|
TPoX Home
Benchmark
Specification
TPoX
XML Data
Workload Driver
·
Documentation
·
Download
Benchmark
Results
TPoX
Project Site
|
Transaction Processing over
XML (TPoX) is an application-level XML database benchmark based on a
financial application scenario. It is used to evaluate the performance of XML
database systems, focusing on XQuery, SQL/XML, XML storage, XML indexing, XML
Schema support, XML updates, logging, concurrency and other database aspects.
(Overview paper)
TPoX consists of the
following parts:
- A
toolset for XML data generation to efficiently generate millions of XML
documents with well-defined value distributions and referential
consistency across documents.
- XML
Schemas for all document types used in the benchmark.
- A set
of transactions to be run on the generated data. This includes queries
in XQuery and SQL/XML notation as well as insert, update and delete
operations.
- A
workload driver which spawns 1 to n parallel threads that simulate
concurrent database users. Each user connects to the database and
submits a mix of transactions. The transactions are picked randomly from
a set of transaction templates. At run time, parameter markers in the
templates are replaced by actual values drawn from configurable random
value distributions. The workload driver collects and reports
performance metrics, such as min/max/avg response time and overall
throughput.
- Documentation
for all of the above-mentioned pieces.
TPoX simulates a security
trading scenario and uses a real-world XML Schema (FIXML) to
model some of its data. TPoX is purposefully simplified, yet still realistic
in terms of documents, transactions, and usage of XML Schemas.
TPoX is very flexible and
extensible. The data distributions, transactions, workload composition, data
and multi-user scaling, commit frequencies, think times, etc. are all
configurable. With specific settings we propose a reference workload in the
first version of TPoX. But, any parameter can be changed and different
transactions can be run over the same or modified data. This makes TPoX a
versatile performance test harness for XML databases.
We welcome any feedback or
contributions to extend and refine the benchmark. You can use forums, email, or
connect to the TPoX SVN.
Possible contributions include but are not limited to:
- Additional queries
- New workloads, i.e. new query/transaction
mixes on the existing data, e.g. an analytical workload
- Enhancements to the data generation
- Query syntax adjusted to your favorite
database system or XQuery engine
- Enhancements to the workload driver code, e.g.
a class to connect the driver to your favorite database
- TPoX benchmark results
Matthias Nicola, mnicola@us.ibm.com
Irina Kogan, ikogan@ca.ibm.com
Rekha Raghu, rekha.raghu@intel.com
Agustin Gonzalez, agustin.gonzalez@intel.com
Mike Liu, mikezliu@ca.ibm.com
Berni Schiefer, schiefer@ca.ibm.com
Guangjun (Kevin) Xie, kxie@ca.ibm.com
|