When using Parallel Query, you should selectively apply it to those tables
whose data is well distributed over many physical devices. Also, you should avoid using it for all
tables; as noted earlier, a single query may use all the available parallel query server processes,
eliminating the parallelism for all the rest of the transactions in your database.
Effective Application Design
In addition to the application design topics described later in this chapter are several general
guidelines for Oracle applications.
First, they should minimize the number of times they request data from the database. Options
include the use of sequences, PL/SQL blocks, and the denormalization of tables. You can use
distributed database objects such as materialized views to help reduce the number of times a
database is queried.
246 Oracle Database 11g DBA Handbook
NOTE
Even mildly inefficient SQL can impact your database??™s performance
if it is executed frequently enough. SQL that generates few or no
physical I/O reads still consumes CPU resources.
Second, different users of the same application should query the database in a very similar
fashion. Consistent access paths increase the likelihood that requests may be resolved by
information that is already available in the SGA. The sharing of data includes not only the tables
and rows retrieved but also the queries that are used. If the queries are identical, a parsed version
of a query may already exist in the shared SQL pool, reducing the amount of time needed to process
the query.
Pages:
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412