반응형
Query per Second is really important to estimating web site's scale
Let's assume the DAU is approximately 1,000.
24 hour = 1,000
1hour = 1000/20 = 50
60 minutes = 50
1 minutes = 1 requests.
it can simplify the calculation.
but this web sites use the AI requests.
and it takes about 10~15 seconds.
so, this sites can not process the just 1 requests.
and it is wrong to design the architecutures.
but if you use the cache, it can be processed the normaly.
because most of the parts are in the cache.
and cache hit rates are over the 95%.
1 minutes = 60 requests
cache hit reate 95%
not cache rates is 60* 0.05 = 3
so, 1 minutes = 1 requests = 10~15 seconds api limit
Assuming a scenario where the cache hit rate is low...
it can processed normally during 2~3 requests at the same time
반응형
'IT' 카테고리의 다른 글
| stateless architecture (1) | 2026.08.05 |
|---|---|
| 주식 전세계 자금 흐름 분석 (0) | 2026.07.28 |
| 백준 촌수 계산 (0) | 2026.03.22 |
| 맥컬록 피츠뉴런 (0) | 2026.03.20 |
| 파이썬 데이터 코드 분석#1 (0) | 2026.03.17 |