Benchmark instructions

1) Install XBench
-----------------
Visit http://db.uwaterloo.ca/~ddbms/projects/xbench/. Note that there are
separate downloads for Windows and Linux.

Follow the instructions. Note that the TPC-W population generator needs 
building under Linux. Set an appopriate value for MIN_HEAP in "toxgene".

2) Build the DB XML Utility Programs
------------------------------------
Build "load_container" and "query_runner"

=========================================================================
EITHER  (work done by Gareth July 2004)

3) Preparing the dbxml_xbench script
------------------------------------
  - ensure java is installed and the $JAVA_HOME/bin is on the path
  - delete the readme file form the xbench/output directory
  - install XML::XPath

  If under unix
    - copy dbxml/src/test/benchmark/toxgene.sh to xbench/toxgene
    - copy dbxml/src/test/benchmark/ to xbench/toxgene

2004-10-06 ARW xdbgen.pl is the Windows version of this script (the Unix
variant works out of the box - use the downloaded version)

This script does everything - from using XBench to generate data sets,
loading the containers and running the queries.
=========================================================================
OR (recent work done by Andy October 2004)

3) Generate the XBench Data
---------------------------
Follow the instructions in the XBench documentation. I generated the "small"
(about 10MB) and "normal" (about 100MB) data sets, and then derived the "medium" (about 50MB), "tiny" (about 1MB) and "miniscule" (about 20K) data sets.

TODO tell everybody where this data is (way too much for CVS) 

4) Loading the Containers
-------------------------

Use the Tcl script "xbench.tcl" - type "load_containers ?"

Takes a long time - best to run for various container types and then keep the
results!

5) Query Workload
-----------------

Use the Tcl script "xbench.tcl" - type "query_container ?"

=========================================================================
