Database Reference
In-Depth Information
Figure 2-4.
The database engine takes advantage of several types of I/O operations
When a data file is stored on an Exadata storage server, a database engine can also take advantage of a fifth type
of disk I/O operation:
smart scans
. Simply put, on an Exadata system, a database engine can use smart scans instead
of direct reads. From a database engine point of view, the main difference between the two disk I/O operations is
that direct reads return regular blocks, and smart scans return different data structures. The aim of smart scans is to
offload part of the processing that would otherwise be done by a database engine to the storage tier. With that in mind,
there are three main goals of using smart scans:
•
To avoid moving irrelevant data between Exadata storage servers and database instances
•
To allow the Exadata storage servers to avoid reading from disk data that isn't required
•
To offload CPU intensive operations to Exadata storage servers and thereby reduce the CPU
utilized by database engines
What'S OraCLe eXaData?
Exadata
is a database machine engineered by Oracle that is composed of database servers, exadata storage
servers, an InfiniBand fabric for storage networking, and all the other components required to host Oracle
Database. however, exadata isn't only a hardware solution. When Oracle Database runs on exadata hardware,
specific software features that are designed to result in better performance are enabled. One of the key design
decisions was to offload part of the processing that would otherwise be done by database servers to the
storage servers.
