Virtual Machine Resource Allocation for Service Hosting on Heterogeneous Distributed Platforms

Abstract

International audienceWe propose algorithms for allocating multiple resources to competing services running in virtual machines on heterogeneous distributed platforms. We develop a theoretical problem formulation and compare these algorithms via simulation experiments based in part on workload data supplied by Google. Our main finding is that vector packing approaches proposed in the homogeneous case can be extended to provide high-quality solutions in the heterogeneous case, and combined to provide a single efficient algorithm. We also consider the case when there may be bounded errors in estimates of performance-related resource needs. We provide a heuristic for compensating for such errors that performs well in simulation, as well as a proof of the worst-case competitive ratio for the single-resource, single-node case when there is no bound on the error

Exact and Hybrid Solutions for the Multi-Objective VM Reassignment Problem

“…Beloglazov et al 1 worked on the reduction of cloud computing data centres' energy footprint and proposed resource allocation heuristics to tackle this problem. Stillwell et al 21 pushed the boundaries further by considering a resource allocation with heterogeneous machines.…”

The HARNESS Platform: A Hardware- and Network-Enhanced Software System for Cloud Computing

“…While work in this area generally acknowledges the underlying variations in capabilities between classes of resources, and much has been made of the differences in how time-and space-shared resources (e.g., CPUs vs memory) are partitioned between users and tasks, these works usually assume uniformity in semantics presented by the interfaces used to allocate these resources. For example, the authors of [39] assume that a multi-resource allocation can be mapped to a normalized Euclidean vector, and that any combination of resource allocation requests can be reasonably serviced by a compute node so long as the vector sum of the allocations assigned to a node does not exceed the vector representing that node's capacity in any dimension.…”

Deadlock Detection for Resource Allocation in Heterogeneous Distributed Platforms

“…For some resources, such as system memory or hard drive space, it is relatively easy for by the different elements together in the hypervisor or the operating system to organize the virtual machine efficiency can interact with only a single major factor. For other types of resources, such as CPU cores, the situation becomes more complicated [2].These resources can be partitioned arbitrarily among virtual elements, but we cannot be effectively pooled together to provide a single virtual element with a greater resource capacity than that of a physical element. For these types of resources, it is necessary to consider the maximum capacity allocated to individual virtual elements, as well as the aggregate allocation to all vital elements of the same type.…”

“…These cloud computing environments provide an illusion of infinite computing resources to cloud users that they can increase or decrease their resources. In many cases, the need for these resources only exists in a very short period of time" [1], [2], [3]. Since them system of information and communication technology (ICT) was introduced, and has played a significant role in the lives of smart cities, the role of information technology infrastructure virtualization has contributed significantly to the solution of the major problem of the succession system of distributed computing, grid computing and parallel computing.…”