VM Requests Research Articles

POF-SVLM: pareto optimized framework for seamless VM live migration

Live migration helps to achieve resource consolidation and fault tolerance. It transfers VM storage together with VM memory and CPU status. During migration, a dirty page rate also delays the period of live migration, and it affects the performance of migration by increasing migration time, network bandwidth consumption, CPU processing overheads and application downtime. Experimental results after comparing with existing methods of VM live migration, reflects that with high data transfer rate, prolonged migration time and downtime make it infeasible to achieve seamless live migration. This paper provides a detailed analysis of the KVM strategy for live migration. It shows that the KVM iterative copy method, where all RAM data is marked as dirty and transferred during the first iteration, initially generating majority of additional overheads, mainly due to large data transfer. Based on these findings, the innovative Pareto Optimized framework [POF-SVLM] was developed and deployed as a standardized VM storage network, such as the Network Attached Storage (NAS), shared between the source and the target machine. Only the VM primary storage is to be transferred during live migration. Additionally, the architecture effectively monitors all I/O VM requests to determine unique pages in the primary memory that are only transferred during the iterations. All duplicate pages are downloaded directly from the NAS to the target machine. Detailed experimental evaluation shows that the proposed mechanism reduces the VM live migration overheads by 55–60%. Experimental results also shows that the downtime of POF-SVLM based live migration is in the range less than 60 s even for highly intense workloads. This is the most significant contribution of POF-SVLM because since the downtime less than 60 s, the user will not be able to notice it and results in seamless VM live migration.

Secure virtual machine placement in cloud data centers

Abstract Due to an increasing number of avenues for conducting cross-VM side-channel attacks, the security of multi-tenant public IaaS cloud environments is a growing concern. These attacks allow an adversary to steal private information from a target user whose VM instance is co-located with that of the adversary. In this paper, we focus on secure VM placement algorithms which a cloud provider can use for the automatic enforcement of security against such co-location based attacks. To do so, we first establish a metric for evaluating and quantifying co-location security of multi-tenant public IaaS clouds, and then propose a novel VM placement algorithm called “Previously Co-Located Users First” which aims to reduce the probability of malicious VM co-location. Thereafter, we perform a theoretical and empirical analysis of our proposed algorithm to evaluate its efficiency and security. Our results, obtained using real-world cloud traces containing millions of VM requests and thousands of actual users, indicate that the proposed algorithm provides a significant increase in the cloud’s co-location resistance with little compromise in resource utilization, compared to existing approaches. We also explore the potential for cloud providers to leverage passive cache monitoring techniques as an additional security measure in order to automatically improve the co-location resistance provided by general VM placement algorithms.

Network affinity aware energy efficient virtual machine placement algorithm

Efficient mapping of virtual machine request to the available physical machine is an optimisation problem in data centres. It is solved by aiming to minimise the number of physical machines and utilising them to their maximum capacity. Another avenue of optimisation in data centre is the energy consumption. Energy consumption can be reduced by using fewer physical machines for a given set of VM requests. An attempt is made in this work to propose an energy efficient VM placement algorithm that is also network affinity aware. Considering the network affinity between VMs during the placement will reduce the communication cost and the network overhead. The proposed algorithm is evaluated using the Cloudsim toolkit and the performance in terms of energy consumed, communication cost and number of active PMs, is compared with the standard first fit greedy algorithm.

Network Affinity Aware Energy Efficient Virtual Machine Placement Algorithm

Efficient mapping of virtual machine request to the available physical machine is an optimisation problem in data centres. It is solved by aiming to minimise the number of physical machines and utilising them to their maximum capacity. Another avenue of optimisation in data centre is the energy consumption. Energy consumption can be reduced by using fewer physical machines for a given set of VM requests. An attempt is made in this work to propose an energy efficient VM placement algorithm that is also network affinity aware. Considering the network affinity between VMs during the placement will reduce the communication cost and the network overhead. The proposed algorithm is evaluated using the Cloudsim toolkit and the performance in terms of energy consumed, communication cost and number of active PMs, is compared with the standard first fit greedy algorithm.

A Two-Dimensional SLA for Services Scheduling in Multiple IaaS Cloud Providers

Customers of cloud services choose the VMs profiles (SLAs) offered by the provider, and pay according to how long these VMs are utilized. Many works deal with how to decrease the cost of VM requests scheduling, but consider solely the charging models in the SLA. However, other characteristics in the SLA must be taken into account when choosing a VM to execute users' applications (e.g. processing capacity). In order to fulfill the user needs and allow proper utilization of resources available in IaaS providers, this paper models a two-dimension SLA, namely charging model and VM type. The problem is modeled as an integer linear program to compute the scheduling regarding this SLA model. Simulations show that the proposed approach computes schedules that better fits the user needs and allow better utilization of VMs, resulting in a higher number of fulfilled requests than alternative approaches.

Overload Avoidance Model using Optimal Placement of Virtual Machines in Cloud Data Cetres

Cloud data centres improve CPU utilization of their servers (physical machines or PMs) through Virtualization (virtual machines or VMs). Over virtualised and under virtualized PMs suffer performance degradation and power dissipation respectively. This work presents a stochastic modular scheme for allocating VM requests to a PM by avoiding overloading of PM and keeping the global load characteristics under specified QoS goal. The proposed approach categorizes PMs into three groups (Under Load, Normal Load, Over Load) in a way that minimizes number of PMs in Under Load and Over Load groups and maximizes number of PMs in Normal Load group. We compute VM request rejection probability, response time, service time and number of PMs that are overload or under loaded for evaluating the performance of our model. The results show that these parameters do not degrade with increasing arrival rate. Thus the proposed model is simple yet efficient approach for VM placement problem.