The main research area is now (Jan 2011) on the use of Openflow to implement cloud network services. 

A few software delivered by my research:

Kestrel (Used by STAR for a large run on our cloud)

scp tsunami (Used at CERN to distributed VM images on 500 hosts)

scp wave (First implementation of tsunami)

Students develop various tools for distributed systems as applied to "clouds". Such as, a distributed file system within VMs, a reverse shell framework, a Fibonacci heap file transfer mechanisms used at CERN and a publish subscribe system based on AMQP (Used in sensor networks).

XMPP frameworks (Lance, graduated 2011)

What started as a class project is now an interesting research thrust. While comparing push vs. pull job submission framework on the grid we concluded that new pilot job frameworks (pull model) are very similar to botnet architectures. In this thrust we try to look at techniques and ideas used in network security to build a distributed system to schedule jobs on overlays as well as monitor compute agents. Currently we used XMPP (aka Jabber) to build a Condor like system, all classads are sent using XMPP and the central manager is an XMPP server.

Virtualization (Mike , Michael, Linton, graduated 2009):

Virtual machines and virtualization in general is being actively investigated , especially in the cloud computing framework. It is seen as a way to reduce complexity of managing, operating and provisioning large distributed systems. It offers strong encapsulation of application and users as well as benefits such as suspend-resume and migration. Our research centers around the provisioning of virtual clusters -overlays of virtual machines dsitributed across wide-are networks-, their monitoring and autonomic migration. We currently operate a testbed of 14 Xen enabled servers and a 10 TB lustre file sytem for potential migration over WAN. We currently decommissioning this cluster and making use exclusively of cloud resources including the Clemson Palmetto cluster (~12000 cores) which now supports KVM. 

Service Coordination and Provisioning (Vikas, graduated 2010):

The use of services to build infrastructures is very promising. But most services offered by current service providers such as TeraGrid fall quite short of being value-add services to build future research environments. In parallel the corporate world is seeing an explosion of APIs offered to the public and easily composed to form new value added applications (e.g mashups). In this research thrust we look at principles and implementations of composing interoperable services as well as dynamic provisioning of missing services. In cloud terminology we work on a platform as a service tool that uses IaaS providers to build specific platforms/infrastructures on demand.