You are evaluating the performance of a SQL statement that accesses a very large table.
You run this query:
eligible for predicate offload” statistic.
A.
There is an index on the column used in the WHERE clause, causing “cell multiblock physical reads” to be
requested by the database instance, resulting in additional I/O.
B.
The table is an IOT and has an overflow segment, causing “cell multiblock physical reads” to be requested
by the database instance, resulting in additional I/O.
C.
There is an uncommitted transaction that has modified some of the table blocks, causing some “cell single
block physical reads” to be requested by the database instance, resulting in additional I/O.
D.
The table is an index clustered table, causing “cell single block physical reads” to be requested by the
database instance, resulting in additional I/O.
E.
There are migrated rows in the table, causing some “cell single block physical reads” to be requested by the
database instance, resulting in additional I/O.
Explanation:
Note:
physical read total bytes: the size of the segment to read is known by the database, and must be read entirely from the database’s perspective.
cell physical IO bytes eligible for predicate offload: this statistic shows the amount of data which the cell server is able to process on behalf of the database, instead of the database processing and the cell server just delivering blocks.
Cell physical IO bytes eligible for predicate offload — This number should be high
The higher the number more MB/GB is filtered out at the cell level itself rather sending it to the buffer cache to filter the rows.
In this case, all bytes are processed on the cellserver (cell physical IO bytes eligible for predicate offload=physical read total bytes)
Cell Offloading:-
The storage cells are intelligent enough to process some workload inside them, saving the database nodes from that work. This process is referred to as cell offloading.