Another way that this can be done is via the Tomcat access logs. (http://tomcat.apache.org/tomcat-5.5-doc/config/valve.html) By default the access logs do not capture "Time taken for request", but this can be easily configured (I can send you the configuration that we are using on some servers.

How this works is that for each page request to the server, the server will calculate the time take to process the request(start time=when server receives request from the browser. end time=when the server closes the network connection to the browser). There are a few things to note about using this number

1) This number is calculated from the point of view of server, so it does not include things like the time taken for the browser to render the page once it has received all the data, or images/data on the page that are coming from another website

2) A request to the server for a given page can be comprised of number of different server request; one for the page itself, another for the CSS files, etc. All of these individual requests will need to be rolled up to determine the "total request" time for your view.
  - The logic for how to rollup this information is pretty well known and available in many open-source projects (awstats is a good one)


There are many open-source and commercial solutions available that can help you calculate, manage and build reports about this information. I will send you an email and we can discuss some of the options.