Resource management and fault tolerance principles for supporting distributed real-time and embedded systems in the cloud

  • Authors:
  • Kyoungho An

  • Affiliations:
  • Vanderbilt University, Nashville, TN

  • Venue:
  • Proceedings of the 9th Middleware Doctoral Symposium of the 13th ACM/IFIP/USENIX International Middleware Conference
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Cloud computing provides an attractive solution to host enterprise applications due to its cost effectiveness, and its ability to seamlessly adjust to changing application work-loads while providing the desired performance assurances using elastic and dynamic resource management. These benefits, however, do not yet readily carry over to distributed, real-time and embedded (DRE) systems, which are a class of systems that require stringent assurances on quality of service (QoS) properties including timeliness, reliability and security all at once. This doctoral research is investigating the sources of these limitations that make it hard to host DRE systems in the cloud, and developing solutions to overcome them. This paper makes three contributions in this regard. First, it outlines the key challenges that must be resolved in supporting DRE systems in the cloud and surveys related literature. Second, it presents ongoing work that addresses one key challenge stemming from the need for real-time and scalable resource monitoring in the cloud. Third, it outlines our proposed ideas on resolving the remainder of the challenges.