How the parallel star join works?
(Download Reference Research Paper)
Star Join Works Involving:
i) JDI (Join DataIndexing)
ii) i-th dimension tables
iii) Use of conventional relational approach
As the situation is a process so described to the point as follows:
They proposed and demonstrated the implementation by partitioning the table into sub-tables. For example the SALES table sub-divided into further 7 tables and SALES table still associated with fact table. Its dimensional key the primary key of SALES table is linked with the foreign key which exists in fact table helps to locate the desired row or column partitioned.
Parallel star join demonstration describes that by dividing the tables in fact table and dimension tables enhance the query performance but by sub-dividing the dimension tables according to the rest of dimension tables add a lot in it.
In research paper SALES table although was divided into further 7 tables but still one attribute was associated for the related dimension table too. We can get exact idea of star join working by that query
SELECT TIME.Year, TIME.Month, average(SALES.Discount)
FROM TIME, SALES
WHERE TIME.TimeKey = SALES.shipDate
GROUP BY TIME.Year, TIME.Month
Conventional relational approach is used. As I mentioned before that although dimension tables are sub-divided into further small tables but still the association among dimension tables is maintained. Like in this query the tables TIME and SALES are presented in association by implementing through the primary key/foreign key relationship linking the ShipDate and TimeKey. The approach as used with join and DataIndexing so in research paper it is called as JDI. If we use the approach DataIndexing without star join so it will be BDI, Basic DataIndexing.
Last updated: March 19, 2014