The primary motivation for uptake of virtualization has been resourceisolation, capacity management and resource customization allowing resourceproviders to consolidate their resources in virtual machines. Variousapproaches have been taken to integrate virtualization in to scientific Gridsespecially in the arena of High Performance Computing (HPC) to run grid jobs invirtual machines, thus enabling better provisioning of the underlying resourcesand customization of the execution environment on runtime. Despite the gains,virtualization layer also incur a performance penalty and its not very wellunderstood that how such an overhead will impact the performance of systemswhere jobs are scheduled with tight deadlines. Since this overhead varies thetypes of workload whether they are memory intensive, CPU intensive or networkI/O bound, and could lead to unpredictable deadline estimation for the runningjobs in the system. In our study, we have attempted to tackle this problem bydeveloping an intelligent scheduling technique for virtual machines whichmonitors the workload types and deadlines, and calculate the system over headin real time to maximize number of jobs finishing within their agreeddeadlines.
展开▼
