Udemy pragmatic-system-design

https://colin-scott.github.io/personal_website/research/interactive_latency.html

Tutorial https://commscope1.udemy.com/course/pragmatic-system-design/learn/lecture/23340674#notes

Estimation

Component	Reads per second	Writes per second
RDBMS	10,000	5,000
No-SQL	10,000 - 50,000	10,000 - 25,000
Distributed Cache	100,000	10,0000
Queue	100,000	100,000

Componenent	Space Size
RDBMS	3TB
No-SQL
Distributed Cache	16GB-128GB

Caching - 

Readthru (commonly used)

Benefits - 1. reduce cpu consumption by caching result of complex algor  2. reduce io consumption by caching resource

 

 

Writethru, only write cache when there is a CUD operation, might be wasted if the data is not visited after update.

Benefit of in memory cache (ease of maintainence, timely, no network overhead)   Satisfy AP (Service Availability and Partition Tolerance) 

Drawback of in memory cache: Lacking consistency  

 

Beneift of dist cache (Highly consistence) Satisfy CP (Service Availability and Partition Tolerance)