SQL Server Always on Availability Group Scalability Experiment and Result on Azure VM


The purpose of this scalability experiment is to evaluate the performance of AG cluster when it has large number of databases and some of the databases undergo large amount of traffic, and measure 1) OLTP transaction performance, 2) Failover success rate and time, and 3) Resource consumption, such that we can provide reference for customer’s large scale AG cluster configuration based on the experiment result. For an AG cluster, based on the specific configuration of CPU cores, memory, and disk capacity, we want to know the maximum number of databases an AG can maintain healthy state without breaching resources