System Software Lab.

About

The main focus of System Software Laboratory is to design, develop, and evaluate system software technologies that can support the diversity of emerging applications, including high performance scientific applications and data-intensive applications, and new computer architectures. Our laboratory is particularly interested in innovative technologies for the efficient and reliable management of large scale systems based on virtualization.


Lab.

Engineering BLDG 3. 106 Rm. 605 


Research Interests


System software technologies for next generation computing systems to support emerging applications and new computer architectures.


  • Cloud computing
  • Big data analytics platforms
  • Virtualization
  • Resource management for large scale systems
  • Scientific computing
  • Next generation supercomputing systems


Cloud Resource Management



In cloud systems based on virtualization, virtual machines (VMs) share physical resources. Although resource sharing can improve the overall utilization of limited resources, contentions on the resources often lead to significant performance degradation. However, cloud systems with virtualization open a new opportunity to widen the scope of contention-aware scheduling, as VMs can cross legacy system boundaries with live migration. We are working on resource scheduling techniques to optimize the system overall performance as well as application (or VM) performance, using VM live migration. 



Science Clouds



Virtualization has become popular to improve system utilization by consolidating multiple servers into a physical system. In addition to the improved utilization, other benefits of virtualization, such as flexible resource management, fault isolation, and support for different operating systems, have led to the increase of interest in the virtualization of computing clusters for scientific computing. However, the characteristics of scientific applications are quite different from those of I/O-intensive server applications, which have been the main target of prior performance  optimization for virtualization. To adopt virtualization for scientific applications, we are working on thorough analysis of their performance characteristics in virtualized systems, and optimization techniques for clouds for scientific applications - science clouds 



Current Students
Alumni
  • Jaewon Kwak, Master's degree 2017, currently at NEXON Korea
  • Taekyung Yoo,  Master's degree 2015, currently at NEXON Korea