TPoX 2.0 released and results published !

2009-09-09T21:33:25Z

An updated version of the TPoX benchmark kit (TPoX 2.0) was released on July 31st, 2009. The enhancements include easier and faster data generation for TB-scale databases.

In August 2009 IBM has published TPoX 2.0 benchmark results. A full report is available on this page:
http://tpox.sourceforge.net/tpoxresults.htm

TPoX v1.3 released !

2008-12-09T10:51:09Z

TPoX 1.3 is being released to include new source files that were contributed
to the open source project by the University of Furtwangen to enable TPoX
for Oracle and MS SQL Server. See the release notes for further details.

1TB TPoX Benchmark published !

2008-11-08T19:39:11Z

Intel has been running the TPoX benchmark with a raw data volume of 1 TeraByte. The hardware system was powered by their latest 6-core CPUs (Xeon 7400 series). To the best of our knowledge, this is the industry's first public report of a 1TB XML database benchmark.

Details can be found on this page:

http://tpox.sourceforge.net/tpoxresults.htm

TPoX 1.2 is available now!

2008-06-13T16:37:42Z

New Features and Enhancements in TPoX 1.2

TPoX Version 1.2 contains the following enhancements over the previous
version (1.1):

(1) The mixed workload, described in workload property file
TPoX/WorkloadDriver/properties/newmixed.properties, has been added with
following features:

- XML Schema validation feature is added to statement I2, U2 and U4
- Explicit type casting of some values into xs:decimal in statement U5 and U6
to avoid potential error when validating updated documents agains XML
schemas
- Statement U5 now updates only the first account holding position for a
customer. Without specifying the first holding position to update may cause
an XQuery dynamic error since a customer may have multiple holdings

(2) All order xml documents generated by TPoX datagen tool now contain realistic
currency symbol for "Order" elements

(3) All workload property files in TPoX/WorkloadDriver/properties now have
settings based on an XML document set of scale XXS with 15% extra documents.
The XML document set of scale XXS with 15% extra documents is pre-generated
and downloadable from TPoX sourceforge website.

(4) The following new features have been added to TPoX workload driver:

- Allow max order and account ids to be 12 digits long. This is necessary for
TPoX large scale factors.

- The workload driver accepts a new command line paramenter: -sc <schema name>,
which allows the workload to run on a set of TPoX database objects
(tables/indexes/XML schema objects) in a specific database schema.

- Allow TPoX-style parameter markers ("|1", "|2", etc) in statements other
than XQuery. Other statements, such as SELECT, DELETE, UDPATE, should be
allowed to use TPoX-style parameters (ie. "|1", "|2", etc) if you need to.

- Allow p1|2 = p1|1 specification for TPoX-specific parameter markers in workload
properties file. p1|2=p1|1 means that two parameters use the same value.
Previously such specification was only allowed for ?-denoted parameter markers
and caused problems with TPoX-style parameter markers (e.g., |1, |2, ...).
This has now been fixed in TPoX 1.2.

- Allow the user to specify a comment about the workload in the workload driver.
We added the -c option to the workload driver that allows the user to specify
a comment about the workload. Then comment.txt containing this comment is
saved in the output directory of the run. This way it is easy to see what was
special about each particular run (e.g. changed db/dbm configs, etc.).

TPoX 1.1 just got released!

2007-06-13T06:19:26Z

New Features in TPoX 1.1

TPoX Version 1.1 contains the following enhancements over the previous version (1.0):

(1) TPoX 1.1 contains six sub-document level updates (U1 through U6) which supersede
the previous updates that perform full-document replacmeent. The old updates
are still included.

(2) The data generation now also produces XML fragments for "accounts". These are required
by Update2 which inserts new accounts into existing custacc documents.

(3) The workload driver has new options -pc, -cl and -m.
-pc <n> computes the n-th percentile of the response times for each transaction type.
-cl <l> computes the confidence interval with confidence level l for the response
times of each transaction type.
If -pc and/or -cl are used, then by default 50MB of main memory are used to collect
and compute these statistics. The -m option can be used to change the memory.

(4) The workload driver also has the following enhancements:

- Enhancements to support the new updates, in particular Update2 which inserts a new account into
an existing custacc document.
- Two new lists with possible parameter values for Update4 (order_SolFlag.txt, order_src_values.txt)
- Shows throughput in transactions per seconds (in addition to the previous transactions per minute)
- Shows number of result rows returned for each query if verbosity level 2 is used. This output
is in the userN.txt files.
- Allows XML inserts into CLOB columns (for comparison purposes).
- Allows uniformly distributed decimal values as input for parameter markers in query predicates.
- Various bug fixes (see https://sourceforge.net/tracker/?group_id=185925&atid=915227\).

(5) Small correction in the data generation for "orders":
- Some orders did not have a trade date (TrdDt attribute). This is now fixed by setting minOccurs to 1.
- Some orders did not have a cash value (Cash attribute). This is also fixed by setting minOccurs to 1.

(6) Updated documentation to reflect all of the above.

Feedback, questions, suggestions for improvement and active contributions are all welcome.

Thanks.

The first release of TPoX is now available!

2007-01-18T21:14:09Z

Transaction Processing over XML (TPoX) database benchmark is going public!

The download of the TPoX package and sample data (v1.0) can be found here: https://sourceforge.net/project/showfiles.php?group_id=185925

Feedback, questions, suggestions for improvement and active contributions are all welcome. Thanks.

Recent posts to news