No shedding vs load shedding
Without shedding, a sale spike queues every request at the API. Load shedding rejects part of the low-priority browsing so checkout still finishes. Checkout reaches 600 requests/s and browsing 1,000. Without shedding, p99 passes 10 s and availability is 84 percent. With it, p99 is 44.5 ms.
Buffer 1,000 jobs · last job waits 1.0 s · 17% turned away
Case 1
A sale starts at 0:20. Checkout rises to 600 requests/s and browsing to 1,000. The shop API serves about 1,000 requests/s. Checkout needs 99.9 percent availability and p99 below 300 ms. Browsing must keep 45 percent.
| Measure | Goal | No load shedding | Rate limit on browsing |
|---|---|---|---|
| p99 latency (Checkout) | under 300 ms | 10.1 s Fail | 45 ms Pass |
| Availability (Checkout) | at least 99.9% | 84.4% Fail | 100% Pass |
| Availability (Browse) | at least 45% | 82% Pass | 53.7% Pass |
| Monthly cost | 8 USD/mo | 9 USD/mo | |
| Error rate | 15.6% | 0% |