As mentioned in my previous post, you should run the log parser on logs for a specific day. You can do that by copying the log files from the different load balanced Front-end servers  to a centralized location and run the logparser queries in there.

Since you will get the results for a specific day, then the choice of that day is critical. Weekend days usually have lower use than the weekdays. Moreover, sometimes there are low months due to higher vacation rates or business reasons. A good sample for this is summer months where the number of users using the system is lower than other months.

A good strategy is to apply the scripts for different days to get clearer view on the current system usage. Keep in mind when designing the new environment that you need it to be able to handle the peak load at the peak period; moreover, you need to keep room for future growth. Thus, it is important to identify peak period and apply usage analysis on it.

 

The output of logparser queries created in the last blog is in tabular format, but you can use them to build graph that would highlight the required values. Samples of the results of the pervious queries in graph format are as follows:

Peak Requests Per Second (Peak RPS)

 

Average Requests Per Second (Average RPS)

Distinct Users Per Hour:

 

Using these graphs and the accompanying tables, it should be clear how to build the modeling tables mentioned earlier. They should be enough for identifying the current load on the system and be the basis for estimating the future load for the new system.