additional disk space when setting up your environment.

    Prerequisites

    Before you begin, ensure that you have:

    • An Elastic Compute Service (ECS) instance to generate data, load it into StarRocks, and run client queries

    • An EMR Serverless StarRocks instance

    ECS instance requirements

    Requirement Details
    Operating system CentOS or Alibaba Cloud Linux
    Data disk Enhanced SSD (ESSD); capacity based on your target data volume
    Network Same region and Virtual Private Cloud (VPC) as the StarRocks instance
    Public IP Required

    For instructions on creating an ECS instance, see Create an instance.

    StarRocks instance requirements

    The reference results in this topic used 8 CUs (8 cores, 32 GB memory). Adjust compute resources to match your testing needs.

    To create a StarRocks instance, go to the E-MapReduce console, navigate to EMR Serverless StarRocks, and click Create Instance. For detailed instructions, see Create an instance.

    Important

    Use a freshly created instance for each test run. Instances that have been upgraded or downgraded may produce inconsistent results.

    Run the SSB performance test

    Step 1: Download and configure the test toolkit

    1. Log on to the ECS instance. See Connect to an ECS instance.

    2. Download and extract the test toolkit.

      sudo wget https://starrocks-oss.oss-cn-beijing.aliyuncs.com/public-access/starrocks-ssb-benchmark-for-serverless.tar.gz
tar xzvf starrocks-ssb-benchmark-for-serverless.tar.gz

    3. Navigate to the toolkit directory.

      cd starrocks-ssb-benchmark-for-serverless

    4. Open the configuration file.

      vim conf/starrocks.conf

    5. Update the parameters for your environment.

      # for mysql cmd
mysql_host: fe-c-7b6f92276******-internal.starrocks.aliyuncs.com
mysql_port: 9030
mysql_user: admin
mysql_password: 1****
database: ssb_sf

# cluster ports
http_port: 8030

# Specifies whether to enable the bitmap index.
bitmap_index: false

# benchmark config
scale_factor: 1
dataset_generate_root_path: /mnt/disk1/starrocks-benchmark/datasets
      Parameter Description Default
      mysql_host Internal endpoint of the frontend (FE). Find it on the Instance Details tab under Gateway Information > Internal endpoint. —
      mysql_port Query port of the FE. Find it on the Instance Details tab under FE Details > Query Port. 9030
      mysql_user StarRocks instance user. admin
      mysql_password Password for the StarRocks instance user. —
      database Database name to create in the StarRocks instance. ssb_sf
      scale_factor Data volume in GB. 1 = 1 GB of raw data. Make sure your disk has enough capacity for the generated dataset. 1
      dataset_generate_root_path Local path where the generated test dataset is stored. /mnt/disk1/starrocks-benchmark/datasets
      bitmap_index Whether to enable the Bitmap Index feature. Enabling it improves query performance for specific patterns, particularly Q2.2, Q2.3, and Q3.3. false
      http_port HTTP port of the FE. Find it on the Instance Details tab under FE Details > HTTP Port. 8030
      Important

      Set mysql_host to the internal endpoint, not the public endpoint.

    Step 2: Generate and load test data

    Run the following command to create the database, tables, and dataset, and load the data into StarRocks.

    sh ssb_dataload.sh

    Step 3: Run queries and view results

    Run queries using one of the following modes. An overview is printed to the terminal, and a result.csv file with full details is saved in the current directory.

    • Wide table mode — queries the pre-joined flat table:

      sh ssb_query.sh ssb-flat

    • Standard mode — queries the original star schema tables:

      sh ssb_query.sh ssb

    Each value in the output is the query response time in milliseconds. Lower values indicate better performance.

    What's next

    • Adjust scale_factor in conf/starrocks.conf to test with larger data volumes and observe how performance scales.

    • Enable bitmap_index: true and re-run the benchmark to measure the impact of the Bitmap Index feature on Q2.2, Q2.3, and Q3.3.