About SSB Performance Test
SSB (Star Schema Benchmark) is a lightweight performance test suite for data warehouse scenarios. It simplifies the TPC-H snowflake schema into a star schema and changes the benchmark queries from TPC-H's complex Ad-Hoc queries to more structured OLAP queries. This suite is primarily used to test the performance of multi-table join queries under the star schema model. The Tencent Cloud TCHouse-C team flattened the SSB star schema into a wide table, creating a single-table test set (hereinafter referred to: SSB FLAT) to evaluate query engine performance. This document presents the performance test results of Tencent Cloud TCHouse-C on the SSB single-table dataset.
Performance Testing Scheme
Monitoring Tools
Use the cluster monitoring page of Tencent Cloud TCHouse-C to obtain cluster and node monitoring information.
Test Content
Step 1: Use the dbgen tool to initialize data of a specified size and store it on the disk space of the cluster to be tested.
Step 2: Purchase a Tencent Cloud TCHouse-C cluster on Tencent Cloud, and create the required tables within that cluster.
Step 3: Import the data generated in Step 1 into the test cluster.
Step 4: Execute the benchmark performance test SQL Q1.1-Q4.3 commands.
Step 5: Execute the benchmark performance test SQL Q1.1-Q4.3 commands three times repeatedly.
Step 6: Record the data volume, time, and speed from the three benchmark SQL executions.
View the result
After the benchmark SQL execution, the following information is output, including: the number of rows in the query result, the time consumed by the SQL query execution, the number of records queried by SQL, the data volume queried by SQL, the number of data records rate of the SQL query, and the data volume rate of the SQL query.
This test uses the SQL query execution time (s) metric as the output metric.
Test Environment
Hardware Environment
This document conducts performance tests for three machine scenarios: Standard + Premium Disk, Standard + Enhanced SSD, and High-Performance + NVMe SSD Local Disk. The specific configurations are as follows:
Machine and scenario combination | Node Specifications |
Standard Instance + Premium Disk | CPU:32 cores Memory: 128 GB Disk: Premium Disk 3000 GB |
Standard Instance + Enhanced SSD CBS | CPU:32 cores Memory: 128 GB Disk: Enhanced SSD CBS 3000 GB |
High-performance Instance + NVMe SSD Local Disk | CPU:32 cores Memory: 128 GB Disk: NVMe SSD Local Disk 7140 GB * 2 |
Software Version
Tencent Cloud TChouse-C 24.8.14.39
Test data set
With the -s 1000 parameter, the generated dataset size is:
SSB Table Name | Number of rows | Remarks |
LINEORDER | 6 billion | Product Order Details Table |
CUSTOMER | 30 million | Customer Information Table |
PART | 2 million | Parts Information Table |
SUPPLIER | 2 million | Supplier Information Table |
DATE | 2,556 | Date Table |
LINEORDER_FLAT | 6 billion | SSB Flattened Wide Table |
Performance Test Results (Unit: Seconds)
Query Number | Standard Instance + Premium Disk | Standard Instance + Enhanced SSD CBS | High-performance Instance + NVMe SSD Disk |
|
|
|
|
|
|
|
|
|
Q1.1 | 0.892 | 0.850 | 0.594 |
|
|
|
|
|
|
|
|
|
Q1.2 | 0.088 | 0.083 | 0.066 |
|
|
|
|
|
|
|
|
|
Q1.3 | 0.037 | 0.036 | 0.033 |
|
|
|
|
|
|
|
|
|
Q2.1 | 6.430 | 6.300 | 4.271 |
|
|
|
|
|
|
|
|
|
Q2.2 | 5.266 | 5.348 | 3.518 |
|
|
|
|
|
|
|
|
|
Q2.3 | 4.813 | 4.891 | 3.112 |
|
|
|
|
|
|
|
|
|
Q3.1 | 6.262 | 6.144 | 3.831 |
|
|
|
|
|
|
|
|
|
Q3.2 | 5.203 | 5.243 | 3.438 |
|
|
|
|
|
|
|
|
|
Q3.3 | 3.634 | 3.567 | 2.374 |
|
|
|
|
|
|
|
|
|
Q3.4 | 0.074 | 0.072 | 0.061 |
|
|
|
|
|
|
|
|
|
Q4.1 | 8.667 | 8.713 | 5.785 |
|
|
|
|
|
|
|
|
|
Q4.2 | 2.459 | 2.434 | 1.386 |
|
|
|
|
|
|
|
|
|
Q4.3 | 2.512 | 2.559 | 1.564 |
|
|
|
|
|
|
|
|
|
Total time | 46.337 | 46.24 | 30.033 |
|
|
|
|
|
|
|
|
|