- 框架层面 框架焦点在于处理流程，系统交互，性能等
- 数据流 数据流关注于数据的转换、合并、shuffle、结果
- Master 中心节点，记录worker、Driver信息
- Worker 任务执行的节点，取Master发送的Driver，运行任务
- Client SparkSubmit启动，把用户的application(包含main的类)打包，分装成Driver, 提交给Master
- Driver Client向Master注册，Master随机取非满负荷的Worker，绑定Worker和此Driver，向此Worker发送任务消息
Cluster Programming Models distributed storage abstraction
To recap previous chapter, RDDs provide the following facilities:
- Immutable storage of arbitrary records across a cluster (in Spark, the records are Java objects).
- Control over data partitioning, using a key for each record.
- Coarse-grained operators that take into account partitioning. • Low latency due to in-memory storage.
As long as each batched record fits in the CPU cache, this leads to fast transformations and conversions.