|Thursday, May 11|
|3:45 pm - 5:15 pm|
Mastering the Hexapus - Data Access and Integration with OGSA-DAI (1/2)
Amy Krause, Mike Jackson
OGSA-DAI is a widely-used middleware product used for data access and integration within Grids. This tutorial will give a general introduction to OGSA-DAI presenting:
o Data access and integration requirements for Grids.
o How OGSA-DAI aims to satisfy some of these requirements.
o An overview of the OGSA-DAI architecture.
o OGSA-DAI extensibility points and how to exploit these.
o Features of the current release.
o Results of recent performance analyses.
o A demonstration of data integration with OGSA-DQP.
o How OGSA-DAI is being used in other projects.
By the end of the tutorial the attendees should have a strong understanding of key OGSA-DAI concepts and the suitability of OGSA-DAI for addressing their own data access and integration requirements within Grid environments.
The UK-based OGSA-DAI (Open Grid Service Architecture - Data Access and Integration) project is producing middleware to access and integrate existing data resources in Grid environments using web services. OGSA-DAI is already being used by a number of large projects within the UK, the US and Asia to satisfy their data access and integration requirements. In addition to this, the OGSA-DAI
project is working in close collaboration with other major Grid middleware providers, such as Globus and the OMII UK, to ensure that OGSA-DAI integrates seamlessly with their products.
OGSA-DAI currently supports access to data held in various types of data resources such as relational databases (MySQL, PostgreSQL, IBM DB2, Oracle, Microsoft SQL Server), data in XML repositories (eXist), and data in flat files (SwissProt, OMIM). These are just the officially supported databases - OGSA-DAI has also been shown to work with other databases.
The current releases are available in two flavours:
-OGSA-DAI WSRF 2.2 which integrates with Globus 4.0.1, and
-OGSA-DAI WSI 2.2 which is compatible with Axis 1.2.1, Axis 1.2RC3, OMII 2.0 or OMII 2.1.
Features of the new release will be discussed. These include:
-Various performance analyses and improvements, in particular when delivering results from relational databases.
-Activities for handling BLOBs.
-A set of activities that can be composed easily to perform batch-like transformations on data.
-Prototype support for aggregating relational data resources and running SQL queries over these aggregated data resources as if they were a single data resource.
An introduction will be given to the key concepts of OGSA-DAI along with an overview of how data resources are exposed by the layered architecture of OGSA-DAI and the role of each of the layers. The functional capabilities of OGSA-DAI services, enabled by the underlying layers, will be described.
The OGSA-DAI architecture has been designed to be highly extensible as it is unlikely that all functional requirements for a given project could be met by the base distribution. These extensibility points and how they may be exploited will be discussed.
The OGSA-DAI project has released eight major distributions. After the latest release the team has again entered into a dialogue with users and interested groups to see how they are currently using OGSA-DAI and to determine their future requirements. This will allow OGSA-DAI to develop in a way that attempts to take into account actual usage practice and informs the OGSA-DAI team as to how the software is actually being used in the field. Some of these real use cases will be described.
In the last six months the OGSA-DAI team have been profiling OGSA-DAI. They have focused on the internal and external representation of SQL query results and their delivery. Some of the findings of this profiling will be presented.
Data integration capabilities are also available through OGSA-DQP (Distributed Query Processing) middleware layered on top of OGSA-DAI. OGSA DQP uses OGSA-DAI for consistent access to database metadata and to interact with databases on the Grid in order to efficiently evaluate distributed queries and process complex data-intensive requests. A new release of DQP, building on OGSA-DAI WSRF/WSI 2.1, was published in December 2005. An overview of the features of OGSA DQP will be given along with a demonstration of
running distributed queries over a number of distributed data resources.
By the end of this tutorial attendees should have:
o An understanding of what some of the data access and integration requirements for applications on Grids are and the possible solutions to satisfy these.
o Background knowledge of what OGSA-DAI is, its scope and its relationship to other Grid middleware products.
o Knowledge of the key components of OGSA-DAI and how these are used to expose data resources onto a Grid.
o An awareness of the key extensibility points of OGSA-DAI.
o A knowledge of how OGSA-DAI is used in other projects.
For more details, please visit the OGSA-DAI website at
Welcome + Introduction (10m)
-Who needs data access and integration within a Grid?
-What is OGSA-DAI?
-How could OGSA-DAI help?
-The OGSA-DAI layer cake - from data resources to data services.
-Doing the work - OGSA-DAI activities and the engine.
-Making OGSA-DAI your own - extending OGSA-DAI.
-OGSA-DAI today - release 8
-OGSA-DQP - Building on OGSA-DAI (30m)
-How OGSA-DAI Performs (30m)
-Who else uses OGSA-DAI? (15m)
-What do/would _you_ want?
| Slides: Architecture and Extensibility|
| Slides: OGSA-DAI Overview|
| Slides: OGSA-DAI Today - Version 2.2|