Virtual machine migration is widely used in cloud data centers to scale and maintain the stability of cloud services. However, the performance metrics of virtual machine (VM) applications during migration that are set in the Service Level Agreements may deteriorate. Before starting a migration, it is necessary to evaluate the migration characteristics that affect the quality of service. These characteristics are the total migration time and virtual machine downtime, which are random variables that depend on a variety of factors. The prediction is based on the VM monitoring data. In this paper, we select the most suitable factors for forecasting five types of migrations: precopy migration, postcopy migration, and modification of precopy migration such as CPU throttling, data compression, and delta compression of modified memory pages. To do this, we analyzed a dataset that includes data on five types of migrations, approximately 8000 records of each type. Using correlation analysis, the factors that mostly affect the total migration time and the VM downtime are chosen. These characteristics are predicted using machine learning methods such as linear regression and the support vector machine. It is shown that the number of factors can be reduced almost twice with the same quality of the forecast. In general, linear regression provides relatively high accuracy in predicting the total migration time and the duration of virtual machine downtime. At the same time, the observed nonlinearity in the correlations shows that it is advisable to use the support vector machine to improve the quality of the forecast.
The problem of virtual machine placement on physical servers in cloud data centers is considered. The resource management system has a two-level architecture consisting of global and local controllers. Local controllers analyze the state of the physical servers on which they are located and determine possible underloading, overloading, and overheating states based on the forecast for the next observation window. If one of the listed states is detected, the local controller notifies the global controller, which selects the destination servers to host the virtual machines via migration. It is proposed to place virtual machines based on the criteria of minimum remaining unused resources and violation of SLA agreements. A mathematical formulation of the optimization problem is given, which is equivalent to the known main assignment problem in terms of structure, necessary conditions, and the nature of variables. Reducing the assignment problem to a closed transport problem allowed us to solve the problem of hosting virtual machines under many criteria in real time and significantly increase its dimension in comparison with heuristic algorithms, which makes it possible to maintain the quality of modern cloud services in the conditions of rapid growth of physical and virtual resources of data centers. The developed mathematical formulation of the problem and the results of computational experiments can be included in the mathematical software of virtual machine live migration.
Dynamic resource allocation of cloud data centers is implemented with the use of virtual machine migration. Selected virtual machines (VM) should be migrated on appropriate destination servers. This is a critical step and should be performed according to several criteria. It is proposed to use the criteria of minimum resource wastage and service level agreement violation. The optimization problem of the VM placement according to two criteria is formulated, which is equivalent to the well-known main assignment problem in terms of the structure, necessary conditions, and the nature of variables. It is suggested to use the Hungarian method or to reduce the problem to a closed transport problem. This allows the exact solution to be obtained in real time. Simulation has shown that the proposed approach outperforms widely used bin-packing heuristics in both criteria.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.