A read-only application is in development and is using a test database on a Database Machine.
You are examining SQL statements from this application in an attempt to determine which ones
will benefit from the Exadata Smart scan capability.
The following is true about the tables used by the application:
1. The data for the tables has just been loaded.
2. There are no applications accessing the tables currently.
3. None of the indexes are compressed or reverse key indexes.
4. The tables use the default organization type data.
5. The only data types for the table are varchar (2), number, or date.
6. The largest number of columns for any table is 29.
7. No access is based on ROWID, or virtual columns.
Which two access paths will always generate either a set of “cell smart table scan” or a set of “cell
smart index scan” requested?
A.
Full scans on sorted hash clustered tables executed in parallel
B.
Full table scans on index organized tables executed in parallel.
C.
Full table scans on heap tables executed in parallel
D.
Full scans on index clustered tables executed in parallel
E.
full scans on hash clustered tables executed in serially
F.
fast full scans on B*Tree indexes executed in parallel
G.
full index scans on B*Tree indexes executed in parallel
Explanation:
Note:
*Exadata is built to efficiently use the resources to return results to the end user. Toward that end
it has been designed to offload query tasks to the storage server, when possible, to reduce the
amount of data passed to the calling session. Certain criteria must be met to offload to the storage
server:
/Full table or index scans must be used
/Direct path reads must be used to access the data
If the first requirement is not met another option can trigger the offloading mechanism:
/Parallel query is in use
*How do you know you’ve used a Smart Scan? The ‘cell smart table scan’ wait event or the ‘cell
smart index scan’ wait event will show activity indicating a Smart Scan occurred. Nothing in the
execution plan would be displayed to prove a Smart Scan was used; querying v$waitstat would be
the only way to confirm that a Smart Scan was implemented.
These types of scans can smart scan be applied.
– Full table scans
– Fast full index scans
– Fast full bitmap index scans
And Smart scan cannot be used on …
– Clustered tables
– Index organized tables
– table with ROWDEPENDENCIES enable
According to “Explanations”,
” The ‘cell smart table scan’ wait event or the ‘cell smart index scan’ wait event
will show activity indicating a Smart Scan occurred.”
Therefore, C and F are correct answer, I think…
Correct Answers are B & F
B is clearly wrong
Smart Scan can not be used on a clustered table or an index organized table. Therefor A, B, D, and E are incorrect. This leaves C, F, and G as possible answers. I’m not really sure which 2 are correct. I think C & F.
Smart Scan is possible only for full segment scans; that is full table scans, fast full index scans and fast full bitmap index scans.
This makes Sugar right! C & F are right.
C, F
Yes C,F right
c f
Smart Scan is possible when
full table scans,
fast full index scans
fast full bitmap index scans.
C & F
A is wrong. No clustered tables can use smart scan
B is wrong. No IOT can use smart scan
C is right. Full table scan on heap tables can use smart scan
D is wrong. No clustered tables can use smart scan
E is wrong. No clustered tables can use smart scan
F is right. Fast full btree indexes scan can use smart scan
G is wrong. full btree index scan is NOT fast full btree index scan
So C.F. are correct