Back on the topic - in the mentioned document there's a possible explanation for this. Syncing the cache maybe your problem. I still think it's worth to play with the JFS parameters. In your case I would even suggest you to email Sjoerd.
That I saw... I am going to play with some of those figures to ensure the cache is sync'd. I'll see what that does... though I am sure there's still a lot of fragmentation. And with the sheer volume of data moving back and forth (between web outgoing, logging, mail in both directions, and the ftp server transferring gigs - of sometimes tens of thousands of small to medium sized images) I am guessing that, fragmentation aside, it's gonna have slowdowns when the cache is not sync'd.