Certificate-Renewal Throughput at Fleet Scale

Ada re-derives this chapter’s own numbers step by step, at full precision

foundations
math-foundations
calculation-audit
auth-labs
Ada ADA · CALCULATION AUDIT

Certificate-Renewal Throughput at Fleet Scale

A utility rolls fresh X.509 certificates across a fleet of 50,000 smart meters inside a 30-day renewal window — 1,667 renewals a day, 42.5 GB of transfer, and $1,500 in EST server fees. The failure table treats a server rate-limited past 69/hour as a real risk, while the steady rate this fleet must actually hold is 69.44/hour. This audit rebuilds each figure at full precision and asks whether that 69.44/hour demand already sits above the 69/hour line, forcing a pre-committed jittered retry rather than a rare edge case.

Companion to the chapter Lab: Access Control Advanced Topics — every number here comes from that chapter.

Ada: The 50,000-meter rotation plan quotes a daily rate, an hourly rate, a total transfer, and a cost. All four fall out of one fleet size and one 30-day window, so let me carry full precision and check whether the hourly figure really lands where the failure table says it does.

The fleet is 50,000 meters renewed over a 30-day window:

  • Daily capacity: 50,000 / 30 = 1,666.6667 renewals/day, which the chapter rounds up to 1,667/day so no meter is stranded past its window.
  • Hourly capacity at full precision: 50,000 / (30 x 24) = 50,000 / 720 = 69.4444 renewals/hour. (Dividing the pre-rounded 1,667/day by 24 gives 69.46/hour; the chapter’s stated 69.5/hour is that same rounding carried one step further – the exact fleet-wide rate is 69.44/hour.)
  • Total transfer: 850 KB x 50,000 = 42,500,000 KB = 42.5 GB at 1 GB = 1,000,000 KB.
  • Server-fee cost: $0.03 x 50,000 = $1,500 for the whole fleet.

The audit’s point is that hourly number. The failure table flags “EST server rate-limited (>69/hour)” as a real scenario – and the steady rate this fleet must sustain, 69.44/hour, already sits above that 69 threshold, which is precisely why the design pre-commits to jittered retry instead of treating the rate limit as a rare edge case.

Every number above is taken from the chapter’s own material and re-derived step by step.