We are currently running 3 slave databases with hyperdb for our Wordpress sites. According to NLoad, each server outputs about 50mbit/s where there are 30,000 people real-time on our websites. This totals to 150 mbit/s SQL traffic, which kinda sounds insane to me.
We have reverse nginx servers in front of our apaches and we are caching the pages for at least 60 seconds. Some pages are even cached between 2 to 10 minutes. So how come are we generating 150 mbit/sec SQL traffic?
What do you guys think?