Proposed scheduling algorithm for deployment of fail-safe cloud services
Simon Kollberg · KTH Publication Database DiVA (KTH Royal Institute of Technology) · 2014
The cloud is a fairly new concept in computer science, altough almost everyone has been or is in contact with it on a regular basis. As more and more systems and applications are migrating from the desktop and into the cloud, keeping a high availability in cloud services is becoming increasingly important. Seeing that the cloud users are dependant on the cloud services being online and accessable via the Internet, creating fault tolerant cloud systems is key to keeping the cloud stable and trustworthy. This thesis researches different aspects of the cloud infrastructure such as virtualization which detaches the services from the physical hardware, scheduling algorithms that decides the mapping of virtual machines onto physical machines and fault tolerance in the context of availability in the cloud. It then proposes a new scheduling algorithm with the purpose of deploying virtual machines in a way that active-passive replication can be sustained.