# MIT6.824 Distributed System

# 1. Intro

  • parallelism
  • Fault tolerance
  • physical
  • Security / isolated

Challenges

  • concurrency
  • partial failure
  • performance

# MapReduce

# 2. RPC& Threads

  • Go
    • type safe & memery safe
    • Garbage collected
  • 线程:
    • 不同的线程拥有不同的栈空间, 程序计数器,寄存器
    • 不同的线程在同一个地址空间
    • 不同线程之间可以通过某种方式访问对方的栈空间
  • Thread ?
    • IO concurrency
      • 为每个RPC调用创建一个线程
    • Parallelism
    • Convenicence
  • OR event-driven 事件驱动编程

# 3. GFS

  • BIG STORAGE
  • Why Hard
  • Performance -- > Sharding
  • Faults --> Tolerance
  • Tolerance --> Replication
  • Replication --> In Consistency
  • Consistency -> Low Performance
Last Updated: 1/16/2021, 9:27:12 AM