Database Autonomy Service (DAS) provides the intelligent stress testing feature. This feature helps you check whether you need to scale up or scale out your instance to effectively handle workloads during peak hours.

Scenarios

  • Check whether your database instance needs to be scaled up or scaled out to handle workloads during peak hours.
  • After the schema, especially the indexes, of a database is changed, check the performance of the database in real business scenarios to reduce the risk of failures.
  • After a database failure occurs, use this feature to recur the failure on a cloned database of the source database to identify the failure cause.

Architecture

The following figure shows the architecture of the intelligent stress testing feature.s
  • Source database: a database that will experience a significantly increased workload or a business change, or a database whose traffic needs to be captured to identify failures.
  • Destination database: a cloned database of the source database, which is used to play back the traffic captured from the source database.

Overview

Conventional stress testing solutions are implemented based on existing stress testing tools such as sysbench and TPCC. Users can monitor how the performance of the tested object changes with different levels of stress by changing the rate of data reads and writes in restricted scenarios. These solutions automate operations such as data generating, stress testing, and report analyzing. However, the testing results cannot indicate how the performance of a database instance changes in a real business scenario due to the significant differences between the SQL statements used by the stress testing tools and those executed in real business scenarios.

In comparison, the intelligent stress testing feature uses business traffic captured from production databases to recur the same or higher stress. Therefore, the testing results of DAS are convincing and more helpful for users to make decisions on business O&M. The intelligent stress testing is built based on the sophisticated architecture that consists of a production database, a stress testing client, and the tested objects. This is designed to capture the traffic of the production database so that real traffic stress during a time period can be simulated for the tested object. In addition, you can specify the palyback speed of the traffic.

Compared with conventional stress testing solutions, the intelligent stress testing feature provided by DAS has the following advantages:
  • Traffic capture with small loads
  • Stress testing by playing back traffic
  • Stress testing with the original transaction sequence
  • Ready-to-use

Prerequisites

Make sure that your database instances meet the following requirements:
  • The source database instance is an ApsaraDB RDS for MySQL instance of the High-availability Edition, an ApsaraDB RDS for MySQL instance of the Enterprise Edition, a PolarDB for MySQL cluster of the Cluster Edition, a PolarDB for MySQL cluster of the Archive Database Edition, or a PolarDB-X 2.0 instance.
  • The destination database instance is an ApsaraDB RDS for MySQL instance, a PolarDB for MySQL cluster, or a PolarDB-X 2.0 instance.
  • The source instance and the destination instance are connected to DAS. For information about how to connect database instances to DAS, see Connect an Alibaba Cloud database instance to DAS.
  • DAS Professional Edition is enabled for the source instance and the destination instance. For more information, see Usage notes on DAS Professional Edition.
Note PolarDB-X 2.0 is supported only in the China (Hangzhou), China (Shanghai), China (Beijing), China (Shenzhen), and Singapore (Singapore) regions.