Exadata Smart Scan – Just The Columns I Want
Posted by Tyler Muth on June 23, 2010
In my last post I talked about the Smart Scan component of cell offload processing. I showed how much network traffic this can save between storage and the Database. That savings comes from the fact that the storage filters the results down just the rows and columns you asked for. This is just a quick post to demonstrate the difference in bytes shipped across the wire between selecting one column and selecting all columns.
-- query v$mystat select * from all_objs where object_type = 'TABLE' and object_name like '%RESOURCE%'; -- query v$mystat select object_name from all_objs where object_type = 'TABLE' and object_name like '%RESOURCE%'; -- query v$mystat ALL_COLUMNS_MB ONE_COLUMN_MB FACTOR_OF_IMPROVEMENT ------------------ ------------------ --------------------- 5.4 1.8 3.0
Evgeny said
Hi, Tyler!
I can’t reproduce this example. It seems that if the row is located in several blocks Smart Scan in this case works.
Regards,
Evgeny / RuOUG
sex said
Very good blog ..