Scaling a star-schemed tables and separate tables to exabyte of data on Vertica?
Vertica's Wikipedia entry markets itself to be able to scale itself to exabyte of data.
Suppose a traditional star-schemed relational database without loops. The db has
separate tables and some tables with common keys. I could combine all tables into
one table by introducing an extra column called "Category" like in the picture.
- Is this reshaping called "column-oriented"?
- Does the column-oriented way of thinking infer to make separate tables into one wide table?
- Does Vertica lie on some other ideas or is this the basic logic behind Vertica (to make straight tables into wide tables and merge everything into one big columnar table)?
- Is it possible to scale traditional star-schemed tables and separate tables from SQL db to exabyte of data on Vertica?
- What are common approaches to convert a working relational db into a columnar db of Vertica?
- What are ways by which Vertica stores its non-zero entries and zero entries? Can I force Vertica to store everything as key-value -pairs or as some sparse data structure? Which sparse data structures does Vertica offer to merge many different tables together that result into many zero/NA entries?