Lecture notes on cloud computing IV b. Tech-1 st semester prepared by
Download 1.49 Mb. View original pdf
|
english grammar pdf 60
| Grid and Cloud Integration To understand why grids and clouds should be integrated, we have to start by considering what the users want and what these two technologies can provide. Then we can try to understand how cloud and grid can complement each other and why their integration is the goal of intensive research. The integration of cloud and grid, or at least their integrated utilization, has been proposed since there is a trade-off between application turnaround and system utilization, and sometimes it is useful to choose the right compromise between them. Some issues to be investigated have been pointed out ● Integration of virtualization into existing e-infrastructures ● Deployment of grid services on top of virtual infrastructures ● Integration of cloud-base services in e-infrastructures ● Promotion of open-source components to build clouds ● Grid technology for cloud federation In light of the above, the integration of the two environments is a debated issue . At the state of the art, two main approaches have been proposed ● Grid on Cloud. A cloud IaaS (Infrastructure as a Service) approach is 84 adopted to buildup and to manage a flexible grid system Doing so, the grid middleware runs on a virtual machine. Hence the main drawback of this approach is performance. Virtualization inevitably entails performance losses as compared to the direct use of physical resources. ● Cloud on Grid: The stable grid infrastructure is exploited to buildup a cloud environment. This solution is usually preferred because the cloud approach mitigates the inherent complexity of the grid. In this case, a set of grid services is offered to manage (create, migrate, etc) virtual machines. The use of Globus workspaces [16], along with a set of grid services for the Globus Toolkit 4, is the prominent solution, as in the Nimbus project . The integration could simplify the task of the HPC user to select, to configure, and to manage resources according to the application requirements. It adds flexibility to exploit available resources, but both of the above-presented approaches have serious problems for overall system management, due to the complexity of the resulting architectures. Performance prediction, application tuning, and benchmarking are some of the relevant activities that become critical and that cannot be performed in the absence of performance evaluation of clouds. Download 1.49 Mb. Share with your friends:
|