Adaptive Evaluation of Virtual Machine Placement and Migration Scheduling Algorithms Using Stochastic Petri Nets

Abstract

More and more mobile applications rely on the combination of both mobile and cloud computing technology to bring out their full potential. The cloud is usually used for providing additional computing resources that cannot be handled efficiently by the mobile devices. Cloud usage, however, results in several challenges related to the management of virtualized resources. A large number of scheduling algorithms are proposed to balance between performance and cost of data center. Due to huge cost and time consuming of measure-based and simulation method, this paper proposes an adaptive method to evaluate scheduling algorithms. In this method, the virtual machine placement and migration process are modeled by using Stochastic Reward Nets. Different scheduling methods are described as reward functions to perform the adaptive evaluation. Two types of performance metrics are also discussed: one is about quality of service, such as system availability, mean waiting time, and mean service time, and the other is the cost of runtime, such as energy consumption and cost of migration. Compared to a simulation method, the analysis model in this paper only modifies the reward function for different scheduling algorithms and does not need to reconstruct the process. The numeric results suggest that it also has a good accuracy and can quantify the influence of scheduling algorithms on both quality of service and cost of runtime. INDEX TERMS Adaptive evaluation, virtual machine placement and migration, scheduling algorithms, stochastic reward net.

Enhancing the Noise Robustness of the Optimal Computing Budget Allocation Approach

“…Algorithm 1 Sequential Update Procedure for Selecting the Single Best Design Out of k Design Alternatives [18] Control parameters: n 0 and Output: x e (the estimated best design) Procedure: Contrary to the assumptions used by the OCBA procedure, T is actually limited and σ 2 i is typically unknown. In order to effectively utilize the allocation rules in such situations, the OCBA procedure applied a heuristic sequential update procedure, as shown in Algorithm 1, where σ 2 i in (7) and (8) is approximated by the sample variance s 2 i . That is, until a given T is depleted, a small number of simulation replications are allocated in a sequential manner such that the optimal allocation depending on (7) and (8) can be calculated based on more accurate values ofμ i and s 2 i in each iteration.…”

“…In order to effectively utilize the allocation rules in such situations, the OCBA procedure applied a heuristic sequential update procedure, as shown in Algorithm 1, where σ 2 i in (7) and (8) is approximated by the sample variance s 2 i . That is, until a given T is depleted, a small number of simulation replications are allocated in a sequential manner such that the optimal allocation depending on (7) and (8) can be calculated based on more accurate values ofμ i and s 2 i in each iteration. There are two parameters to control this procedure: n 0 and .…”

“…Discrete-event system simulation is a powerful tool for analyzing modern industrial systems such as telecommunication [1], manufacturing [2], microgrid [3], transportation [4], healthcare [5], and military [6] systems that cannot be described as a closed-form analytic model [7]. As one of the essential applications of simulation, simulation-based optimization (SBO) finds the optimal configurations of the systems' decision variables (i.e., the optimal designs) that satisfy the given optimal requirements of the system performance (e.g., maximize the performance) using simulations [8].…”

A Simulation Budget Allocation Procedure for Finding Both Extreme Designs Simultaneously in Discrete-Event Simulation

“…Discrete-event simulation is widely used to analyze modern industrial complex systems, such as manufacturing [1], military [2], smart grid [3], telecommunications [4], and transportation [5]. The most significant advantage of the simulation is that, with just a few assumptions, it can accurately analyze complex systems that cannot be described using closed-form analytic models [6].…”

An Improved Budget Allocation Procedure for Selecting the Best-Simulated Design in the Presence of Large Stochastic Noise

“…Discrete-event system simulation plays a key role in evaluating, analyzing, and optimizing the performance of modern industrial systems such as the telecommunication [1], military [2], manufacturing [3], smart grid [4], and transportation [5] systems, which scarcely meet the assumptions of closed-form analytic models [6]. However, conducting a simulation experiment can be expensive and time-consuming.…”