Skip to content
Find file
Fetching contributors…
Cannot retrieve contributors at this time
91 lines (61 sloc) 4.56 KB
Note: this is a modified version of BenchmarkSQL (for which the standard README file can be read below). It supports non-interactive operation of the database benchmark, and record results in various CSV files.
Change Log:
version 2.3.3 11/19/2010 sjm - Added DB2 LUW V9.7 support, and supercedes patch 2983892
- no other changes from 2.3.2
Instructions for building
Use of JDK 1.5 is recommended, build with "ant jar" from the command line of
the base directory or use your favorite IDE such as Netbeans or Eclipse.
Instructions for running
The below scripts all use relative paths, but, they depend on JAVA_HOME
environment variable being set so that the correct runtime can be found.
JDBC drivers and sample "?.properties" files are included to make it extremely easy for
you to test out the performance of EnterpriseDB, PostgreSQL, MySQL, Oracle, DB2, & SQL Svr
in your environment.
Any special instructions by DBMS are prefaced by <dbms_name>, else just substitute the appropriate properties file in the command lines below
.sh or .bat script suffixes are NOT indicated below; specify as appropriate
1. Go to the 'run' directory, edit the appropriate "??????.properties" file to
point to the database instance you'd like to test.
2. Run the "sqlTableCreates" to create the base tables.
$ runSQL sqlTableCreates
3. Run the "loadData" command file to load all of the default data for a benchmark:
A.) Approximately half a million rows in total will be loaded across 9 tables
per Warehouse. (The default is numWarehouses=1) A decent test size of data
totaling about 1 GB is 10 warehouses as follows:
$ loadData numWarehouses 10
Note: "loadData" will truncate all tables prior to inserting data through the JDBC program
There is also a "sqlTableDrops" script if you need it. If needed, run
$ runSQL sqlTableDrops
B.) Alternatively, you may choose to generate the load data out to CSV files that can be
bulk loaded as follows:
$ loadData numWarehouses 10 fileLocation ./csv/ (directory BenchmarkSQL-2.3.2/run/csv used in this example)
These CSV files can be bulk loaded into EDB-Postgres via the following:
$ runSQL sqlTableCopies
<DB2 LUW> runSQL sqlTableCopies_DB2 //first check comments at top of sqlTableCopies_DB2
You may clean out the data in the tables without dropping them via:
$ runSQL sqlTableTruncates
<DB2 LUW> runSQL sqlTableTruncates_DB2 //first check comments on top of sqlTableTruncates_DB2
4. Run the "runSQL" command file to execute the SQL script "sqlIndexCreates" to
create the primary keys & other indexes on the tables.
$ runSQL sqlIndexCreates
<DB2 LUW> runSQL sqlIndexCreates_DB2 //first check comments at top of sqlIndexCreates_DB2
5. Run the "runBenchmark" command file to execute the swing GUI application to
test the database. Don't forget to set the number of warehouses equal to the
number you created in step 3. For each run, a report will be placed in run/reports.
A sample report is included.
$ runBenchmark
6. Operational Notes to minimize problems:
(a) executing runBenchmark will start the GUI.
Click the Database button to view properties file settings. No changes are needed if the properties settings are correct.
Click the Terminals button and specify desired settings.
Specify the same number of warehouses as you created.
Select either "Minutes" or "Transactions per terminal" and blank out the other setting.
Click the Weights button and specify desired settings
Click the Controls button, then click Create Terminals. One DB connection per Terminal is created. Click Start Transactions to start the benchmark.
(b) If changing the number of terminals between runs, it is best to close the GUI window and re-execute runBenchmark
(c) If the benchmark runs properly, all database connections are terminated at completion. You may need to manually
terminate connections if this is not the case
(d) When done, close the GUI window
Something went wrong with that request. Please try again.