Exercise 2: The "Floor" Technique — Possible Solution ==================================================================== WHAT THE TECHNIQUE IS ------------------------------ Per this chapter, "instead of watching the peaks, watch the low points - the quiet periods, night after night, week after week." Rather than comparing how high a metric climbs each day, the technique compares how low it settles back down to during each quiet period. WHY THE FLOOR REVEALS WHAT THE PEAK CAN HIDE ------------------------------ Per this chapter's finding-box, "two servers can show visually similar-looking daily graphs... and still be in completely different states. One returns to exactly the same overnight floor every single night: healthy, bursty, normal. The other's overnight floor creeps very slightly higher each night: a real, sustained problem hiding underneath what looks, at a glance, like the same repeating pattern." Peaks driven by normal daily load can look nearly identical day to day regardless of whether an underlying leak exists, since the peak is dominated by that day's own traffic/activity - the floor, by contrast, reflects what's left over once that day's activity has fully died down, which is exactly where a leak's own accumulation becomes visible. WHY THIS REVEALS A LEAK PEAK-WATCHING WOULD MISS ------------------------------ If only the peaks are being compared, a slowly rising floor underneath similar-looking daily peaks would go completely unnoticed - the peaks alone don't distinguish "returned fully to baseline" from "settled slightly higher than last time." Only checking the low points, repeatedly over time, surfaces that slow upward creep. WHY THIS WORKS AS AN ANSWER ------------------------------ It explains specifically what to compare (successive quiet-period low points, not peaks) and why that specific comparison surfaces a leak that comparing peaks alone would hide, using the chapter's own reasoning about why peaks can look deceptively similar.