The document details advanced database administration topics including partitioning strategies, additional data structures, and query optimization techniques for Oracle databases. It covers various partitioning methods, such as interval and composite partitioning, their benefits, and the usage of sequences, synonyms, and clusters for data management. Additionally, it explains query optimization processes, contrasting rule-based and cost-based optimization approaches, highlighting the importance of statistics in determining optimal execution paths.

1. 1
Database Administration
& Management
Lecture 11
Chapter 4 – Additional Data Structures and Query
Optimization

2. 2
Partitioning
Divide the information in tables or indexes among multiple
physical storage areas.
Data is divided based on column values.
• Range of values (usually date and numbers)
• List of values
• Hash Function
Partition Advisor can recommend a partitioning strategy for
a table(s) that can help to improve the performance of
Oracle Database.

3. 3
Partitioning
Interval partitioning
• Automatic generation of a new partition after a fixed interval
or range when data to be inserted does not fit into existing
partition ranges.
Composite partitioning
• Defines two levels of partitions using a combination of list,
range and hash partitioning.
Equipartitioning
• Partition a table and an index identically.
• Automatic implementation by specifying an index for the
partitioned table as LOCAL index.

4. 4
Partitioning - Benefits
Faster SQL
• Oracle won’t access partitions that do not contain any data to
satisfy the query.
Reduce scope of maintenance operations.
Increased data availability.
Data partitions based on the time span are specially useful
in a data warehouse environment.

5. 5
Additional Data Structures
Sequences
Synonyms
Clusters
Hash Clusters

6. 6
Sequences
 Sequence object is used to generate unique numbers for use as keys
or identifiers.
 Whenever a value is requested from a sequence, it returns a value
and increments its internal value.
 Sequence numbers are defined with a name, an incremental value,
and some additional information about the sequence.
 Sequences exist independently of any particular table, so more than
one table can use the same sequence number.
 Syntax:
CREATE SEQUENCE sequence_name
MINVALUE value
MAXVALUE value
START WITH value
INCREMENT BY value

7. 7
Synonyms
Synonym is an alternative name for a table, view, sequence,
procedure, stored function, package, materialized view, Java
class schema object, user-defined object type, or another
synonym.
Qualified names for objects can get confusing, synonym can
be created for any table or view to make name simpler and
more readable.
Synonyms simplify user access to a data structure.
Public synonym - all users of a database can use it.
Private synonym - only the user whose schema contains the
synonym can use it.

8. 8
Clusters
 Cluster is a data structure that improves retrieval performance by
storing related data values together on disk.
 Storing related values together reduces the number of I/O
operations needed to retrieve related values.
 A cluster is composed of one or more tables.
 The cluster includes a cluster index, which stores all the values for
the corresponding cluster key.
 Each value in the cluster index points to a data block that contains
only rows with the same value for the cluster key.
 Clusters may not be appropriate for tables that regularly require
full table scans.

9. 9
Hash Clusters
Each request for data in a clustered table involves at least
two I/O operations, one for the cluster index and one for the
data.
A hash cluster stores related data rows together, but groups
the rows according to a hash value for the cluster key.
Each retrieval operation starts with a calculation of the hash
value and then goes directly to the data block that contains
the relevant rows.
A hash-clustered table is faster for retrieving data than a
clustered table.

10. 10
Query Optimization
The process of determining the best way to access the data
requested by query.
Execution path - optimal way to retrieve data by Oracle.
Approaches to retrive data:
• Use index to find ROWIDs and retrieve relevant rows.
– ORDER BY can be automatically implementes in presorted index.
• Full table scan
– Bypass the index for small table to avoid additional I/O step.
Complexity in selecting the right execution path can grow rapidly.
Two different Oracle query optimizers:
1. Rule-based optimizer
2. Cost-based optimizer

11. 11
Rule-Based Optimization
Uses a set of predefined rules for query optimization decisions.
No longer supported from version 10g and onwards.
Oracle rule-based optimizer had about 20 rules.
• The rule-based optimizer assigns an optimization score to each
potential execution path and then takes the path with the best
optimization score.
• When two paths have the same optimization score, the rule-
based optimizer selects the execution path based on the order in
which the tables occurred in the SQL statement.
Simple set of rules cannot deal with complex queries effectively.

12. 12
Figure 4-4. The effect of optimization choices

13. 13
Cost-Based Optimization
Introducted with Oracle version 7.
Selects the execution path that requires the least number of
logical I/O operations.
Uses statistics about the composition of the relevant data
structures.
Use of statistics enables cost-based optimizer to make a much
more well-informed choice of the optimal execution plan.
Automatic Workload Repository (AWR) automatically gathers
the statistics by default.
• Database access and usage statistics, system and session
statistics, SQL statements that produce the greatest loads...

14. 14
CBO - How statistics are used
Optimal execution plan is determined by assigning an
optimization score for each of the potential execution plan
using internal rules and logic along with statistics.
These statistics relate to the tables, columns, and indexes
involved in the execution plan.
Overall system statistics collected in Data Dictionary Tables,
also determine the overall cost of the I/O required by an
execution plan.
• I/O and CPU performance and utilization
Oracle database 10g and onwards, default cost basis is
calculated on the CPU cost plus the I/O cost for a plan.

15. 15
Table 4-1. Database statistics