I was debugging an issue with an engineer at Broadcom once. He opened a log in Notepad and applied some simple search & replace transformations to turn it into CSV, then graphed the data in Excel. I was blown away by how simple it was. Ever since then I do the same thing. I don't use Notepad, I use sed, or perl, or python, depending on my mood and the complexity, but the basic idea is the same: transform a noisy log into CSV to isolate the number I am investigating.