Software Defined Networking

Design and Deployment

Patricia A. Morreale, James M. Anderson

• Publisher: CRC Press
• Year: 2014
• eBook ISBN: 9781498760393
• Topic: Computer Science
• Subtopic: Computer Networking
• Index: Computer Science

Software Defined Networking: Design and Deployment provides a comprehensive treatment of software defined networking (SDN) suitable for new network managers and experienced network professionals. Presenting SDN in context with more familiar network services and challenges, this accessible text:Explains the importance of virtualization, particularly

Chapter 1

Virtualization

1.1 Introduction

Virtualization has been an integral part of computing since distributed computing began. Users were initially provided with virtual resources, which were effectively logical subdivisions of a single shared device or devices. The concept of virtualization provided individual, dedicated resources from a larger common pool of resources and provided users with the desired customization and control. The field of virtualization expanded in part because of the limitations of shared resources. Although virtualization has been used since distributed computing started, the integration of virtualization and networking, the heart of software-defined networking (SDN), has been driven and enabled by reductions in hardware cost, advances in software, and limitations in current network configurations.

Servers and hardware costs have declined over time. When a company’s information technology (IT) team created a new application, additional processing needs were handled by purchasing more servers and installing the servers in the corporate data center. Once the new hardware was assembled and available for use, often called “racked and stacked,” the hardware was plugged in, with the new application software installed on the hardware, and the additional software functionality was off and running. The steep decline in the cost of server hardware beginning in the late 1990s and continuing into the first half of the twenty-first century reduced the cost of adding additional servers to corporate IT infrastructure.

This unparalleled, unchecked hardware growth began to slow once data center costs exponentially increased, as overall data center costs also include operating personnel, software licenses, maintenance costs, as well as utility and building costs. Chief information officers (CIOs) and other senior managers concluded, after reviewing data center costs, that although hardware costs had fallen, their total costs for operating a data center had risen and would continue to increase.

Corporate data centers continued expanding without constraints and were still requiring more space. All companies with IT interests were going to have to make sizable investments in building new data centers, which would be an ongoing, increasing cost. However, few of the servers that were in the existing data centers were fully utilized. In fact, many of the servers were running at less than 10% utilization. With the realization that this was an enormous waste of computing resources, it was clear that a new approach was needed to provide resources without excessive capitalization costs and with a sustainable, scalable growth plan. The answer was virtualization.

1.2 Virtual memory

In the mid-twentieth century, when mainframes first appeared in corporate processing centers, a problem quickly appeared. The programs that were being written for the mainframe computers were too large. With the software instructions needed for execution, program variables to be stored, and extra control information that the mainframe’s operating system (OS) used, available memory was quickly exceeded. The IBM 650 computer (circa 1953–1969) had a memory size of only 2,000 words (a word was 36 bits) (Viking Waters, n.d.).

With the arrival of shared computing resources that supported multiple simultaneous users, memory constraints became more visible. Unlike today’s memory expansion approach, for which additional memory is available and affordable, memory was both expensive and limited at the birth of mainframes because early computers could not support memory expansion. This was a significant problem for computer designers.

This problem was solved by “virtualizing” the computer’s memory. The virtual memory technology used permitted a computer with a limited amount of physical memory to use the local hard disk storage system along with physical memory to make it appear that the computer had more memory available for use by application software.

Initially, virtual memory referred to the concept of using a computer’s hard disk to extend a computer’s physical memory. The idea was that programs running on the computer would distinguish between whether the memory was “real” memory (i.e., random access memory or RAM) or disk based. The OS and hardware would determine the actual memory location. This solved the problem of trying to run programs that were larger than the computer’s memory.

Later, virtual memory was used as a means of memory protection. Every program uses a range of addresses called the address space. The use of virtual memory for memory protection solved the problem of allowing multiple users to use the same computer at the same time. The use of virtual memory prevents programs from interfering with each other. If a user’s process tries to access an address that is not part of available address space, then an error occurs. The OS assumes control. The process is usually killed or terminated as a safeguard.

A computer that has been programmed to make use of virtual memory has the additional task of managing its memory. The computer will inspect RAM to determine which programs or data that have been loaded into actual physical memory space have not been used recently. Once such areas have been identified, the central processing unit (CPU) will then copy the programs from RAM to the computer’s hard drive. The RAM that this information had been using is now available for use by other applications.

The copying of RAM contents to the hard drive happens both automatically and quickly. The result is that the end use...

Table of contents

• Preface
• About the authors
• List of abbreviations
• Chapter 1 - Virtualization
• Chapter 2 - Software-defined networking
• Chapter 3 - SDN implementation
• Chapter 4 - Service providers and SDN
• Chapter 5 - SDN development
• Chapter 6 - Network vendors and SDN
• Chapter 7 - Google and SDN
• Chapter 8 - OpenFlow
• Chapter 9 - SDN evolution
• References