◦ Designing distributed systems using NALSD flashcards ▪ May 2, 2020 ▪ https://cloud.google.com/blog/products/management-tool s/sre-principles-and-flashcards-to-design-nalsd
reads. • Global shared data is expensive. This is a fundamental limitation of distributed systems. The lock contention in shared heavily written objects kills performance as transactions become serialized and slow. • Architect for scaling writes. • Optimize for low write contention. • Optimize wide. Make writes as parallel as you can. http://highscalability.com/numbers-everyone-should-know Feb, 2009
from one server? • 画像読み込み回数 15回 ◦ 30 images / 2 disks per machine = 15 • HDDから画像1枚を読み込む時間 11.28 ms ◦ (256KB / 1MB) * 5 ms + 10 ms seek = 11.28 ms • トータル時間 169.2 ms ◦ 15 reads * 11.28 ms = 169.2 ms
from one server? • 画像読み込み回数 15回 ◦ 30 images / 2 disks per machine = 15 • HDDから画像1枚を読み込む時間 11.28 ms ◦ (256KB / 1MB) * 5 ms + 10 ms seek = 11.28 ms • トータル時間 169.2 ms ◦ 15 reads * 11.28 ms = 169.2 ms 1000 ms / 169.2 ms ≒ 5 => 1秒あたり 5 つの画像を取得できる