PReServ 0.3.1
P-assertion Recording for Services

Provenance Query interface to the provenance store: functional specification of this interface can be found here.

XQuery interface to the provenance store: documentation on using this interface can be found here.

Motivation

The importance of understanding the process by which a result was generated in an experiment is fundamental to science. Without such information, other scientists cannot replicate, validate, or duplicate an experiment. For generations, scientists have used lab notebooks to record the process by which an experimental result was produced. These scientists were capturing provenance, which for some data is the documentation of process that led to that data. However, with large scale insilico experiments, it may difficult for a scientist to capture provenance especially when the experiment is conducted using thousands of computers owned and operated by different organisations.

In order to facilitate the recording and querying of provenance, PASOA has developed PReServ an implementation of the Provenance Recording Protocol (PReP).

Software Components

The following is a description of the software components that make up PReServ. The interaction between these components is shown in figure 1.

Figure 1: PReServ Component Interactions

The Provenance Store Web Service is a repository for provenance documentation. It handles requests both to store and query such documentation and provides well defined interfaces for doing so. Figure 2 shows the layered design of the Provenance Store.

Figure 2: PReServ Layers

Normally, a SOAP message is sent to PReServ to either record or query provenance documentation. (SOAP is the Web Services messaging layer.) Based on the port that the message was sent to, the SOAP Message Translator strips off the HTTP and SOAP Headers and passes the contents of the SOAP body to an appropriate PlugIn. Each PlugIn is responsible for knowing how to handle the content of the message it receives. The content, however, must conform to the schemas distributed with PReServ. For example, the Store PlugIn handles messages in order to record data in the Provenance Store. The selected PlugIn then makes calls on the backend data store to generate an appropriate response for the received message. Currently, PReServ comes with in-memory, and database backends. Each of these backends implements the same API, the Provenance Store Interface. This abstraction makes it easy to integrate new backend stores without having to change already developed PlugIns.

The Provenance Store is accessed directly using Web Service calls. (See figure 1) However, developers may not want to deal with the intricacies of building their own SOAP messages to send to the Provenance Store. Therefore, PReServ provides a set of Java libraries to both record and query documentation from the Store. Figure 1 shows how the Java libraries (PS Client Side Library) can be used to interact with the Provenance Store. A recording API (Application Programming Interface) presents methods that are directly mapped to the messages that specified by PReP. A query API presents a set of basic methods for browsing the Provenance Store's hierarchy of data. Both APIs take method calls and translate them into Web Service invocations to the Store.

The third software component is an Axis Handler. Axis is one of the most common set of libraries for developing Java based Web Services. Therefore, PReServ provides a mechanism for "automatically" recording interactions for services developed using Axis. A developer can add this functionality by adding a provided Jar file to their classpath. The developer's Web Service is then wrapped by PReServ's Axis Handler, which then intercepts all incoming and outgoing data and records it to a configured Store. This "automatic" recording is shown in Figure 1, where two Axis Handlers wrap the WS Client and Web Service in order to record their interaction in the Provenance Store.

Get PReServ

Download PReServ 0.3.1

Warning: This release may not be complete and is not completely documented. It is intended only as a technology preview. We assume you have had experience developing web services and using Apache Tomcat.

Dependencies:

Java Version 1.5
Apache Tomcat 5 or greater
Apache Ant 1.62

PReSev and Java 1.4:

PReSev and Java 1.4

If you need to run PReServ under Java 1.4 please contact us. We are moving towards adopting Java 1.5 completely on the Provenance Service side while maintaining Java 1.4 compatibility on the client side.

Change Log:

Changes between 0.3 and 0.3.1

Bug fixes and performance improvements

Changes between 0.2.6 and 0.3

Build file now contains targets to deploy/undeploy preserv from the command line
Support for clients to server gzip compression
The backend automatically compresses data
We now use an in house xquery engine, LAXQuery, which provides greater scalability than the engine we used previously, IPSIXQ.
Support for changing between LAXQuery and IPSIXQ in the web.xml configuration file
Changes to the code to allow for easier development of multiple backends
Code reorganisation to allow for a better set of distributions
Fixed several threading bugs
Compatible with the 025 Provenance Schemas

Changes between 0.2.5 and 0.2.6

This release is compatible with a version 024 Provenance Schemas. These schemas increase the extensibility of the schemas.
This release of PReServ is compatible with the 0.4 Client Side Library produced by the EU Provenance project
Minor bug fixes in querying and recording

Changes between 0.2.4 and 0.2.5

Maintenance release
Fixed minor bugs with dataAccessors

Changes between 0.2.3 and 0.2.4

Supports Provenance Query Inteface
WSDL now designed for extensibility
Fixed bugs in DOMWriter handling

Changes between 0.2.2 and 0.2.3

Improved Developer Friendly WSDL
WSDL is now JAX-RPC compatible
Provenance Service handles JAX-RPC requests

Changes between 0.2.1 and 0.2.2

Improved client library documentation
Added ClientLibExample
Improved building of provenance store (no need to modify any configuration files)

Changes between 0.2 and 0.2.1

Changes in WSDL interfaces to be clearer and consise

Changes between 0.1.5 and 0.2

Compatible with new structure of process documentation
XQuery interface
WSDL interfaces
New database backend (easier to redeploy)
Refactored code for future development, easier to add backends
AxisHandler now supports the transmission of multiple tracers (replacing sessions)

Changes between 0.1.4 and 0.1.5