How group by works internally

Question

Hi,
I would like to know how group by works internally using data and index cache( informatica uses data and index cache... not sure oracle uses the same). Consider the following table T with columns A and B.
A      B
A1    1
A2    2
A1    3I am selecting SUM(B) GROUP BY A
My understanding is as follows
1. Reads row 1 and stores in data cache.
2. scans entire table to find the A1s. As an when a row is found, cache them to data cache.
3. Perform SUM for A1 and store result in index cache.
4. Above steps repeated for A2 also.
Please let me know if this is true.
Thanks,
Saff
Edited by: saffron on Nov 19, 2009 9:18 AM

Answer 1

saffron wrote:
I would like to know how group by works internallyWhy?
using data and index cache( informatica uses data and index cache... not sure oracle uses the same).It caches data and indexes but it doesn't use those terms
I am selecting SUM(B) GROUP BY A
My understanding is as follows
1. Reads row 1 and stores in data cache.
2. scans entire table to find the A1s. As an when a row is found, cache them to data cache.
3. Perform SUM for A1 and store result in index cache.
4. Above steps repeated for A2 also.
Please let me know if this is true.
Very unlikely.
The grouping algorithms are internal, undocumented and subject to change and of little use to know outside of the Oracle development team.
If you just want to learn about Oracle you would probably be better off reading about the documented features for your database version.
http://tahiti.oracle.com/