Presenting the migration scheduling algorithm of virtual machines to simultaneously optimize energy consumption and pollutant production in the cloud computing network

Computer Engineering Master's Thesis

Abstract:

In recent years, due to the growing demand and new customers joining the world of computing, computing systems must also change and be more powerful and flexible than before. In the meantime, cloud computing was presented as a model beyond a system that currently has the ability to respond to most requests and requirements. Virtualization solutions are widely used to solve various problems of modern data centers, which include: less use of hardware, optimal use of data center space, high system management and maintenance costs. The main challenges that large servers face are the lack of high system reliability. And high operating costs due to high energy consumption. Therefore, energy-aware VM deployment and scheduling is an urgent necessity to achieve these goals.  Work scheduling has been studied by various researchers for several years, but the development of virtual clusters and the cloud environment have opened a new window to researchers for new scheduling approaches. One of the techniques required to increase the flexibility and scalability of cloud data centers is migration. The act of migration is carried out with various goals, including load balancing and distribution, tolerance against failure, energy management, reducing response time and increasing service quality, server maintenance.

The main components of work scheduling in the virtual environment include: the deployment of VMs among physical machines and dynamic load balancing with the help of work migration across the nodes of the data center cluster.

In this thesis, we focus on the migration schedule of virtual machines in the cloud data center using algorithms. It is hereditary. The simulation results confirm the feasibility and efficiency of this scheduling algorithm and lead to a significant reduction in total energy consumption compared to other strategies. And since our focus is on the operational energy of data centers, with the reduction of operational energy consumption, the production of carbon biopollutants has also decreased, which plays a significant role in reducing the user's cost. 1969 Leonard Kleinrock [1], one of the chief scientists of the Agency's Advanced Research Projects Network (ARPANET), the founder of the Internet, said: "Now, computer networks are still in their infancy, but when they grow and become more complex, we will probably see the expansion of the general computer industry" that, like electricity and telephones today, will serve homes and offices in the country. This vision of the ubiquitous computing industry based on the service provider model predicts a massive transformation of the entire computing industry in the 21st century whereby computing services will become on-demand [1], as readily available as other utility services in today's society. Similarly, users (consumers) pay providers' computing services only when they access them. In addition, consumers will no longer need to invest heavily in building and maintaining complex IT infrastructure. In such a model, users can access the services they need regardless of where those hosting services are located.  This model is referred to as universal computing or recently cloud computing [2]. The latter represents the infrastructure as a "cloud" from which users can access applications as a service on demand from anywhere in the world. Therefore, cloud computing can be introduced as a new paradigm for the dynamic provision of computer services with the support of data centers, which usually uses virtual machine technology3 for stabilization and environmental purposes.[3] .

Until recently, the only concern was deploying high-performance cloud data centers without any consideration for energy consumption. On average, data centers consume as much energy as 25,000 households[4] A significant portion of the electrical energy consumed by computing resources is converted to heat. High temperature leads

to a number of problems, such as reduced system reliability and availability, as well as reduced lifespan of devices.

Data centers are not only expensive, but also cause serious damage to the environment.  The carbon produced by these centers is currently more than the carbon produced by a country with all its industrial facilities [5]. The high carbon footprint is due to the large amount of electricity required to cool the many servers hosted in data centers. Cloud service providers need to take necessary measures to ensure that their profits do not decrease significantly due to high energy costs, and at the same time, they can provide services with higher quality and speed, also due to the increasing pressure from world governments to reduce the amount of carbon, which has a significant impact on climate change. [6].

In the meantime, virtualization solutions, in addition to reducing energy costs in data center infrastructures, dramatically increase server utilization and efficiency. Virtualization using the migration technique of virtual machines, by integrating and balancing the load between physical servers and running ten or more virtual machine applications on one x86 server, can transfer virtual resources between physical servers with great flexibility [7].

Generalities

Introduction

In this chapter, an introduction about the emergence of the cloud computing network and a series of general discussions about virtualization[3] and the migration of virtual machines[4] in the cloud environment and genetic algorithm[5] are stated. In the following, an overview of the cloud, familiarity with the important issues and challenges in the cloud, the necessity of conducting this research and the research objectives are given. This chapter is finished with the summary and conclusion of the chapter as well as the description of the structure of the thesis.

Overview of cloud computing

The evolution of computing is such that it can be assumed as the fifth basic element after water, electricity, gas and telephone. , access it. There are various examples of computing systems that try to provide such services to users. Some of these computing systems include: cluster computing [6], network computing [7] and recently mass computing [8] which is referred to as cloud computing. The popularity of these three computing approaches has been evaluated from the perspective of the Google search engine, the result of which is shown in Figure 1-1, and it indicates that the popularity of cloud computing, after the emergence of its initial concepts in 2007, is increasing by a large distance compared to other computing approaches [8].

To better understand cloud computing from the perspective of infrastructure, we first take a look at the evolution of computing systems from the beginning until now so that we can place it among others. Identify systems. If we consider main computers [9] as the first generation of computing systems, we were faced with a very large system that users accessed through a single terminal [10]. Over time, these systems became smaller and became available to all users as personal computers with more processing power. Then it became possible to connect a set of these small systems to provide a network with more processing power to meet more and heavier processing needs. But the processing needs were increasing and the need for bigger and stronger computing systems was felt. Therefore, a large number of these networks were connected exclusively throughout the Internet and created the grid computing network. In the meantime, it was observed that there are millions of users on the Internet who do not use the full power of their computers most of the time, and another computing system was formed so that users who wish to donate their idle time for public computing tasks. Therefore, a large number of small computing resources joined together in a network called voluntary computing and created a huge processing power.

But there were still many other resources in organizations and Internet data centers whose full capacity was not fully utilized. These resources could not be used exclusively in the grid computing network, because another task was defined for them.