# Appendix I
<a name="appendix-i"></a>

All the tests are performed on an Amazon RDS for MySQL instance running on the `db.r6g.8xlarge` instance class.

The following [sysbench](https://github.com/akopytov/sysbench) commands were used to prepare and run the load on the database.

```
sysbench oltp_read_write --db-driver=mysql --mysql-db=<DATABASE> --mysql-user=<USER> --mysql-password=<PASSWORD> --mysql-host=<ENDPOINT> --tables=500 --table-size=2000000 --threads=500 prepare 
sysbench oltp_read_write --db-driver=mysql --mysql-db=employees --mysql-user=admin --mysql-password=qwertyuiop --mysql-host=mysql8.cbbhujzeoxed.us-east-1.rds.amazonaws.com --tables=500 --rate=500 --time=7200 run
```

In the following graph, an OLTP workload was running, and the pt-archiver process started where arrow is marked.



![\[At the arrow marking the start, the line for the number of rows inserted per second rises slightly.\]](http://docs.aws.amazon.com/prescriptive-guidance/latest/archiving-mysql-data/images/AppendixI_image1.png)


There is no significant change in the CPU utilization with pt-archiver running in parallel, which infers that pt-archiver doesn't impact OLTP queries while running.



![\[The graph shows only a small amount of activity waiting for handler commit\]](http://docs.aws.amazon.com/prescriptive-guidance/latest/archiving-mysql-data/images/AppendixI_image2.png)