URL for further information:
Provide a set of generic keywords that define your contribution (e.g. Data Management, Workflows, High Energy Physics)
grid, API, SAGA, OGF
4. Conclusions / Future plans
The SAGA based Service Discovery C++ API supporting SQL style filters has been implemented within gLite. We provide a C wrapper that is compatible with the old gLite API however this is mainly useful for testing to check that the returned set of services is the same. In future we will provide Python and C wrappers and a Java implementation. The wrappers avoid the need for rewriting the plugins. JNI will be considered for a Java wrapper as a short term solution.
The existing gLite Service Discovery API is being used by components of the Workload Management System (WMS) and Data Management Systems. The Logging and Bookkeeping service developers have expressed an interest in using the new SAGA API because of the extra functionality it offers. The SAGA approach is useful from a user perspective because it frees end users from dependency upon specific Grid middleware. We expect that the use of SAGA will grow making the interface increasingly valuable. The SAGA Service Discovery API has been designed to be very easy to use with SQL style filter expressions. It also means that services that use other services require less error-prone configuration as they can find the actual services when they need them. This in turn leads to increased reliability.
1. Short overview
The Simple API for Grid Applications (SAGA) from the Open Grid Forum (OGF) defines standard APIs to allow grid applications to be middleware independent. We have contributed a Service Discovery specification to SAGA and an implementation as a generalisation of the existing gLite component. The purpose of service discovery is to locate services based on various characteristics. The API supports a very general selection mechanism and is built on the GLUE notion of a service.