Dynamic Resource Provisioning Clause Samples

Dynamic Resource Provisioning. This section provides brief overview of recent research in dynamic resource allocation. Kalyvianaki, Charalambous, and Hand (2008) present a feedback controller scheme using Kalman filters for multi-tier applications hosted on virtual machines. They use this scheme for dynamic allocation of CPU resources to virtual machines. The figure 2.5 shows a prototype of their design. Their experiments use the RUBIS Web application deployed on the Xen virtualization platform. Three physical machines are used to host three virtual machines. Tocmcat, JBoss and MySQL are hosted on separate virtual machines. They developed a module called “manager,” that obtains CPU usage information for certain intervals and submit that information to another module called “Controller.” The controller module is responsible for computing the CPU allocation for the next time interval and adjust the CPU resources for specific VMs. In the future, authors hope to demonstrate that their model performs well even when many virtual machines share a single CPU. Nowadays, e-commerce applications are gaining in popularity. E-commerce Web applications contain sensitive resources and require access over HTTPS. Cryptographic techniques are used in e-commerce application to ensure confidentiality and authentication. These techniques are CPU-intensive and may lead to overload condition when many concurrent clients connect to the system. Guitart, Carrera, Beltran, Torres, and Ayguad´e (2008) use admission control and dynamic resource provisioning to develop an overload control strategy for secure Web applications hosted on SMP (Symmetric MultiProcessing) platforms. They implemented a global resource manager for the Linux hosting platform that is responsible for allocating resources to the application servers running on it and ensures the desire QoS in terms of performance stability for extreme overloads. Server machines are able to manage themselves on changes of workload and adapt the load. ▇▇▇▇▇▇ et al. (2007) demonstrate two software systems, ▇▇▇▇▇▇▇ (▇▇▇▇▇ et al., 2006) and ▇▇▇▇ (▇▇▇▇▇▇, ▇▇▇▇▇▇▇▇, & ▇▇▇▇, 2006). Shirako is a Java toolkit for dynamic resource allocation. It allocates virtual resources to guest applications from available resources. ▇▇▇▇ creates an application performance model using active learning techniques. The authors use NIMO to estimate the minimum resources required for a guest application, and they use Shirako to allocate those resources. ▇▇▇▇ builds performance models by capt...