For your convenience Apress has placed some of the front matter material after the index. Please use the Bookmarks and Contents at a Glance links to access them. Contents at a Glance About the Authors ���������������������������������������������������������������������������������������������������������������xv Acknowledgments ������������������������������������������������������������������������������������������������������������xvii Preface ������������������������������������������������������������������������������������������������������������������������������xix Introduction �����������������������������������������������������������������������������������������������������������������������xxi ■ Chapter 1: Understanding Cloud Computing ���������������������������������������������������������������������1 ■ Chapter 2: Cloud Stakeholders and Value Chain ���������������������������������������������������������������9 ■ Chapter 3: Technology that Drives the Cloud �������������������������������������������������������������������15 ■ Chapter 4: Introduction to Capacity Management �����������������������������������������������������������19 ■ Chapter 5: Cloud Capacity Management �������������������������������������������������������������������������35 ■ Chapter 6: Capacity Planning ������������������������������������������������������������������������������������������55 ■ Chapter 7: Determining Capacity Requirements for New Services ���������������������������������61 ■ Chapter 8: Capacity Management Design������������������������������������������������������������������������71 ■ Chapter 9: Capacity Plan �������������������������������������������������������������������������������������������������79 ■ Chapter 10: Ongoing Capacity Management for Existing or Live Services����������������������87 ■ Chapter 11: Capacity Monitoring �������������������������������������������������������������������������������������99 ■ Chapter 12: Capacity Analysis, Tuning, and Review ������������������������������������������������������109 ■ Chapter 13: Capacity Management Database and Integrations ������������������������������������123 v ■ Contents at a GlanCe ■ Chapter 14: Capacity Management Concepts ����������������������������������������������������������������131 ■ Appendix A: Capacity Plan Template �����������������������������������������������������������������������������137 ■ Appendix B: Capacity Implementation Case Study ��������������������������������������������������������145 Index ���������������������������������������������������������������������������������������������������������������������������������157 vi Introduction We started writing this book in 2011. As a part of a cloud solution and product development team, we came across repetitive questions while building cloud solutions. How will capacity be planned and managed by cloud providers? Is infinite capacity possible? We also witnessed chaotic discussions and fierce debates on how service providers will manage resource capacity in a cloud environment at all levels, be it business, service, or component. Is it really possible to have infinite capacity in a cloud environment, and if yes, then how? Every time these basic questions popped up, everybody had an opinion—but no two were the same. We started with our introspection and attempted to answer those questions considering ourselves as service providers. In fact, while writing this, two of our team members were talking about cloud scalability and designed a solution around it for an RFP response. Lunch conversations at the company café with server automation and solution architects, statisticians, process consultants, service owners gave rise to diverse views whenever cloud capacity and scalability were discussed. Since the beginning, cloud computing has created gigabytes of questions in everyone’s mind, be it technology guys or business people. At the very least people were fascinated by the name. So let’s start with a basic definition: cloud computing refers to the usage and access of computer resources over the Internet or a digital network. Technology and business people we encounter everyday are clueless about the workings and techniques behind private and public cloud capacity management, and there is a lot of white space in this area. Capacity management is and will continue to be the basis for the lion’s share of benefits and praise bestowed upon the art of cloud computing presently. All companies providing cloud infrastructure services have to seek an approach on the basis of which they can without any hesitation offer optimized solutions. This is not an area where the suppliers are saying “no.” Resource utilization and planning for capacity are the key areas that sit at the heart of cloud capacity management. Capacity needs, associated with a service, must be addressed at all layers (i.e., business capacity, service capacity, and resource/component capacity). It’s important to have a clear understanding of how these layers are dependent on each other from top to bottom and how these behave when there are changes in service demand, service levels, and other business process changes. There are various types of clouds: public cloud, community cloud, private cloud, etc. The latest analyst reports suggest that the cloud market will continue to grow exponentially. Worldwide companies are investing in cloud services or strategically intend to do so in coming years. This book primarily deals with capacity planning for enterprises interested in setting up a private cloud. Capacity management procedures must address capacity-related needs for both new services and ongoing services. Capacity management for new services must seek inputs from demand data to determine capacity requirements in terms of datacenter space, resource, and performance requirements (i.e., network, storage, compute, etc.); service levels also must be known in order to set up performance benchmarks. If disaster recovery requirements are necessary, capacity has to be tuned accordingly. Similar tuning has to be done to handle seasonality and other demand regulatory measures that are required by the demand management process. On the basis of capacity needs, capacity design and planning are required to take care of all capacity-related needs for running the business. In contrast, capacity management for current services is primarily an iterative process of regulating and tuning capacity along with performance monitoring. Monitoring tools play a major role here in order to keep an eye on resource usage and other performance-related meters. xxi ■ IntroduCtIon What This Book Offers This book first helps readers understand what cloud computing is and the stakeholders involved in delivering value in the cloud value chain. Cloud service delivery is explained through service- and layer-based views of the cloud. Capacity management processes are explained to help readers understand their relevance and importance in delivering IT services in a cloud computing environment. Then capacity management is explained in business, service, and component terms. This book will also help readers understand what capacity management means for the cloud service creator, cloud service aggregator, and the cloud service consumer. As we know, the cloud environment includes a high level of abstraction and virtualization to facilitate rapid and on-demand provisioning of services in a pay-as-you-go cost model. This differs from conventional IT service models, which utilize a traditional approach when planning for service capacity in order to provide optimum services levels. This book offers a helpful blend of IT service management (ITSM) best practices and on-the-ground technical implementation of these practices in the various cloud scenarios for infrastructure capacity planning and optimization activities. The book also educates readers concerning an integrated scenario of how ITSM best practices for capacity planning get addressed in the cloud environment. It also addresses cloud computing basics, cloud computing models, the impact of cloud computing on capacity planning, traditional versus cloud capacity, and capacity management process implementations in cloud environments showcasing toolset capabilities and techniques for capacity planning and performance management. Capacity management from the cloud service provider’s view has been segregated into capacity management for new services and ongoing capacity management of live services. • Procedures are explained for capacity planning of new services to ensure cost-justifiable capacity. These include procedures like understanding capacity requirements, documenting and designing capacity techniques, and producing a capacity plan for new services. • For live services, techniques for infrastructure performance monitoring and optimization are explained to ensure that agreed upon performance levels and appropriate capacity is provided. These include procedures like implementing capacity design, analysis, tuning, and capacity monitoring. Tools and techniques are explained to ensure implementation of • Best practices for capacity planning. • Best practices for ongoing performance monitoring. Techniques like dynamic resource scheduling, scaling, load balancing, and clustering are explained for implementing capacity management; these are aligned with capacity management procedures using best practices. This book also covers emerging techniques in capacity management like self learning systems, yield management, and proactive capacity planning to make readers aware of the latest in the capacity management area. Additionally, emphasis is given on how capacity models like business capacity, service capacity, and component capacity are collectively influenced by service demand, varying performance needs, and SLAs. Before moving ahead, it’s important to understand how a cloud service delivery model is formed. Chapter 1 defines cloud computing and the various deployment models. xxii Chapter 1 Understanding Cloud Computing This chapter covers the basic concepts of cloud computing, cloud technology, and its ingredients. Before diving deeply into capacity management, it’s important to have a clear understanding of the technology definitions. Cloud computing characteristics, deployment, and service models have been simplified in order to set the context for subsequent chapters. Besides cloud basics, this chapter discusses the impact of cloud computing on enterprises. This chapter also makes readers understand how the role of enterprise IT changes when cloud solutions are considered. Under the purview of cloud computing, the traditional IT landscape seeks transformation in order to support business applications efficiently and effectively. Cloud Computing Cloud computing is a buzzword these days. CIOs and key decision makers are seriously considering aligning IT with the cloud—the same cloud many of them discarded as a worthless idea in the year 2008. It is predicted that CIOs will continue to increase investments in cloud computing. In Gartner’s latest quarterly IT spending report, the research firm for the first time broke out cloud computing as a separate forecast category, providing an in-depth analysis of current and future cloud spending trends. The hottest growth in the cloud market in the coming year will be in Infrastructure as a Service (IaaS), which is expected to grow by 41%. Management and security is the second-hottest cloud growth area, expected to rise 27.2%, with platform as a service (26.6%), SaaS (17.4%), and business process as a service (15%) rounding out the top five. “The cloud market is growing at a pretty rapid clip,” Anderson says. “Cloud services within the broader IT spending market are still small, but the growth rate looks promising.” Source: Gartner Quarterly IT spending Report 2012. All technology providers are publicizing their cloud computing capabilities, or at least their strategies for the cloud. Cloud computing hype is inundating the IT world like no other hype before. In fact, the impact of cloud computing has been realized way beyond the critical interpretations. Virtualization technology, which is at the heart of cloud computing, appeared magically, right on cue, when it could be put to excellent use for optimized data center operations. Revolution is the word that truly interprets the impact of cloud computing on the way technology is being offered. As defined by the National Institute of Technology and Standards, cloud computing is a model for enabling convenient, on-demand network access to a shared pool of configurable computing resources (such as networks, servers, storage, applications, and services) that can be rapidly provisioned and released with minimal management effort or service provider interaction. Such a cloud model promotes availability and is composed of five essential characteristics, three service models, and four deployment models. The primary benefit of cloud computing solutions is a lower total cost of ownership (TCO) that results from the more efficient utilization of resources through resource pooling and the leveraging of technology. For example, through the use of virtualization technology, several servers may be consolidated into just one physical server, resulting in reduced cost and enhanced support capabilities via centralized services management. Figure 1-1 depicts a conventional on-premises IT model of a typical enterprise with the IT back bone supporting enterprise functions at all levels (i.e., back office, middle office, and front office). 1 Chapter 1 ■ Understanding CloUd CompUting Figure 1-1. On-premise enterprise IT On-premise IT, if compared with cloud-based IT, generally takes the back seat considering factors like cost, functionality, speed of deployment. Moreover, possibility of errors by IT personnel and lack of standardization have historically caused more worries to CIOs than with large cloud providers today. With the new age of cloud computing, IT is providing business agility, resilience, and profits to the enterprise. Cloud Characteristics There are several variations of the cloud definition and the characteristics associated with cloud computing. We will cover the basics and make it easy for the readers of this book to understand the essential characteristics and why they are important. On-Demand Availability of Services This characteristic is essential in a cloud service In traditional IT environments, there is no on-demand availability of services; typically one has to go through a long procurement process to get an application or an IT service. As an example of Infrastructure as a Service (Iaas), the cloud provides compute, network, storage, and security services on demand. As a result, users that require infrastructure services can order them and get them at the click of a button, rather than wait for a lengthy procurement process to complete before the capacity of a particular service is delivered. In other words, cloud providers have systems that can provide compute and other resources on demand to the users without any hassles. 2 Chapter 1 ■ Understanding CloUd CompUting Standardization and automation in a cloud computing environment enable agility and on-demand provisioning of resources. In fact, one can use a credit card to purchase compute capacity and have a virtual instance running in the cloud in a matter of minutes. Thus, cloud computing enables immediate access to resources when demanded by the users or applications. Network Access The cloud services are available on the Internet and can be accessed through multiple modes of connectivity including dedicated connections. The connectivity is based on open standards. Thus, cloud computing—just like the Internet—crosses national boundaries and provides services to the world. Pooling of Resources The cloud provider creates a pool of compute capacity that is offered to multiple customers or tenants. The cloud provider makes an upfront investment in creating a cloud service and has ready capacity to offer to customers. The pooled resources allow multiple customers to leverage the service and use the shared infrastructure for their requirements. The consumer may not have exact knowledge of the location of the service being offered, though they may choose the continent, country, or approximate location from where services are offered. The cloud provider has automations and dynamic placement engines in place that ensure that the capacity is shared across customers and that they get sufficient resources to run their applications. The cloud provider thus acts as an owner of the cloud computing resources that are rented out to multiple customers, who use it in a shared model. The concept is similar to the way a taxi or a bus service works, wherein a bus or taxi is used by multiple tenants but it is owned by someone else. The cloud provider ensures security and confidentiality of customer information and has systems in place to ensure one customer’s resources or data is not accessed by the other customers hosted on the same physical resources. Thus, network bandwidth, compute cycles, memory, and storage are resources in IaaS that are offered as a shared pool of resources to customers. Elasticity The cloud provider builds the service with scalability in mind. As a result, all aspects of the cloud service should be elastic and scalable. As the usage of an application grows, the customers can order more capacity from the cloud provider in an automated fashion. Applications that require a rapid scale up or down in capacity are ideally suited to cloud environments since the cloud has vast amounts of capacity available on demand. However, it does not mean that the cloud has infinite capacity. The elasticity of the cloud is much higher than a traditional IT environment of an enterprise as the cloud is architected to be scalable and the cloud provider keeps enough capacity to meet the needs of its customers. Also, since it is a shared infrastructure, the peaks and troughs of usage by multiple tenants, particularly over wide geographic ranges, ensure that the infrastructure is used more optimally than a dedicated one. Pay Per Use The cloud providers provide metering and billing so that the customers can be billed on a pay-per-use model. The customers use the capacity and are billed for the usage. This is analogous to the way telecom companies bill their customers by the number and duration of calls made; at the end of the month, a bill is generated detailing the calls, duration, and the cost of each item. The cloud operates in a similar way where transparent billing is available based on various types of plans and customers can pay using various methods including a credit card. 3 Chapter 1 ■ Understanding CloUd CompUting Shared Management Since the cloud provider offers a standardized mass market service, there are aspects of the service that are self-managed. The basic monitoring, provisioning, replication, and availability of service are managed by the cloud provider using advanced technologies. The economies of scale and automation are realized in the cloud world since now it is possible for the cloud provider to provide these advanced technologies for self-managed infrastructure in a standardized fashion. The cloud provider provides basic management features and functions through automated means and the layers above are to be managed by the customer. As an example, the IaaS provider provides the virtualization platform and shared network and storage as a managed service, but the operating system and the applications run on this infrastructure by the customer have to be managed by the customer. Thus, the management of the core cloud platform is done by the provider and the other components are managed by the customers themselves. The cloud provides cost savings and agility to enterprises and consumers so that they can focus on their core business and enjoy the benefits of a highly available and scalable service. Next, we’ll define and explain the service models in cloud computing. Figure 1-2 shows the various models and their characteristics. Figure 1-2. Service models for cloud computing: IaaS, PaaS, and SaaS Service Models Cloud computing service models are layers of services that the cloud provider can provide to customers. It is essential to understand the basic models in which cloud services are delivered. Different models cater to different kinds of requirements and can achieve different business objectives. Depending on the offering and needs of the customer, they can consume pure compute resources on the cloud (which is called Infrastructure as a Service), or consume a platform on which applications can be written and deployed (which is called Platform as a Service), or leverage full-blown applications or services like e-mail from the cloud (which is termed Software as a Service). 4