Security is one of the topmost concerns of any computing model and cloud computing is no exception. Consumers or enterprises moving into cloud need to feel secure about their computing facilities and more importantly about the data they are disclosing to service providers. This is a critical choice for enterprises accustomed to safeguarding data sitting in their own centers in the traditional way of computing. Cloud computing promotes the concept of working on proprietary data and applications outside their jurisdiction.

It is often said that, ‘cloud computing is an excellent idea, but cloud security is not good’. The common perception is that the cloud services are not inherently secure. Cloud computing creates scope for scores of possibilities although like any other technology, there are some risks associated with it which can be overcome if understood correctly.

A cloud computing environment can be formed in different ways and there can be several ‘cloud formations’. Each form is not suitable for every kind of business operations. Consumers must first understand which among these forms which is best-suited for their purpose. This little judgment can make the security of cloud computing even better than that of traditional in-house computing environments. This chapter focusses on these important aspects.

THE SECURITY CONCERN IN CLOUD

Traditional computing systems used to create a security boundary by placing firewalls at the gateways through which the network used to communicate with the outer world. Firewall blocks unwanted traffic trying to access the network it protects and thus only authenticated accesses are allowed into the system. Thus, malicious accesses get blocked at firewalls in the traditional data centers in order to keep the system protected from outside threats.

But this strategy only makes sense when all applications and data reside within one network of security perimeter. Traditional data centers allow perimeterized (i.e. within organization's own network boundary or perimeter) access to computing resources. But, the de-perimeterization (to open-up the interaction with outer network) and erosion of trust boundaries that was happening in enterprise applications, have been amplified and accelerated by cloud computing.

With appropriate capacity planning in cloud computing, the large pool of virtualized resources creates an illusion of an infinite source of computing resources. In the preceding chapters, it has been discussed how applications often need to scale to manage the varying workload and how resources are dynamically provisioned in order to meet the needs of scaling in cloud computing. It has also been discussed that horizontal scaling is more prepared for large scale scaling in distributed environment like cloud.

Upon obtaining provisioned or de-provisioned dynamic resources, the horizontal scaling provides elasticity to cloud computing. But, this flexibility comes with lots of implementation complexity. Load balancing is the mechanism through which these features are implemented. Apart from these, as there are multiple resources available to serve a particular type of service request, and in a distributed computing environment it becomes necessary to distribute proportionately those service requests among the available resources so that none of them becomes overloaded and degrades the performance of the entire system. It is the run-time logic or algorithm of distributing the service requests that has to be capable of evenly distributing the workload. This chapter discusses different types of load balancing techniques along with their pros and cons. The possible parameters for consideration while working out load balancing algorithms have also been discussed. A properly-designed workload distribution and balancing architecture reduces resource underutilization, improves the performance of a system and also brings down the cost of computation.

LOAD BALANCING

Load balancing in distributed computing systems is an essential technique to distribute processing and communication activities evenly across the resources in the network so that no single computing resource gets overloaded. Among different exercises, load balancing is especially important for applications which often deal with a large and unpredictable numbers of service requests. Load balancing is a technique that distributes load evenly among multiple computing resources such as processor, memory etc.

Through load balancing, the incoming service requests are distributed among available computing resources. An efficient load balancing mechanism improves average resource utilization rate and therefore enhances the overall performance of a system.

The easy access to high-performance computing resources in cloud computing has not only made the process-intensive activities smarter but the data-intensive computing activities also have taken center stage. The nature of data have radically changed with this revolutionary utility service; hence their processing and storage requirements vary. Large data-sets are generated and produced everyday are sent for processing in the high-performance computing environments.

Like the traditional storages, users can store and access multimedia files of various formats like text, image, audio and video in cloud also, but the storage requirements have been altered for efficient processing of the large data-sets which are produced in cloud every hour. The traditional enterprise level files and data storage systems were not sufficient to satisfy all of the data-intensive and high-performance computing requirements.

Efficient file handling to support parallel and distributed operations of large data-sets needed an entirely new file system format. Hence, researchers and computing vendors have come up with suitable storage solutions to achieve optimal performance in cloud like high-performance environment. This chapter focuses on all of these advancements made to fulfill those requirements.

REQUIREMENTS OF DATA-INTENSIVE COMPUTING

Data-intensive computing presents a challenge to computing systems in terms of delivering high-performance. Large volume complex data-sets cannot be processed centrally in a single node and require partitioning and distribution over multiple processing nodes. Thus, data-intensive computing is I/O-bound and requires rapid movements of data in large numbers. This requires appropriate management of data in transaction. Data modelling, partitioning, node assignment and accumulation are some of the critical parts of this computing. Consumers’ requirements and technical aspects related to the storage facility in high-performance computing environment are different from the traditional storage system in many ways. Traditional enterprise storage systems are no more sufficient to tackle those issues.

Scalability and high-performance distributed data processing are complementary to each other. Distribution of large data-set among as-many-nodes as required for processing promotes scalability of the application. Suitable file systems are required to support this distribution and scaling proficiently. It has been observed that data-intensive computing often involves process-intensive computing too. Complex data-sets present challenges before the computing system.

The future often arrives silently. In the digital world, cloud computing which emerged silently, has already made a significant place of its own and has shown enough potential to be the future of computing. The advantages it offers are plenty, and this is what makes cloud computing revolutionary.

Books on this evolving subject started appearing in the market towards the end of the last decade. A handful of books have been published so far, and they provide useful resource material. While a few among them have been written in order to discuss conceptual aspects of the subject, others have been written from research or application perspective. As the research or application oriented books serve specific purposes only, they are not appropriate for general readers of the subject.

Cloud computing is not a single technology; rather, it is a combination of multiple technologies. Hence, it is necessary to understand all those individual technologies in order to understand cloud computing as a whole. Many books have been written with their focus on some particular aspects of cloud computing, and thus, they have not been able to cover all the essential topics of cloud computing with equal importance. This book is written from conceptual perspective, for the sake of new learners of the subject, and it tries to fill the void for a book covering all the fundamentals in detail, with simplified explanations in a learner friendly manner.

The author felt the need for such a book while teaching the subject to the undergraduate engineering students, as well as software professionals who wanted to develop a basic understanding of the subject. It is either inadequate coverage or a complex approach of most books available in the market, that makes it difficult for new learners to develop a clear conception about the various aspects of these technologies, and they therefore do not satisfy the fundamental queries of students. Thus, it is a lack of books covering all the fundamentals in detail which encouraged the author to write this book.

The author is aware of the fact that computing and IT industry are fast merging into a singular cloud computing environment, and in the near future there will be a requirement for a large pool of workforce with sound insight into this subject.

Cloud computing delivers virtualized resources. In previous chapter, it has been discussed how virtualization technology has revolutionized the computing environment. It is important to note that in reality the virtualized resources are effective representations of actual physical resources. Cloud data centers maintain a large pool of physical computing resources which is presented in virtual mode as infrastructure service. This chapter focuses on how those resource pools are organized in cloud data centers.

Resource sharing is another important feature of cloud computing. Any resource component from the pool of resources is not dedicatedly used to serve one particular user or application; rather they are shared among multiple users or applications. Through this increment in resource utilization rate, a comparatively smaller number of resources can serve a large number of users.

Resource pooling, virtualization and sharing of resources ultimately converge to produce the dynamic behaviour of cloud systems. Cloud supplies the resources dynamically to users. This dynamic resource provisioning eliminates the need of permanent procurement of computing resources. This turns out to be a very effective technique as application demands or loads during computation do not remain same over a period of time.

RESOURCE POOLING

The utility service model of cloud computing requires to maintain huge amount of all types of computing resources to provide different services to consumers. For this purpose, cloud service providers create pool of computing resources. Effective pooling or grouping of resources requires appropriate system designing and architectural planning.

Resource pooling in cloud needs setting up of strategies by provider for categorizing and managing resources. Before resource pooling concept appeared, earlier people used to maintain discrete and independent set of resources known as silos.

In traditional computing model, the silos are made with very little or no inter-connections. On the other hand in cloud computing, the consumers use well-connected pool of computing resources. They gain almost no knowledge or control over the locations from where physical resources are allotted to them. In its out-of-best scenario, the providers sometimes ask for choice of geographic location (country or continent) from where a consumer wants to get resources. But this choice is only possible in case of large service providers who have data centers at multiple geographic locations around the world.

The first three chapters of this book provide an overview of introductory cloud computing issues. Beginning with the discussion about limitations of traditional computing approach, it focused on the primary benefits of cloud computing adoption. The second chapter also explained how starting from the days of mainframe computers, cloud computing has matured over the years with continuous developments and reformations in the field of computing.

With this basic understanding of the primary concepts and knowledge regarding the technological drivers behind cloud computing, it is now time to delve further. Now, it will be appropriate to understand the architectural model of cloud computing.

In this chapter, the standard definitions and globally accepted NIST (National Institute of Standards and Technology) model of cloud computing have been explored in order to establish a baseline. NIST has identified five characteristics which have collectively revolutionized this computing model. Apart from these, different types of cloud deployments have also been discussed. Understanding the cloud deployment models is important as it is critical for making informed decisions so that appropriate cloud solution can be picked to satisfy personal or business needs.

STANDARD CLOUD MODEL

Cloud computing is a reality now and is being delivered commercially. Decades of research and several innovations have paved the way for its success. Though the concept surfaced several years back, the implementation of the concept has been possible through different phases of remodeling and renovation over the years.

During its early days, the idea was defined in many ways by different sections of people as they presented it from their respective view points. Academicians, computer engineers, system architects, software developers, and consumers had their own ideas and suggestions about the models for this new approach of computing. All these proposals had several limitations as there were many aspects which had to be considered. Formation of a complete definition which could address all of these aspects was thus felt necessary. A standard model was also required as it becomes substantive for the advancement of any technology.

In cloud computing everything moves around the globe through a huge network called Internet. All requested services and data from any public cloud are delivered to consumers via Internet. The delivery of the content from the source to the consumer may involve delays as the two may be far apart geographically from one another. This delay in content delivery in cloud computing could increase the response time barrier in acceptance of the technology. This delay may further increase due to heavy traffic over the network. And, studies have shown that delay in response time can cause huge business loss.

Cloud computing is made not only for network-centric computing but also for network-centric content management. To serve the purpose of instant delivery of content via network to the consumers, technologists have introduced the concept of a dedicated network for faster content delivery. Content generators like cloud service providers can make use of these networks for efficiently delivering their content to millions of consumers across the globe.

CONTENT DELIVERY IN THE CLOUD

Content is any kind of data ranging from text to audio, image, video and so on. Delivering this content to any location at any time is a critical issue for the success of cloud services. Smooth delivery of content and a positive user experience have always been the prime concern of content makers and content providers in world-wide networked environment like cloud computing. The primary objective is to minimize the page (that is the content in a page) load time as studies have shown that even a delay of a second in page loading time can negatively affect customers’ experience significantly.

Forecasting any failure in network delivery and overcoming it has always been a challenge for content suppliers and the scenario has turned more critical with the adoption of network-centric cloud computing environment. The network requirements for delivering content have changed extensively during the last decade when large volumes of multimedia content started pouring in and consumers’ expectation regarding content availability over Internet shot up.

Computing performance over cloud-based services is very important to encourage cloud service adoption as consumers are accustomed to working on their personal work-stations in traditional way of computing.

The introduction of cloud computing creates numerous benefits for consumers of computing services but it also has security concerns associated with it. In addition to most security risks present in the traditional computing environment, cloud computing brings a new set of security issues with it. These issues emerge basically due to the massive sharing of infrastructure and resources which is an inherent feature of the utility service model.

A proper understanding of cloud architecture and appropriate selection of cloud deployment may reduce the security risks to a great extent. It is also important to understand that security maintenance is a combined effort for both the provider and the consumer of cloud. This chapter tries to focus on the security aspects of cloud from different service levels, apart from discussing the cloud security design principles and frameworks.

Cloud-based security systems need to address all the basic needs of an information system like confidentiality, integrity, availability of information, identity management, authentication and authorization. These security requirements are not new, but cloud computing specific standpoint on these issues are essential to analyze as well as implement security in a cloud-based system.

CLOUD SECURITY

For an enterprise, sharing infrastructure in cloud computing environment is like a person going to some public place with valuable belongings. Others with wrong intent may anytime target those valuable belongings. Similarly, moving sensitive files or data out of an enterprise's own network security boundary, in cloud computing, also causes security concerns. Hence, implementation of strong identity management and access control mechanism are important in cloud environment.

Security is developing resistance to damage or protecting a system from any harm. Security is a major concern for any computing system and also for the cloud computing system too. Computing system security means protection of the system itself and also of the data that it stores.

When an IT service consumer, be it an individual or an organization, migrates to cloud computing, especially in public cloud services; much of the computing infrastructure moves into the control of third-party cloud services providers.

EVALUATION OF PARALLEL ALGORITHMS

In order to solve computational problems by appropriate algorithms, we need to carry out their analysis. The goal of analysis is to demonstrate correctness of the algorithms (see Section 1.3) and to examine their properties that are crucial in terms of suitability for solving a problem at hand, as well as to compare performance of the algorithms and select the best one. Another aspect of analysis may be to determine whether the best algorithm we have is optimal, or in other words, no better algorithm exists.

A sequential algorithm is a sequence of precisely defined steps solving a problem, that is computing the output data based on the input data. The precision in describing the steps is necessary to be able to transform the algorithm into a computer program that is the implementation of the algorithm. The computation of a sequential algorithm is performed by one processor.

If we want to solve a problem by using multiple processors, then the problem should be decomposed into subproblems that can be solved in parallel. Each of the subproblems is solved by a separate algorithm, which is a component of a parallel algorithm. So the parallel algorithm consists of a certain number of separate algorithms that can be executed simultaneously. The parallel algorithm also specifies how its components use intermediate results of computation and how they synchronize their work.

The basic criteria for evaluating sequential algorithms include the running time and memory requirement. The running time—or more precisely, the compute time of a processor—and the occupied memory are the amounts of resources necessary to execute an algorithm. Other evaluation criteria of sequential algorithms include universality, understood as possibility of applying an algorithm to solve a class of problems, simplicity, ease of implementation, and modification.

Evaluating a parallel algorithm is more difficult than a sequential one, because an additional factor (except the size of input data) is to be considered, which is the number of processors of a parallel computer where the algorithm is implemented. A number of evaluation criteria is also greater.

INTRODUCTION

One of the types of parallel processing is distributed computing. This type of processing can be conducted in integrated computers with distributed memory, or in clusters, which are systems of homogeneous or heterogeneous networked computers. In distributed computing the tasks communicate via communication channels (or links). The channels form an interconnection network between processors or computers. Processors or computers that are vertices of a network perform computing tasks, as well as send and receive messages.

In this chapter we explore how to implement parallel programs that consist of tasks cooperating with each other using message passing. Parallel programs should be written in a suitable programming language. Probably the only language specially developed to describe parallel computing with message passing was occam. This language proposed by May et al. in Inmos company was based on the CSP notation (acronym for Communicating Sequential Processes) defined by Hoare. In the 1980s occam was used as the programming language for transputers—systems of large-scale integration, each combining a processor and four communication channels. Along with development of computer hardware, it turned out that occam due to certain weaknesses and restrictions is insufficient to describe distributed computing. Nowadays, these are often carried out using C or Fortran languages augmented by functions intended for cooperation of parallel processes. The most popular libraries of such functions are PVM (Parallel Virtual Machine) and MPI (Message Passing Interface).

The PVM library was developed in Oak Ridge National Laboratory. It permits the creation and execution of parallel programs in heterogeneous networks consisting of sequential and parallel computers. Another popular and largely universal library that is used to build distributed programs is MPI. It can be applied together with the OpenMP interface in computers with distributed memory (see Section 5.4.3), in particular in clusters composed of multicore processors or SMP nodes (see Sections 5.4.2 and 5.4.3, and Chapter 7). The library is highly portably enabling to build scalable programs for applications where achieving high computational performance is essential.