IBM Db2 Analytics Accelerator for z/OS, Version 8.1.0

Basic concepts

Get familiar with the basic concepts of IBM Db2 Analytics Accelerator for z/OS.

Accelerator
IBM Db2 Analytics Accelerator for z/OS consists of a high-performance hardware platform and an optimized database query engine. These components work together to support a variety of data analysis and business reporting tasks.
Static SQL query support
Static SQL query support means that you can accelerate SQL queries that are embedded within Db2 for z/OS application programs.
Accelerator-shadow table
An accelerator-shadow table represents a table in your database. When you define accelerator-shadow tables in your administration client, you instruct IBM Db2 Analytics Accelerator for z/OS to copy the definition of that table (an empty table) to an accelerator.
How to select tables for query acceleration
Selecting the proper tables for a query to be accelerated is basically simple: You need to define the tables that are referenced by the query. Slightly more thought needs to be spent on the choice of proper distribution keys and organizing keys.
Distribution keys
A distribution key determines how tables rows are distributed among the available working nodes. IBM Db2 Analytics Accelerator for z/OS selects a suitable distribution key automatically when tables are loaded. The distribution varies between the two extremes: an even distribution (called random distribution) without the use of a key and a maximally skewed distribution that uses a distribution key.
Organizing keys
Organizing key also have an impact on the query response time. When choosing an organizing key, you select columns by means of which you group the rows of an accelerator-shadow table within the data slices on the worker nodes. This creates grouped segments or blocks of rows with equal or nearby values in the columns selected as organizing keys. If an incoming SQL query references one of the organizing key columns in a range or equality predicate, the query can run much faster because entire blocks can be skipped. It is not necessary to scan the entire accelerator-shadow table. Thus the time needed for disk output operations related to the query is drastically reduced.
Tables states
A table can go through various states while it is being processed by IBM Db2 Analytics Accelerator for z/OS. The meaning of these states is explained here. In addition, it is shown which operation you can run on a table in a particular state. Finally, a diagram shows the possible shifts from one state to another (state transitions).
IBM Db2 Analytics Accelerator for z/OS system table entries
When you select tables for query acceleration, information about these tables is added to the IBM Db2 Analytics Accelerator for z/OS system tables in the Db2 for z/OS catalog of the connected subsystem. Each entry (system table row) serves as a link between Db2 and the table on the accelerator. An entry contains, among other information, the table name.