Viewing Client Cache Statistics in 11g

The client cache statistics are available from a dynamic view called client_result_cache_stats$.  This view includes information such as the number of result sets cached, number of cached result sets that were invalidated, and the number of cache hits.

SQL> select cache_id, name, value from client_result_cache_stats$; 

CACHE_ID NAME                              VALUE
-------- ------------------------------ --------
      55 Block Size                          256
      55 Block Count Max                    8192
      55 Block Count Current                 128
      55 Hash Bucket Count                  1024
      55 Create Count Success                  2
      55 Create Count Failure                  0
      55 Find Count                            0
      55 Invalidation Count                    1
      55 Delete Count Invalid                  0
      55 Delete Count Valid                    0

A few important values to monitor in this dynamic view are as follows:

• Block Count Max - a computed value that shows the maximum number of blocks that can be allocated in the result cache.  This is based on the configuration parameters. 

• Block Count Current - the number of blocks currently being used by the client result cache.

• The ratio of Block Count Max to Block Count Current shows the utilization of the result cache. 

  • Create Count Success - the number of cached result sets that did not get invalidated prior to caching the result set.

  • Invalidation Count - the number of cached result sets that got invalidated due to database changes that could affect the result set. 

• The ratio of Create Count Success to Invalidation Count will provide a metric which determines if the use of the result cache is inefficient due to frequent changes on the database object.  As mentioned previously, the client cache should be used for tables that are either read-only or near read-only.  If the result cache is used for tables which have frequent changes like inserts, updates and deletes, it will force the database to spend resources to ensure consistency between the client cache and the database object. 

Further detail about this dynamic view can be found in the documentation for Oracle Database Reference. 

Virtual columns

A virtual column is an expression based on one or more existing columns in the table.  As previously discussed in the section for Virtual Column-Based Partitioning, using a virtual column in a table provides advantages in disk space utilization.  While a virtual column is only stored as metadata, and does not consume physical space, it can be indexed.  The virtual column also contains optimizer statistics and histograms. 

Using a virtual column also simplifies the use of derived columns. This is done by transparently deriving the values instead of requiring the application to calculate and insert an additional value.  It also prevents the need to use a trigger on the table to provide an alternate implementation of this functionality.  Another benefit of using virtual columns in tables is eliminating the need to use views to display derived column values. 

To define a new table with a virtual column, use the following syntax:

create table <table_name>(
   <column_name> <data_type>,
   ?
   <column_name> [<data_type>] [generated always] as (<column_expression>) [virtual]
); 

In order to add a virtual column to an existing table, use the alter table syntax:

alter table <table_name>

add (<column_name> [<data_type>] [generated always] as (<column_expression>) [virtual]);

When defining a virtual column in a table, a DBA could either include the datatype or let the database determine the datatype based on the expression.  The phrases ?generated always? and ?virtual? can be optionally used for syntactic clarity. Keep in mind that the column expression must reference columns defined on the same table.  However, the column expression can refer to a PL/SQL function if the function is designated DETERMINISTIC during its creation.   

As an example, a virtual column might be helpful in a table that stores employee information:

SQL> create table employees(
  2     employee_name varchar2(30),
  3     start_date date,
  4     end_date date,
  5     hourly_rate generated always as (annual_salary/2080),
  6     annual_salary number,
  7     active as (case when end_date is null then 'Y' else 'N' end));

Table created.

SQL> insert into employees
  2     (employee_name,
  3     start_date,
  4     end_date,
  5     annual_salary)
  6  values
  7  ('V.J. JAIN', '01-AUG-2005', NULL, 100000);

1 row created.

SQL> select * from employees
  2  /

EMPLOYEE_NAME   START_DATE END_DATE  HOURLY_RATE ANNUAL_SALARY ACTIVE
--------------- ---------- --------- ----------- ------------- ----------
V.J. JAIN       01-AUG-05            48.0769231        100000 Y

SQL>

While the insert statement required to create a new record in this table only requires four values, a query of this table displays all six values.  This includes the two virtual columns derived from other columns.  The first virtual column, hourly rate, is an expression of the annual salary divided by 2080 hours in a year.  The second virtual column displays if the employee is active by examining the row's end date. 

Virtual columns can be used for partitioning, indexing, constraints and foreign keys.  However, virtual columns cannot be used for index-organized, external, object, cluster, or temporary tables.

Note that you cannot perform updates or deletes directly on virtual columns since they are expressions.