Sensitive instructions: the sensitive instruction sets are those which tends to interact with the system hardware. Both the control and behavior sensitive instructions fall in this category of instruction set (Pan and Liu 2015). The control sensitive instructions are those which can change the configuration of the resources. On the other hand the instructions that are dependent on the resource configuration for its result, is called the behavior sensitive instruction set.
Benefits and the drawbacks of the Oversubscription
For the virtualization it can be said Virtualization empowers the servers or specifically the data centers to concentrate on the requirements of business applications. In addition with that the virtualization also makes the resource allocation to the application more complex (Ferdaus et al., 2014). To overcome these complexities of allocation of the resources, the administrators try to oversubscribe the physical assets on a host with the goal of expanding the quantity of workloads on a host. The oversubscription of the resources happens when the requests from the clients for the resource exceeds the actual amount of the available physical resources. By doing this the different resources like the servers, services, storage spaces are shared. While the different strategies are developed to use the resources optimally, its observed that the most of the resources remains idle at the highest work load. So the different benefits of suing the oversubscription of the resources are
- Proper use of resources: The oversubscription helps in reducing the waste and operational cost of the resources and improving the amount of revenues from the business (Beloglazov 2013).
- More number of users: By using the oversubscription method for the resources it can be possible to serve more number of customers by proper utilization of smaller portion of the resources.
- High availability: As the virtual machines are scalable according to the need, if one of the servers go down or fails to serve due to the maintenance the client request can be served by another server resource with the minimal downtime and loss of data.
Figure1: Utilization of the physical resources.
(Source: Householder et al., 2014 pp-2)
Drawbacks of Oversubscription: The process of oversubscribing the available resources is not free from different flaws and drawbacks. Like
- Performance issues: If the memory is overloaded with the excessive client requests, it is possible that restrains the progress of any other application. Expansive overheads and can truly block and affect the execution speed of the system
- Economic issues: Again if the oversubscription of the resources is not managed properly then it can lead to the violation of the service level agreement, which is signed between the client and the service provider (Ho et al. 2015). In rare cases if the SLA are legally bonded then it is possible that the service provider organization is forced to compensate the client for not meeting the service levels. This process leads to the economical and reputational loss of the service provider.
- Downtime Due to crash: It is almost impossible that all the resources in virtual computing fail at the same time (Xiao et al., 2013). In addition with the above statement it is also true that if there is cascaded hardware failure then the total system needs a lot of time to go live again after recovering from the crash.
Utilization of resources
Impact of virtualization on measuring the CPU, RAM, and I/O utilization
The virtualization of the CPU utilization is a complex issue. Since the allocation of the different tasks to the core are done using the Hypervisor than the guest OS. So to measure the CPU utilization the user has to use different hypervisor aware performance counters. This counter can be used to tack the CPU utilization of a virtual processor that is assigned to some virtual machine (Mishra et al. 2012 ). The different components of the hypervisor components are guest runtime, hypervisor run time, total runtime, total intercepts /sec, hyper calls cost, hypercalls/sec etc.
RAM utilization: Both in the host and virtual machines the measurement of the memory utilization is difficult. In case of the virtual machines if it is suspected that the memory is leaked or is not optimally oversubscribed then at first it is determined that what is to be checked. After that the counters like, Ballooning, VMkernel, vSphere etc. (Baucke et al. 2013).
I/O utilization: In the virtual networks the input and output operations are limited and the users have a limited scope for the I/o operation to the virtual services.
Indication for configuration change depending on the resource consumption by the physical machines
A server or a physical machine with a low load will run cooler, and have lower chances of failure. Once the CPU achieves the most extreme temperature, the quality of the processes become much lower with higher amount of burdens (Oki et al. 2015). The reality of the matter is that the cooling framework will make fans move speedier and that will raise the likelihood of a disappointment on the fans. There is another issue, with such a high load, if the heat is not taken out of the closed area in an effective way, then the entire server will have a higher temperature, influencing different parts of the virtual system. So for reliable processing and great speed of processing, it’s recommended to increase the capacity of the system.
Indication for the configuration change depending on the resource allocation by the virtual machine
It is often seen that the high CPU usage helps in increasing the host ready time and the queuing of the processors too. Moreover it can be said that if the CPU is constantly used at a high rate (like 90%) of its capacity then the service provided by it is impacted. Therefore comparing the CPU usage of a virtual machine with other machines available in the host, increasing the CPU limit, and upgrading the number CPUs or cores would help in maintaining the same level of services by the provider.
Justification about the change in the configuration when the resource consumption is 30%
As the host installed have the different number of available processors and cores. When the resource consumption is at 30% then there is no need of increasing the capacity. Moreover it can be said that the provider can accommodate more number of requests from the clients and process them
Description of the behavioral characteristics of the three instruction set architectures proposed by Popek and Goldberg
Goldberg and Popek proposed some requirements that are helpful for a computer system to efficiently manage the system virtualization. They proposed three types of instruction sets, which are privileged, non- privileged and sensitive.
Privileged instruction: These are the instructions that are trapped in the system if the system is in user mode and is executed if the system is in kernel mode. This privileged instruction sets are used to separate the user space kernel space.it is required because the kernel space accesses hardware resources directly and the user space do that in an indirect way (Gracia-Moran et al. 2014). If the privilege instructions are tried to execute in the non-privileged mode then the system triggers an alarm and execute that in the privileged environment.
Non privileged instructions: these are the instructions that are executed in the user mode in the virtual system. For the execution it requires the user space rather than the kernel space.
References
Bai, T. and Rabara, S.A., 2015, August. Design and Development of Integrated, Secured and Intelligent Architecture for Internet of Things and Cloud Computing. In Future Internet of Things and Cloud (FiCloud), 2015
Bari, M.F., Boutaba, R., Esteves, R., Granville, L.Z., Podlesny, M., Rabbani, M.G., Zhang, Q. and Zhani, M.F., 2013. Data center network virtualization: A survey. Communications Surveys & Tutorials, IEEE, 15(2), pp.909-928.
Barroso, L.A., Clidaras, J. and Hölzle, U., 2013. The datacenter as a computer: An introduction to the design of warehouse-scale machines.Synthesis lectures on computer architecture, 8(3), pp.1-154.
Baucke, S., Green, H., Kempf, J. and Tatipamula, M., Telefonaktiebolaget LM Ericsson (Publ), 2013. Using MPLS for virtual private cloud network isolation in openflow-enabled cloud computing. U.S. Patent 8,560,663.
Beloglazov, A., 2013. Energy-efficient management of virtual machines in data centers for cloud computing.
Beloglazov, A., Abawajy, J. and Buyya, R., 2012. Energy-aware resource allocation heuristics for efficient management of data centers for cloud computing. Future generation computer systems, 28(5), pp.755-768.
