Creating Star Schema from snowflake

[justinrush]

I have built an application to monitor database performance and run scalability tests to verify the scalability of various hardware/software configurations. I have a very strong background in traditional relational database structures and as such my database backend for this tool utilizes this type of data structure. I would like to create a star schema from this to facilitate the creation of cubes and other traditional OLAP structures that aid with analysis of mass quantities of data. However, I am having a problem wrapping my head around the proper way to do this so I am seeking some help/suggestions...

As an example, here are three tables indicative of my data structure:

performance_runs
-----------------
run_id
set_id
start_time
end_time
cores
ram
db_size

io_stat
----------
run_id
set_id
sample_dt
disk_name
reads_per_sec
writes_per_sec

vm_stat
--------
run_id
set_id
sample_dt
cpu_system
cpu_user



Each run consists of 1 or more sets, every few runs something is changed and performance is measured in between. Currently I use several large, ugly, stored procedures to calculate various averages and percentile numbers. I would like to be able to slice and dice based on any of the hundred or so metrics we are collecting and drill through between runs and sets.

How would one go about converting this to a star schema with facts and dimensions?

I apologize for the long post, but I wasn't sure how much information to post. If I need to post more, please let me know.