IBM® DB2® Version 10.5 for Linux, UNIX, and Windows offers
accelerated analytic processing by introducing a new processing paradigm
and data format within the DB2 database
product. Advantages include significant reductions in time-to-value
and increased consumability, which can be achieved through minimal
DBA design requirements and reduced query tuning and debugging efforts.
Industry-leading compression, large performance gains for analytic
queries, and large reductions in performance variation round out the
benefits of deploying this technology.
DB2 Version 10.5 fix
packs deliver additional enhancements and product changes. For
a summary of all the new and changed functionality provided in each Version 10.5 fix
pack, see DB2 Version 10.5 for Linux, UNIX, and Windows fix pack summary.
DB2 column-organized tables
DB2 Version 10.5 introduces
compressed column-organized tables
for DB2 databases. The release
also includes broad support for data mart (analytic) workloads with
complex queries that are commonly characterized by multi-table joins,
grouping and aggregation, and table scans over a star schema.
Column-organized
tables are tables whose data pages contain column data instead of
row data. This new capability is in-memory optimized, CPU optimized,
and I/O optimized.
The DB2 product
continues to provide industry-leading performance across multiple
workloads by combining unique ideas with several of the best technological
developments of the past 10 years of database research. BLU Acceleration,
a combination of innovations from IBM Research
and the development labs that simplifies and speeds up reporting and
analytics, is a new, fully integrated capability in DB2 Version 10.5. Easy to set up and self-optimizing, BLU Acceleration
can typically eliminate the need for indexes, aggregates, or time-consuming
database tuning to achieve top performance and storage efficiency.
In most cases, no SQL or schema changes are required to take advantage
of this breakthrough technology.
DB2
Version 10.5 provides
the following enhancements:
- The ANALYTICS option of the DB2_WORKLOAD registry
variable, which provides a single setting to enable column organization,
space reclamation, and automatic workload management, and to automatically
configure memory, page size, and extent size (see System environment variables)
- New syntax for the CREATE TABLE statement to specify table storage
organization (see CREATE
TABLE statement)
- A new database configuration parameter (dft_table_org)
to change the default table organization (see dft_table_org - Default table organization)
- The new db2convert utility to convert row-organized tables
into column-organized tables (see db2convert
- Convert row-organized tables
into column-organized tables)
- The ability to use the REORG TABLE command
to reclaim storage in column-organized tables (see Space reclamation for column-organized tables)
- Automated workload management, which can significantly improve
the performance of workloads with several queries running at the same
time (see Default
query concurrency management)
- Dynamic list prefetching, a new prefetching type that is used
in query execution plans that access column-organized tables (see Prefetching data from column-organized tables)
- Support for NOT ENFORCED primary key and unique constraints, which
you can use to reduce performance costs and space requirements when
it is known that the data already conforms to the constraint (see Informational constraints), and for ENFORCED
primary key and unique constraints on column-organized tables
The following additional enhancements pertain to
column-organized tables
specifically:
- Semi-join support for queries to reduce the amount
of memory that is consumed by large inner tables; in some cases, queries
that use semi-joins might run faster. A semi-join is an optimization
technique that uses one or more predicates on the outer table and
other inner tables in the query. The combined filtering effect of
all eligible predicates is pushed down to one or more large inner
tables to reduce memory consumption. DB2 explain
output is enhanced to indicate whether the inner table of a particular
hash join operator is a large inner table for which the semi-join
optimization technique is applied (see EXPLAIN_ARGUMENT
table).
- Late decompression, the ability to operate directly on compressed
data for certain operations, thereby reducing memory usage.
- Multiplied CPU power that uses single instruction, multiple data
(SIMD) processing for many operations.
- A vector processing engine for processing vectors of column data
instead of individual values.
- Improved system scaling across cores.
- An enhanced caching strategy for buffer pools to substantially
reduce I/O.
- A smaller footprint for database storage than in DB2 Version 10.1.
- In-memory optimization for maximal performance.
- A system that is unconstrained by main memory size, in which the
amount of space used by the data that is being processed can be considerably
larger than the RAM.
For
the most up-to-date installation requirements for BLU Acceleration
, see DB2 Advanced
Enterprise Server Edition 10.5 - Detailed system requirements for
the BLU deployable component .
IBM DB2 pureScale Feature enhancements
The DB2 pureScale® Feature was
first introduced in Version 9.8. V10.5 builds
on DB2 pureScale Feature support.
The
DB2 pureScale Feature provides
extreme capacity and application transparency which exceed even the
strictest industry standard. Continued improvements in continuous
availability and high availability, the
DB2 pureScale Feature tolerates
both planned maintenance and component failure with ease. The DB2
Version 10.5 release
provides:
- Support for high availability disaster recovery (HADR)
- Increased availability
- Improved workload balancing
- Restore mobility between DB2 pureScale Feature
and DB2 Enterprise
Server Edition
In
addition, V10.5 introduces DB2 pureScale Feature enhancements
in high availability, performance, workload management, and installation.
For more information, see DB2 pureScale Feature enhancements.
Performance improvements
Continuing
to build on prior release performance improvements, V10.5 performance
improvements focus on explain information for column-organized tables
and extension to the CREATE INDEX statement to create an index containing
an expression-based key.
For more
information, see Performance enhancements.
SQL compatibility
If
you work with relational database products other than DB2 products,V10.5 builds
on existing functionality, interfaces, and compatibility features
to provide additional enhancements that make DB2 products more familiar to you. These enhancements
reduce the time and complexity of enabling applications that are written
for other relational database products to run quickly in a DB2 environment.
For more information, see SQL compatibility enhancements.
Product packaging simplified
Changes
to product packaging results in fewer packages, and greater value
through more function and features that are included in the base DB2 editions such as warehouse functionality.
For
more information, see Product packaging enhancements.