外部表加载-Greenplum 数据库基础培训V1.2

时间:2021-04-25 21:28:08
【文件属性】:
文件名称:外部表加载-Greenplum 数据库基础培训V1.2
文件大小:7.92MB
文件格式:PPT
更新时间:2021-04-25 21:28:08
Greenplum 中文 外部表加载 外部表的特征 Read-only(4.0后可写外部表) 数据存放在数据库外 可执行SELECT, JOIN, SORT等命令,类似正规表的操作 外部表的优点 并行方式加载 ETL的灵活性 格式错误行的容错处理 支持多种数据源 两种方式 External Tables: 基于文件 Web Tables: 基于URL或指令 With the other loading mechanism such as COPY and Bizgres Loader, data is loaded through the master in a single process. This does not take advantage of the parallel processing power of the Greenplum segments. External tables provide a means of leveraging the parallel processing power of the segments for data loading. Also, unlike other loading mechanisms you can access multiple data sources with one SELECT of an external table. There is a lot of flexibility when defining an external table as to where, how, and what data will be used. External Tables: Static data, rescannable. Allows file:// or gpfdist:// protocols. gpfdist is a file server program that serves files in parallel. Web Tables: Allow http:// protocol or EXECUTE clause to execute an OS command or script. Data is assumed to be dynamic, meaning that query plans involving web tables do not allow rescanning as the data could change during the course of query execution. Possibly slower plans as data must be materialized (I/O) if can’t fit in memory.

网友评论