See the new TdBench 7.2 version for Linux at http://downloads.teradata.com/download/tools/tdben
Teradata Benchmark Query Driver (TdBench) provides a set of tools to help you compare the performance within a data warehouse:
The tools provide a framework for executing benchmarks driven by a Windows Server or PC and reporting on the results using DBQL. There are also tools for extracting a cohesive set of Queries and Tables from DBQL to define the benchmark.
This reference guide explains how to install, configure and use the TdBench package.
TdBench is provided free of charge in the belief that good benchmarks are an accurate reflection of your real or planned production load. If Teradata customers and prospects can do a better job of executing such tests with the use of this software, they will be able to make better decisions on selection of tools, planning for rollout, and managing system growth.
This package was developed by the Teradata Americas Benchmark Center of Expertise in support of competitive benchmarks. However, we found that customers recommended its use in support of software and database changes. As a result, this package was presented at the Teradata Partners 2010 Conference in the session “Build Your Own Benchmark Center”.
.lstfile located in the
The QueryDriver manages the execution of a set of tests which are defined by
.sqlfiles), generally executions of macros or calls to stored procedures
.batfiles as defined in a
Each execution of a test results in a “Run” which is identified by a sequentially numbered
RunID in directories on the Windows Server or PC and in a
TestTracking table on Teradata. The TestTracking table provides the precise start and stop timestamps for each test along with the identification of the test and your description of the conditions for that test.
A unique feature of this query driver is the support of one or more queue tables for SQL commands to be executed by worker sessions. This simplifies the definition and control of the tests by allowing different query workloads to be defined (e.g. heavy analytics, medium reporting, and light tactical), and arbitrarily changing the number of worker sessions without needing to change the scripts. It also supports different testing models such as fixed period, fixed work, and decay.
Reporting views are defined which uses the
TestTracking table and DBQL or
Resusage tables for creating reports. The design allows for the benchmark database to extract the historical data from the
DBC database and retain multiple benchmark databases for future analysis.