Cache data sing larang kanggo dibuat, mula piyambak, lan banget sering dibaca, tur arang-arang owah — iku sing ngasilake manfaat caching paling gedhe. TTL (time to live) minangka trade-off antarane freshness (kepiye saiki data kudu dadi) lan staleness tolerance (kepiye salah dheweke bisa yen nganti pirang lama).
Cadidat cache sing apik ngasilake skor dhuwur ing kabeh telu iki:
Jangan cache sing murah-murah kanggo diitung, write-heavy, utawa data rahasia per-user sing freshness-ne kudu persis (conto: balance akun ing wektu transaksi).
Staleness tolerance → TTL:
config / static reference data → long (hours to days)
product listings, article body → medium (minutes)
prices, stock, leaderboards → short (seconds)
must always be exact → don't cache, or use explicit invalidation
Untuk kebenaran data sing owah ing cara sing ora bisa diprediksi, gabung TTL karo explicit invalidation: write-through utawa delete-on-write supaya cache dibusak sasuwene sumber owah, tinimbang nunggu expiry.
on update(record):
db.save(record)
cache.delete(key(record)) // invalidate now, don't serve stale until TTL
hit ratio = hits / (hits + misses)
high (>90%) → cache is doing its job
low → TTL too short, keys too granular, or data not cacheable
Awasi cache hit ratio kanggo tune TTLs: ratio sing kurang berarti sampeyan arang caching; ratio tinggi karo keluhan stale berarti TTLs dawa banget.
Kakuatan apa-sing-kudu-di-cache lan TTL bener-bener adalah inti caching sing efektif. Cache bab sing salah lan sampeyan nambahe kompleksitas tanpa gain; set TTLs dawa banget lan sampeyan nglayani data stale; sing cendhak lan sampeyan ilang manfaat. Nggunakake staleness tolerance kanggo pilih TTLs, nambah explicit invalidation kanggo kebenaran, lan awasi hit ratio menehi cara sing principled, measurable kanggo cache.