Please enable JavaScript.

Coggle requires JavaScript to display documents.

Presto, " If it is good enough for facebook it is good enough for you…

- - - - ANSI compatible sql statement
      - presto converts statement to worker tasks
      - workers converts statements to queries
    - - queries is divided into stages, tasks, splits and connectors
      - interconnected stages running on workers
    - - the distributed work to execute the query
      - combined together to get the result of a query as a tree
      - implemented as a series of tasks distributed over workers
    - - sections of datasets
      - lowest level in the architecture
      - the coordinators query a connector for splits of a table
    - - acts on splits
      - the work horse of presto
      - executed in parallel
      - has inputs and outputs
    - - Task contain one or more parallel drivers
      - the physical set of operators in memory
      - lowest level of parallelism
    - - consume transform and produce data
    - - transfer data between presto nodes
  - - - adapt presto to data sources
      - every catalog is associated with a specific connector
      - Ex. Hive connector, system, JMX, TPCH, Kafka
    - - contains schemas
      - references a datasource through connector
      - Ex. Hive, JMX ...
    - - a way to organize tables
      - catalog and schema define tables to be queried
    - - set of unordered rows
      - the mapping from source data to tables is is defined by the connector
  - - - parsing statements
      - planning queries
      - managing workers
      - communicate with clients and workers using REST API
    - - execute tasks
      - process data
      - fetch data from connectors and exchange with each other