doubt about CarbonData index and query process

[李寅威]

Hi all:


  I have some doubts about the design of index and the process of Carbondata:


  1. The global table level index(index blocks) is located at the driver of spark app and file level index(index blocket) is located at the executor of spark worker. Are they both rebuild from .carbonindex files at the first query of a table?


  2. Where are the blocket level indexes(index columns/column groups) located and how it effect the efficiency of a query?


  3.When a query request calls, what are the detail steps does carbondata do with the help of the indexes and dictionary maps?


  thx~

[CreativeDataStudio]

If an app fails to engage users, it can lead to lower user retention and a diminished return on investment. Adding interactive elements, personalization features, and rewards can increase user engagement. Frequent updates based on user feedback and market trends can also help keep the app fresh and engaging.