Distributed Computing Pearls. Gadi Taubenfeld

Чтение книги онлайн.

СКАЧАТЬ biometrics and send them to the patient’s physician; infant monitors that provide parents with real-time information about their babies; and much more.

1.1 WINDS OF CHANGE

      Computers and computer networks are one of the most incredible inventions of the 20th century, having an ever-expanding role in our daily lives by enabling complex human activities in areas such as entertainment, education, and commerce. One of the most challenging problems in computer science for the 21st century is to improve the design of systems where communicating devices interact with one another, and in particular, to invent new applications that will take advantage of these new capacities.

      The fundamental changes in computing architectures, where the whole hardware industry shifted to both mobile and multicore platforms expanding the arena of communicating devices to an unprecedented scale, require a fundamental change in how applications are written for computer networks and multicore computers. Concurrency and synchronization are fundamental issues that are critical for the design of such applications—the Internet could not work without synchronized protocols, database systems would be unthinkable without them, operating systems would crash even more frequently, and concurrent programming would be an impossibility.

      Concurrency exists, of course, not only in computer systems but also in nature and living organisms. We can find concurrency at the molecular level as well as at those of cells, organs, individuals, communities, and ecological systems. Much of the future of computers in the 21st century will be told by how well programmers can take advantage of concurrency.

      This book gently introduces the general reader to some of the basic principles and some of the fascinating results of computer science which involve interactions between computing devices (or processors on the same or different devices), so that the reader can get a feel of the nature of this exciting and interesting field called distributed computing. These fundamental principles and results, underlying the design of algorithms for distributed systems, are presented by comparing issues that arise when dealing with groups of computing devices to those that arise when a group of people has to share their resources and work as a team to solve various problems. The book is self-contained; no reliance is made on previous knowledge in computer science. In particular, all the figures in the book which include code segments can be skipped without loss of continuity.

1.2 THE INTERNET

      The Internet is a vast collection of computer networks which are connected into a single huge network. It is used for the transport of data and messages across distances which can be anywhere from the same room to anywhere in the world. It provides an infrastructure for the use of E-mail, banking, commerce, education, entertainment, telecommunications, manufacturing, and social networks and enables the sharing of remote resources such as databases, files, discs, printers, and computational resources. It also enables several computers to solve problems together much faster than any computer can do alone.

      Historically, the Internet was designed as a research network without the expectation that it would have a significant role in our daily lives. The scale and heterogeneity of the Internet have far surpassed all expectations. As the Internet continues to grow, more and more business-critical functions rely on its availability. The vast majority of communications traffic, including telephone, television, radio, business data, and government data, rely on an Internet infrastructure that is available and secure.

      Many applications of large-scale computer networks, such as the Internet, require a high level of reliability and security, for example, an air traffic control system, spacecraft navigation systems, or an integrated corporate management system. As the system size increases, so does the need to design fault-tolerant applications, for the probability that the entire system functions correctly at any given time rapidly approaches zero. Such applications enable the system as a whole to continue to function despite the failure of a limited number of components.

      Today’s economy involves manufacturing, distributing, and retailing of goods. However, it also has to do with creating and disseminating information, for example, publishing books, filmmaking, etc. Future economy is likely to be dominated by information. Information is a representation of knowledge, and can be represented in two ways: analog—a book that you can hold in your hand; or digital—a book that is stored as a file in your computer. The digital revolution is about converting analog information to digital information and use computer networks such as the Internet to move the digital information around. Such networks are required to be able to move the information in large quantities, everywhere, cheaply, securely, and as fast as possible.

1.3 COMPUTERS WITH MULTIPLE PROCESSORS

      A processor is the brain of the computer. It is a component in a computer that interprets and execute computer program instructions and processes data. Throughout the history of modern computing, application developers have been able to rely on improved processor design to deliver significant performance improvements while reducing costs at the same time. That is, as processors got faster so did the applications. Unfortunately, increasing difficulty with heat and power consumption of the new smaller and faster processors along with the limits imposed by quantum physics has made this progression increasingly more difficult.

      Until a few years ago mainstream computers were built with a single processor only. This situation has changed, as it became more difficult to increase the (clock) speed of uniprocessor computers. Hence, all microprocessor companies have been forced to bet their futures on multiple processors (also called multicores¹) which resides inside a single computer.

      Essentially, all computer manufacturers are now offering a new generation of multiprocessor computers where a single computer includes several processors all executing concurrently, and interact and collaborate with one another. Several computer manufacturers have been building, for many years now, costly high-end computers where each computer includes many processors,² however relatively cheap multiprocessor computers are available today as mainstream computers and can be found in many homes.³ The switch from uniprocessors to multiprocessors is a milestone in the history of computing, and researchers have the rare opportunity to re-invent some of the cornerstones of computing.

      This fundamental change in computing architecture requires a fundamental change in how such computers are programmed. Writing a concurrent application for a multiprocessor computer that takes advantage of having multiple processors to increase speed and get better performance is much more challenging and complex than programming a uniprocessor computer, and requires an understanding of new basic principles. Much of the future of computers with multiple processors will be told by how well programmers can take advantage of the new concurrent computing architecture.

1.4 SYNCHRONIZATION

      Computation on computer networks like the Internet and computation on a single multiprocessor computer have many aspects in common. The key issue in both cases is the need to understand how separate computers on the Internet or, similarly, separate processors within a single computer, interact and synchronize with one another. Synchronization techniques are perceived as essential to design and support the working activities of groups of computers and processors.

      Many of our daily interactions with other people involve synchronization. You and your spouse may have to synchronize on who will buy the groceries, empty the garbage can, take the kids to school, which one of you will be the first to take a shower (assuming you only have one shower at home), will take the car, or use the single computer you have. Assume that you have a cat and your neighbor has a dog and you and your neighbor are sharing a yard, then you and your neighbor might want to coordinate to make sure that both pets are never in the yard at the same time.

      In these examples, synchronization is used to ensure that only one participant (and not both) will take a specific СКАЧАТЬ