While we are performance testing the engineering key, we need to come up with load strategies to commence the test. We'll help to monitor the test, and afterward, we'll help to make all the outcomes, and if they are new, we'll do lots and lots of interpretation and analysis across various servers, to look at response times, and impact. For example, whatever the observations we had during the test, we need to implement it. We'll have to help to catch what exactly is the issues were, and we'll help to see how they can be reduced. Everything is very manual. It's up to us to find out exactly what the issues are. The solution needs better monitoring, especially of CPU.