Part 4 – Using the Diff Router
This multipart series breaks down each section into easy logical steps.
Now that you have the Diff Router ready to be used, we need to have a load on our system, the idea is that this is your standard production load, however, for the purposes of this blog I am going to create my load using sysbench.
If you would like to use sysbench to test your load, you can see the instructions here. Just make sure you change the IP address to use your MaxScale Server.
For my testing, I am going to increase the run time from 60 seconds to 600, so that I can run a sustained load test.
Compare the old server with the new!
Server Load
To generate my server load, I am running SysBench. You should be using your standard production workload.
Enable Service
If you have not yet got the service enabled, you need to re-enable it:
maxctrl call command diff create Diff-Service Application-Service server1 server4
Once your load is running on the server, and you are ready to test the load, you need to start the service.
maxctrl call command diff start Diff-Service
End of Test
Once your testing period has completed. You need to stop the service, to return the MaxScale configuration and stop traffic going through the Diff Router Service.
maxctrl call command diff stop Diff-Service
Generate comparison files
If you have followed the instructions and tested everything correctly, then you should be OK, and you will be ready to create the log files.
You can generate the comparison files at any point by running this command on the MaxScale Server:
maxctrl call command diff summary Diff-Service
You should have an OK returned.
You will find the files in a folder on the MaxScale server to make sure the JSON files containing the information are getting generated.
ls /var/lib/maxscale/diff/Diff-Service/
Once you have finished your testing, you will want to remove the service you added:
maxctrl call command diff destroy Diff-Service
Now that you have the results, let’s have a look at how we review the results to give us confidence in the testing in Part 5 of this series.
Leave a Reply
You must be logged in to post a comment.