Design and implement a fast and scalable backend to SAGA'a Advert API package

Google Summer of Code 2009 ideas

Primary Mentor: Andre Merzky andremerzky@gmail.com.
Secondary Mentor: Shantenu Jha sjha@cct.lsu.edu
OMII Project: OMII-SAGA.

Background

OMII-UK funds SAGA development to further code stability and feature completeness of the SAGA API standard implementation. To further this goal, we look for a fast and scalable implementation of an backend to SAGA's Advert API extension.

Project Goals

The project should come up with an backend service which maps the functionality required by the SAGA Advert API. That service implementation must be proven to be fast (>O.10 transactions/second) and scalable (~O.100 congruent remote clients). The DB should scale to millions of entries, in deep hierarchies, w/o significant performance loss. The service should be relatively simple to deploy and configure. PAM or X509 security would be great - otherwise user/pass is sufficient. Code documentation is a must-have.

Project Description

The Advert API simply allows R/W access to hierarchically ordered sets of key/value pairs - a multitude of existing technologies could thus be considered to implement a backend. Part of the students work is to make a informed decision on what technology to use. Otherwise, the implementation is probably straight forward, but stability and performance constraints (stated above) will make the work challenging nevertheless.

Project Requirements

The SAGA Advert API is relatively easy to understand. A moderate to good knowledge of C++ is required to be able to write the SAGA adaptor which has to talk to the service to be implemented.

The skills required for the service implementation entirely depend on the technology choosen for the backend. However, experience with distributed systems is a huge bonus. Database experience will help.