CERN considers cloud-based number crunching

The European Organisation for Nuclear Research (CERN) is partnering with Rackspace to do some epic number crunching on the cloud.

CERN and Rackspace are building a hybrid cloud built atop OpenStack, an open-source Infrastructure-as-a-Service (IaaS) platform.

The platform was originally developed by Rackspace as part of a joint effort with NASA. OpenStack is fast becoming the tool that the likes of HP and IBM use to expand their cloud portfolios.

Under the deal Rackspace will fund one full-time CERN member which will also help to push its street cred considerably.

Tim Bell, leader of CERN’s OIS Group IT department said that CERN and Rackspace will initially focus on simulations.

This involves putting into place the theory and then working out what a particle collision will have to look like.

Bell said that there will be investigations into using the cloud for data analysis in the future but there is no timeframe for it at the moment.

So far, the experiences running between the two data centres in Geneva and Budapest gave CERN early indications of the challenges of the more data intensive work, he said.

CERN’s physicists write their own research and analytics software, using a combination of C++ and Python running atop Linux.

Complex physics frameworks and the fundamental nature of the research makes it difficult to use off-the-shelf [software] packages, he said.

The outcomes of the collaboration will help everyone involved better understand the workloads that can be placed on the public cloud.

CERN’s private cloud will use 15,000 hypervisors and 150,000 virtual machines by 2015. It is fairly likely that any public cloud will likely need to handle similarly massive loads with a minimum of latency.

By running small tests with a variety of public-cloud providers, CERN can determine how to best distribute workloads, puzzle out those latency questions, and eventually take on some of its more serious number crunching.