SQL Load generator/simulator with adjustable probability distribution for keys
Client: A large financial institution
from April 2019
until April 2019
In order to verify the performance claims of AuroraDB
I developed a simulator based on JMeter
that generated a load that is representative of the account events from the banks System Of Record.
The key requirement of this simulator was that the probability distribution of the simulated events followed an exponential profile, with some accounts generating 10000 times more events per hour than most. Another requirement was being able to generate up to 20000 inserts and selects per second, with the inserts in size-adjustable batches.
I used the Groovy language to generate the SQL code for the simulator.