That's really a good way to track, but the events I caught were mostly from tempdb. I tried it on another server (server B), which had normal "Full scans/sec" number...