02 · Case chapter

SHR · BANKING

Rate calendar at 300% site speed

Caching + TTL for high-latency data. 10k+ SKUs synced. Sub-second responses that held under real traffic.

95%
Cache hit
10k+
SKUs
0.02s
Responses
Abstract illustration of layered rate calendar panes with cache orbs

Problem

Rate data lived behind a slow upstream. Every browse hit the source. The calendar felt heavy; catalogue scale made it worse.

Hardening

Layered caching with deliberate TTLs. Invalidation that matched how rates actually change. Sync for 10k+ SKUs without hammering origin. Guardrails so stale never meant wrong.

Result

~95% cache hit rate. Responses around 0.02s. Site speed up roughly 300%. Banking UX that felt local even when the source was not.

Let's harden your next system