In that example, I bet WordPress on its own would have spent less than 550ms without having to replenish your cache layer. Also how much of that time was actually PHP vs. PHP waiting for MySQL? I bet if you scattered microtimers throughout WordPress, you would see 99% of your 550ms would be waiting for the disk and MySQL.