Flavours of Ganga

Fri 2nd November 2012

Three particle physics experiments have started using the Ganga front end. The application, partly developed by GridPP, is proving very useful to the physics community.

The kind of research being done at the Large Hadron Collider at CERN would be near impossible without the grid. This has intrigued many other particle physics experiments and they have been looking to get up and running on the grid. However moving from one or two people running test jobs to something that the entire community can use is not easy. “We’ve found ourselves in a bit of a difficult situation in that we’re a small experiment, but we need more computing resources than would be available at one university’s cluster.” says Matthew Mottram who works on SNO+ at the University of Sussex “Alongside this we have members across the world and a number of computing systems with different methods for submitting, running and monitoring computing jobs. Accessing the grid using the scripts developed by the SNO+ community during testing just wasn’t scalable.”

SNO+ are not alone, researchers from both T2K and SuperB are also looking to fit the grid into their workflow. All three look at what is called flavour physics and all have come to the conclusion that Ganga probably fits their needs. Developed for use by ATLAS and LHCb, Ganga is a simple application that is used to create, submit and monitor jobs, most importantly it can be used with just a single machine, a local cluster of computers or the grid. “Ganga allows for jobs to be submitted to a range of computing architectures requiring minimal configuration.” says Matthew “Their system also makes writing plugins to extend its use to other experiments very easy.”

This extendibility is one of Ganga’s major selling points. It was even chosen by a small company called Imense when running their image tagging work on the grid. “We always wanted Ganga to be customisable to meet the requirements of any new user communities, that’s why we implemented the plugin idea” explains Ulrik Egede head of the Ganga project “This means that once someone has investigated and built a solution for their community with Ganga it can be used by an end-user to process tasks without needing detailed knowledge of the grid.”

It was actually SuperB that was first of the three to use Ganga. Armando Fella from the INFN facility in Pisa works on SuperB and has been instrumental in writing their plugin, “SuperB has not been built yet but we already realise the importance of having a reliable, accessible IT infrastructure in place before data taking and for simulated data analysis” he explains “last year we started looking into Ganga as our grid interface for distributed data analysis and simulation purposes. We had a small enough team but last May our code was included in Ganga so our users now have a simple to use interface.”

The SNO+ Detector

T2K is in a similar situation to SNO+ but has been using the grid for a bit longer and now wants to make it accessible to everyone in the collaboration “To help me understand Ganga I have been using scripts that can interact with the application and do the required tasks” says Luke Southwell from Lancaster University “now I just need to mould these into a basic plugin so anyone in T2K can run jobs. The SuperB plugin has been very useful and I’ve been working with the SNO+ people as well.”

Not only has Ganga proved very flexible for physics but smaller teams have used it. “I have worked with a few researchers interested in using the grid” says Mark Slater from the University of Birmingham “and I usually suggest that they use Ganga. It solves a lot of problems and makes life easier once the community reaches a critical mass.” The next experiment on Mark’s list is again particle physics with CERN-based experiment NA62 starting to investigate use of the grid.

If you are interested in using Ganga or the grid for your research you can find more technical information here or contact GridPP on info@gridpp.ac.uk.

© Copyright GridPP
If you wish to reproduce this piece please credit GridPP and contact us to say you are using it