History of MongoDB Engine MMAPV1 (~v2.x) Map based data structure. So it has disadvantage about range-based query, concurrent update. ID is consists fo Diskloc(collection data file name) + offset ...
Don't use DB as Container
DB as container is not good approach (1) Stateless Containers are designed for stateless processes. DB is a stateful system. You have to manage volume. This is so tricky since concurrency and per...
Query execution plan
PostgreSQL Type Description Seq scan Full scan When index doesn’t exist, or expected rows are large (greater than 5~10%) Index sca...
Database Partitioning
What’s Partitioning Split the big table into multiple logical tables in the same table. When to use Partitioning? If table size is bigger and bigger , Indexes should being large, the queries could...
Cache Strategy
1. Cache Aside Application update cache data only when cache miss occurs reading from database When to use Massive read workload Pros Have elasticity in error since cache memory is separated ...
인덱스 컬럼을 어떻게 선택할까?
인덱스란? 데이터의 논리적 분류 인덱스를 잘 고르려면? 1. 분포도가 좋은 컬럼인가? 분포도란? 전체 레코드에서 식별 가능한 수에 대한 백분율 ex) 남녀 성비는 분포도가 50%, 나이는 (1~100세) 1%다. 즉, 백분율이 낮을 수록 분포도가 좋다고 한다. [분포도 = \frac{1}{식별 가능한 수} * 100 (\%)] 선택도 ...
How to decide composite index
What kind of columns are to be included in composite index? Let me introduce my principal about how to choose column of composite index. 1. WHERE clause 2. Always used by ‘=’ If leading colum con...