Database Autonomy Service (DAS) provides traffic playback and stress testing to help you verify whether your instance type can handle peaks in service traffic.
Scenarios
-
Determine whether your current database instance type needs to be scaled out before an upcoming traffic peak.
-
Verify the impact of a database schema change, especially an index change, under real-world traffic before publishing it to production.
-
Reproduce a failure on a cloned database to identify the root cause when it cannot be determined at the time of occurrence.
Feature architecture
The following figure shows the architecture of traffic playback and stress testing:
-
Source database: The database from which traffic is captured. This database may be preparing for sales promotions or business changes, or it may have already experienced a failure.
-
Target database: A clone of the source database with the same table schema and data. The captured traffic from the source database is replayed on this database.
Feature introduction
Traditional stress testing tools such as sysbench and TPCC adjust read and write speeds to measure performance and automate data generation, testing, and reporting. However, the SQL statements they generate differ significantly from real service traffic, so their results cannot accurately predict instance performance and stability under production workloads.
DAS traffic playback and stress testing captures real service traffic and dynamically replicates and amplifies production pressure, delivering more objective results for O&M decisions. Instead of a simple client-to-target model, DAS uses a three-point architecture: production database, stress testing client, and test object. This architecture captures production database pressure and realistically replays it on the test object for a specified period. You can also replay traffic at different speeds.
Compared to traditional stress testing solutions, DAS traffic playback and stress testing offers the following advantages:
-
Low-overhead capture of real service traffic.
-
Support for write traffic playback and stress testing.
-
Preserves transaction ordering during replay.
-
Ready to use out of the box.
Prerequisites
-
Supported source databases:
-
RDS MySQL
-
PolarDB for MySQL
NotePolarDB for MySQL Enterprise Edition single-node instances are not supported.
-
-
Supported destination database instances:
-
RDS MySQL.
-
PolarDB for MySQL.
-
PolarDB-X 2.0.
-
-
The source and destination database instances are connected to DAS. For more information, see Connect an Alibaba Cloud database instance to DAS.
-
For the source database instance, DAS Enterprise Edition and the SQL Explorer feature must be enabled. The storage duration for log indexes must be 7 days.
-
For the destination database instance, DAS Enterprise Edition and the SQL Explorer feature must be enabled. The storage duration for log indexes must be 7 days.