Geo Distributed Analytics Architecture

tags: #geo-distributed

• Architecture of geo-distributed-systems
  • There is a central master where queries are submitted; this is for SparkSQL, HiveQL, Pig Latin, etc
  • Query Optimizer at master is responsible for preparing Query Execution plan.
  • There is also a centralized scheduler which places tasks in Nodes
    • placement is based on resource availability and task dependencies (upstream/ downstream, etc)