The Adaptive Server version 15.0 query processor is self-tuning, requiring fewer interventions than earlier versions. This version of Adaptive Server has less reliance on worktables for materialization between steps since the engine supports data flow between steps. However, more worktables could be used in cases where Adaptive Server determines that hash and merge operations are effective.
New features include support for:
Both vertical and horizontal parallelism for query processing
Improved index selection, especially for joins with OR clauses and joins and search arguments (SARGs) with mismatched but compatible datatypes
More efficient algorithms
Improved costing, using join histograms for joins with data skews in joining columns
Improved query plan selection that enhances performance through:
New index union and index intersection strategies for queries with and/or predicates on different indexes
On-the-fly grouping and ordering using in-memory sorting and hashing for queries with group by and order by clauses
Cost-based pruning and timeout mechanisms that use permutation search strategies for large, multi-way joins, and for star and snowflake schema joins
Improved problem diagnosis and resolution using:
Searchable XML format trace outputs
Diagnostic output from new set commands
Joins involving a large number of tables
Data and index partitioning, which are especially beneficial for very large data sets
Partitioning is the basic building block for parallelism.
Adaptive Server release 15.0 provides roundrobin partitioning. Round robin partitionin is equivalent to the 12.5 style of partitioning. During the upgrade to Adaptive Server release 15.0, all existing partitioned tables are unpartitioned and automatically converted to 1-way round robin partitioned tables.
| Copyright © 2005. Sybase Inc. All rights reserved. |
|
|
