### Chapter 28 Clouds and Grids 云和网格 > We cannot enter into alliances until we are acquainted with the designs of our neighbors.—Sun Tzu > 在我们熟悉邻国的计划之前，我们无法结成联盟。——孙子 At this point in the book, we need to discuss one of perhaps the most importantadvances in application scaling that we have seen since hardware started to becomecommoditized. This most important advancement is the advent of cloud computing.Although most people think of this as a very new technology innovation, the reality isthat this has taken well over a decade to become a reality. In this chapter, we aregoing to cover the history that led up to the launch of cloud computing, provide anoverview of both cloud and grid computing, discuss the common characteristics ofclouds, and finish the chapter with a comparison of grid and cloud computing. 在本书的这一点上，我们需要讨论自硬件开始商品化以来我们所看到的应用程序扩展方面可能最重要的进步之一。最重要的进步是云计算的出现。尽管大多数人认为这是一项非常新的技术创新，但现实是这花了十多年的时间才成为现实。在本章中，我们将介绍云计算推出的历史，概述云和网格计算，讨论云的共同特征，并通过网格和云计算的比较来结束本章。 Cloud computing is important to scalability because it offers the promise of cheap,on-demand storage and compute capacity. As we will discuss in this chapter, this hasmany advantages, and a few disadvantages, to physical hardware scaling. Grid com-puting, similar to cloud computing, although utilized differently, offers a method ofscaling for computationally intensive applications. To be well versed in scaling appli-cations, you must understand these concepts and how they could be implemented toscale an application or service. 云计算对于可扩展性非常重要，因为它提供了廉价、按需存储和计算能力的承诺。正如我们将在本章中讨论的，这对于物理硬件扩展有许多优点，但也有一些缺点。网格计算与云计算类似，尽管用途不同，但它为计算密集型应用程序提供了一种扩展方法。要精通扩展应用程序，您必须了解这些概念以及如何实施它们来扩展应用程序或服务。 Although the development of the technology and concepts for cloud computinghave been in process for many years, the discussion and utilization of it in main-stream technology organizations is still relatively new. Because of this, at times, evendefinitions of the subject are not always completely agreed upon. We have been for-tunate to be around the cloud environment through our clients for quite some timeand have seen many players become involved in this field. As with our discussion ontechnology agnostic designs in Chapter 20, Designing for Any Technology, we believethat it is the architecture and not the technology that is responsible for a product’scapability to scale. As such, we consider clouds and grids as architectural compo-nents and not as a technology. The particular vendor or type of service is the equiva-lent of a type of technology chosen to implement the architecture. We will cover someof the technology components so that you have examples and can become familiarwith them, but we will focus primarily on the architecture. 尽管云计算的技术和概念已经发展多年，但主流技术组织对云计算的讨论和利用仍然相对较新。因此，有时甚至连主题的定义也并不总是完全一致。我们有幸通过我们的客户涉足云环境已有相当长一段时间，并看到许多参与者涉足这一领域。正如我们在第 20 章“为任何技术进行设计”中对技术无关设计的讨论一样，我们认为是架构而不是技术决定了产品的扩展能力。因此，我们将云和网格视为架构组件，而不是技术。特定的供应商或服务类型相当于为实现架构而选择的技术类型。我们将介绍一些技术组件，以便您有示例并可以熟悉它们，但我们将主要关注架构。 ####History and Definitions 历史和定义 The term cloud has been around for decades. No one is exactly sure when it was firstused in relation to technology, but it has been around at least as far back as when net-work diagrams came into vogue. A network diagram is a graphic representation of thephysical or logic layout of a network, such as a telecommunications, routing, or neural.The cloud on network diagrams was used to represent unspecified networks. In theearly 1990s, cloud became a term for ATM networks. Asynchronous Transfer Mode isa packet switching protocol that breaks data into cells and provides OSI layer 2, thedata link. ATM was the core protocol used on the public switched phone network.As the World Wide Web began in 1991 as a CERN project built on top of the Internet,the cloud began to be used as a term and symbol for the underlying infrastructure. 云这个术语已经存在了几十年。没有人确切知道它是什么时候第一次与技术相关的，但它至少可以追溯到网络图流行时。网络图是网络物理或逻辑布局的图形表示，例如电信、路由或神经网络。网络图上的云用于表示未指定的网络。 20 世纪 90 年代初，云成为 ATM 网络的术语。异步传输模式是一种数据包交换协议，它将数据分解为单元并提供 OSI 第 2 层（数据链路）。 ATM 是公共交换电话网络上使用的核心协议。随着万维网于 1991 年作为建立在互联网之上的 CERN 项目开始，云开始被用作底层基础设施的术语和符号。 #####OSI Model 开放系统互连模型 The Open Systems Interconnection Reference Model, or OSI Model, is a descriptive abstrac-tion of the layered model of network architecture. It is used to describe the different compo-nents of the network and how they are interrelated. There are seven layers, and starting fromthe lowest layer, they are as follows: 开放系统互连参考模型（OSI 模型）是网络体系结构分层模型的描述性抽象。它用于描述网络的不同组件以及它们如何相互关联。共有七层，从最底层开始，依次如下 1.Physical. This layer contains the physical devices such as cards, cables, hubs, and repeaters. 1.身体。该层包含物理设备，例如卡、电缆、集线器和中继器。 2.Data Link. This layer is responsible for the functional transmission of data between devices and includes protocols such as Ethernet. 2.数据链路。该层负责设备之间数据的功能传输，包括以太网等协议。 3.Network. This layer provides the switching and routing functionality and includes proto-cols such as Internet Protocol (IP). 3.网络。该层提供交换和路由功能，并包括互联网协议 (IP) 等协议。 4.Transport. This layer provides the reliability by keeping track of transmission, resending if necessary. It includes the Transmission Control Protocol (TCP). 4.运输。该层通过跟踪传输并在必要时重新发送来提供可靠性。它包括传输控制协议（TCP）。 5.Session. This layer controls the communication by establishing, managing, and terminat-ing connections between computers such as with sockets. 5.会议。该层通过建立、管理和终止计算机之间的连接（例如使用套接字）来控制通信。 6.Presentation. This layer provides data presentation and encryption such as with Secure Sockets Layer (SSL). 6.演示。该层提供数据表示和加密，例如使用安全套接字层 (SSL)。 7.Application. This layer interacts between the software application and the network and includes implementation such as Hyper Text Transfer Protocol (HTTP). 7.申请。该层在软件应用程序和网络之间进行交互，并包括超文本传输协议（HTTP）等实现。 Cloud computing can trace its lineage through the application service providers(ASP) of the 1990s, which were the embodiment of the concept of outsourcing com-puter services. This later became known as Software as a Service (SaaS). The ASPmodel is an indirect descendant of the service bureaus of the 1960s and 1970s, whichwere an attempt at fulfilling the vision established by John McCarthy in his 1961 speechat MIT. 云计算的历史可以追溯到20世纪90年代的应用服务提供商(ASP)，它们是外包计算机服务概念的体现。这后来被称为软件即服务 (SaaS)。 ASP 模型是 20 世纪 60 年代和 1970 年代服务局的间接后代，这些局是为了实现约翰?#40614;卡锡 (John McCarthy) 在 1961 年麻省理工学院演讲中确立的愿景的一次尝试。 1 John McCarthy is the inventor of the programming language Lisp, recipientof the 1971 Turing Award, and credited with coining the term Artificial Intelligence. 1 John McCarthy 是编程语言 Lisp 的发明者，1971 年图灵奖获得者，并因创造人工智能一词而受到赞誉。 2The idea of the modern cloud concept was extended in October 2001 by IBM inits Autonomic Computing Manifesto.3 The essence of the paper was that the infor-mation technology infrastructure was becoming too complex and that it could col-lapse under its own weight if the management was not automated. Around this time,the concept of Software as a Service (SaaS) started to grow. SaaS is a software modelwhereby people pay for software based on usage and demand rather than upfrontand recurring license fees. Another confluent event occurred around this time at thebeginning of the 21st century, and this was the dot com bubble. As many tech startupswere burning through capital and shutting down, those that were ultimately going tosurvive and thrive were tightening their belts on capital and operational expendi-tures. Amazon.com was one such company that began modernizing its data centersusing early concepts of virtualization over massive amounts of commodity hardware.Having lots of unused capacity most of the time, to deal with peak usage, Amazondecided to sell this as a service. 2现代云概念的思想在 2001 年 10 月由 IBM 的《自主计算宣言》中得到扩展。3 该论文的本质是，信息技术基础设施变得过于复杂，如果管理没有自动化。大约在这个时候，软件即服务（SaaS）的概念开始发展。 SaaS 是一种软件模型，人们根据使用情况和需求来支付软件费用，而不是预先支付和经常性的许可费用。 21世纪初的这个时候，另一个融合事件发生了，这就是互联网泡沫。随着许多科技初创公司耗尽资金并倒闭，那些最终能够生存和发展的公司却在资本和运营支出方面勒紧裤腰带。 Amazon.com 就是这样一家公司，它开始在大量商用硬件上使用虚拟化的早期概念对其数据中心进行现代化改造。由于大多数时候都有大量未使用的容量，为了应对高峰使用，亚马逊决定将其作为服务出售。 4Out of the offering of spare capacity as a service came the concept and label ofInfrastructure as a Service (IaaS). This term started to appear around 2006 and typi-cally refers to offerings of computer infrastructure such as servers, storage, networks,and bandwidth as a service instead of by subscription or contract. This method was apay-as-you-use model for what previously required either capital expenditure to pur-chase outright, long-term leases, or month-to-month subscriptions for partial tenancyof physical hardware. 4 基础设施即服务 (IaaS) 的概念和标签源自于提供备用容量即服务。该术语于 2006 年左右开始出现，通常指以服务形式提供计算机基础设施，例如服务器、存储、网络和带宽，而不是通过订阅或合同形式提供。这种方法是按使用量付费的模式，以前需要资本支出来购买直接的长期租赁，或者按月订阅部分租赁物理硬件。 #####SaaS, PaaS, IaaS, and EaaS SaaS、PaaS、IaaS 和 EaaS All these concepts of Blah as a Service have common characteristics. Among these are payingfor what you use instead of buying it up front, scaling the amount you need without prior notice,and multiple tenancy or having many different people use the same service. Blah 即服务的所有这些概念都有共同的特征。其中包括按使用量付费而不是预先购买、在不事先通知的情况下扩展您所需的数量、以及多重租赁或让许多不同的人使用相同的服务。 * Software as a Service (SaaS). This was the original Blah as a Service term and started with software such as customer relationship management (CRM) software as some of the earliest offerings. Almost any form of software can be offered in this manner and it can be done either over the Web or via download. * 软件即服务 (SaaS)。这是最初的“Blah 即服务”术语，始于客户关系管理 (CRM) 软件等最早的产品。几乎任何形式的软件都可以通过这种方式提供，并且可以通过网络或通过下载来完成。 * Platform as a Service (PaaS). This model provides all the required components for devel-oping and deploying Web applications and services. These components include workflow management, integrated development environments, testing, deployment, and hosting. * 平台即服务 (PaaS)。该模型提供了开发和部署 Web 应用程序和服务所需的所有组件。这些组件包括工作流管理、集成开发环境、测试、部署和托管。 * Infrastructure as a Service (IaaS). This is the concept of offering computing infrastructure such as servers, storage, network, and bandwidth for use as necessary by clients. Ama-zon’s EC2 was one of the earliest offerings of this service. * 基础设施即服务 (IaaS)。这是提供计算基础设施（例如服务器、存储、网络和带宽）以供客户端根据需要使用的概念。亚马逊的 EC2 是最早提供此服务的产品之一。 * Everything as a Service (XaaS or *aaS). This is the idea of being able to retrieve on demand small components or modules of software that can be pieced together to provide a new Web based application or service. Components could include retail, payments, search, security, and communications. * 一切即服务（XaaS 或 *aaS）。这个想法是能够按需检索软件的小组件或模块，这些组件或模块可以拼凑在一起以提供新的基于 Web 的应用程序或服务。组件可能包括零售、支付、搜索、安全和通信。 As these concepts evolve, they will continue to refine their definitions, and subcategoriesare sure to develop. 随着这些概念的发展，它们的定义将不断完善，子类别也必将不断发展。 From Infrastructure as a Service, we have seen an explosion of Blah as a Serviceofferings. (Blah meaning feel free to fill in the blank with almost any word you can imag-ine.) We even have Everything as a Service (EaaS) now. All of these terms actually doshare some common characteristics such as a purchasing model of pay as you go orpay as you use it, on demand scalability of the amount that you use, and the conceptthat there will be many people or multiple tenants using the service simultaneously. 从基础设施即服务中，我们看到了 Blah 即服务产品的爆炸式增长。 （Blah 的意思是随意用你能想象到的任何词来填空。）我们现在甚至拥有一切即服务（EaaS）。所有这些术语实际上都有一些共同的特征，例如即用即付的购买模式、使用量的按需可扩展性以及将有许多人或多个租户同时使用该服务的概念。 #####Grid Computing 网格计算 Grid computing as a concept has been around for almost two decades. It is used todescribe the use of two or more computers processing individual parts of an overalltask. The tasks that are most well structured for these types of solutions are ones thatare computationally intensive and divisible into smaller tasks. Ian Foster and CarlKesselman are credited as the fathers of the grid from their book The Grid: Blueprintfor a New Computing Infrastructure published in 1998 by Morgan Kaufmann Pub-lishers.5 These two individuals along with others also were instrumental in develop-ing Globus Toolkit. It is an open source product by Globus Alliance and is consideredthe de facto standard for building grids. 网格计算作为一个概念已经存在了近二十年。它用于描述使用两台或多台计算机处理整个任务的各个部分。对于这些类型的解决方案而言，结构最好的任务是计算密集型且可分为较小任务的任务。 Ian Foster 和 CarlKesselman 在 Morgan Kaufmann Publishers 于 1998 年出版的《网格：新计算基础设施蓝图》一书中被誉为网格之父。5 这两个人以及其他人在开发 Globus Toolkit 方面也发挥了重要作用。它是 Globus Alliance 的开源产品，被认为是构建网格的事实上的标准。 For tasks and jobs to be performed in a grid environment, software must be usedto orchestrate the division of labor, monitor the computation of subtasks, and aggre-5.As declared in the April 2004 issue of the University of Chicago Magazine. http://maga-zine.uchicago.edu/0404/features/index.shtml. 对于要在网格环境中执行的任务和工作，必须使用软件来编排劳动分工、监视子任务的计算以及 aggre-5。正如芝加哥大学杂志 2004 年 4 月号中所声明的那样。 http://maga-zine.uchicago.edu/0404/features/index.shtml。 gation of completed jobs. This type of processing can be thought of as parallel pro-cessing that occurs on a network distributed basis. Before the concept of gridcomputing, the only way to achieve this scale of parallel processing was on a main-frame computer system. Modern grids are often composed of many thousands ofnodes across public or private networks of computers. 已完成工作的控制。这种类型的处理可以被认为是在网络分布式基础上发生的并行处理。在网格计算概念出现之前，实现这种规模的并行处理的唯一方法是在大型计算机系统上。现代网格通常由公共或专用计算机网络中的数千个节点组成。 Public networks range from volunteers on the Internet allowing the use of theircomputers’ unused CPU clock cycles to pay for usage grids, such as the Sun Grid that waslaunched in February 2005.Private networks include dedicated farms of small com-modity servers used with grid middleware to allow for parallel computing. Other privatenetworks include corporate offices where personal computers are used after hours forparallel computing. One of the most well known public network examples of gridcomputing is the SETI@home project. This project uses computers connected to theInternet in the Search for Extraterrestrial Intelligence (SETI). Individuals can participateby running a program on their personal computers that downloads and analyzes radiotelescope data. It then sends the results back to the central system that aggregates com-pleted jobs. This type of public network grid computing is known as CPU scavenging. 公共网络包括互联网上的志愿者，允许他们使用计算机未使用的 CPU 时钟周期来支付网格使用费用，例如 2005 年 2 月推出的 Sun Grid。私有网络包括与网格中间件一起使用的小型商品服务器的专用场。以允许并行计算。其他专用网络包括公司办公室，下班后使用个人计算机进行并行计算。 SETI@home 项目是网格计算最著名的公共网络示例之一。该项目在搜寻外星智慧生物（SETI）中使用连接到互联网的计算机。个人可以通过在个人计算机上运行下载和分析射电望远镜数据的程序来参与。然后，它将结果发送回聚合已完成作业的中央系统。这种类型的公共网络网格计算称为 CPU 清理。 As we mentioned previously, there are several middleware providers for buildinggrid systems. One of the earliest is the Globus Toolkit, but others include the SunGrid Engine and UNICORE (UNiform Interface to COmputing REsources). Many ofthese middleware products are used in university, governmental, and industrial appli-cations to create grids. 正如我们之前提到的，有几个用于构建网格系统的中间件提供商。最早的工具包之一是 Globus Toolkit，但其他工具包包括 SunGrid Engine 和 UNICORE（计算资源统一接口）。许多中间件产品在大学、政府和工业应用中用于创建网格。 With all this processing power, how fast can our applications run? Similar in con-cept to Brooks’ Law from the book The Mythical Man-Month by Frederick P.Brooks, Jr., applications can only be divided so many times. Although adding moreprocessors to the job probably won’t make it actually slow down, as projects dowhen adding engineers late within a project, it does nothing to accomplish the jobfaster. There is a law know as Amdahl’s Law developed by Gene Amdahl in 1967that states that the portion of a program that cannot be parallelized will limit thetotal speed up from parallelization. 有了如此强大的处理能力，我们的应用程序可以运行多快？其概念与 Frederick P.Brooks, Jr. 所著《神话人月》一书中的布鲁克斯定律类似，应用程序只能被划分多次。尽管向工作中添加更多处理器可能不会真正减慢工作速度，就像在项目后期添加工程师时所做的那样，但它对于更快地完成工作没有任何帮助。 Gene Amdahl 在 1967 年提出了一条称为“Amdahl 定律”的定律，该定律指出，程序中无法并行化的部分将限制并行化的总速度。 6 Stated another way, the nonsequential parts of aprogram benefit from the parallelization, but the rest of the program does not. Thereactually was no formula provided in the original paper, but it can be approximated,as shown in Figure 28.1, by the total improvement, a factor of the original run time,equaling the inverse of the sequential parts of the program (s). 6 换句话说，程序的非顺序部分受益于并行化，但程序的其余部分则不然。实际上，原始论文中没有提供公式，但可以通过总改进（原始运行时间的一个因素）来近似，如图 28.1 所示，等于程序的顺序部分的倒数。  If 75% of the program can be parallelized, leaving 25% that must be executedsequentially, we should see approximately a 4u increase in the total runtime of theapplication by running it in a parallel computing environment. 如果 75% 的程序可以并行化，剩下 25% 必须顺序执行，那么通过在并行计算环境中运行应用程序，我们应该会看到应用程序的总运行时间增加了大约 4u。 As you might expect, there are problems associated with running applications andprograms on grid systems. This is especially true on public networks, but can also beproblematic on private ones. One of the biggest problems is how to handle the datathat is required by the program to execute. Storage management of data, security ofthe data, and transmission of the data are all areas of concern when dealing with net-work distributed parallel processing or grid computing. Usually, the amount of datathat is sent along with the executable program is small enough that by itself it is nei-ther valuable nor sensitive. Nevertheless, the perceived risk of sending data to com-puter systems that you do not control can be a tough sell, especially when processingsensitive information such as Personally Identifiable Information (PII). 正如您所料，在网格系统上运行应用程序和程序会出现一些问题。这在公共网络上尤其如此，但在私人网络上也可能出现问题。最大的问题之一是如何处理程序执行所需的数据。在处理网络分布式并行处理或网格计算时，数据的存储管理、数据的安全性和数据的传输都是需要关注的领域。通常，与可执行程序一起发送的数据量足够小，其本身既不有价值也不敏感。然而，将数据发送到您无法控制的计算机系统的风险可能很难接受，尤其是在处理个人身份信息 (PII) 等敏感信息时。 #####Public Versus Private Clouds 公共云与私有云 Some of the largest names in technology are providing or have plans to provide cloudcomputing services. These companies include Amazon.com Inc., Google Inc.,Hewlett-Packard Company, and Microsoft Corporation. These are publicly availableclouds, of which anyone from individuals to other corporations can take advantage.However, if you are interested in running your application in a cloud environmentbut have concerns about a public cloud, there is the possibility of running a privatecloud. By private cloud, we mean implementing a cloud on your own hardware inyour own secure environment. With more open source cloud solutions becomingavailable, such as Eucalyptus, this is becoming a realistic solution. There are obvi-ously pros and cons with running your application in a cloud environment. We aregoing to talk more about these in detail in Chapter 29, Soaring in the Clouds. Someof the benefits and drawbacks are present regardless of whether it is a private or pub-lic cloud. Some of the drawbacks are directly related to the fact that it is a publiccloud. Similar to the problem with data security in grid computing, even if the publiccloud is very secure, there is a perception that the data is not as protected as it wouldbe inside of your network. One of the pros of a cloud is that you can allocate just theright amount of memory, CPU, and disk space to a particular application taking bet-ter advantage and improving utilization of the hardware. So, if you want to improveyour hardware utilization and not have to deal with security concerns of a publiccloud, you may want to consider running your own private cloud. 一些科技领域的巨头正在提供或计划提供云计算服务。这些公司包括亚马逊公司、谷歌公司、惠普公司和微软公司。这些是公开可用的云，从个人到其他公司的任何人都可以利用。但是，如果您有兴趣在云环境中运行应用程序但对公共云有顾虑，则可以运行私有云。私有云是指在您自己的安全环境中在您自己的硬件上实施云。随着更多开源云解决方案（例如 Eucalyptus）的出现，这正在成为一个现实的解决方案。在云环境中运行应用程序显然有利有弊。我们将在第 29 章“翱翔云端”中详细讨论这些内容。无论是私有云还是公共云，都存在一些优点和缺点。一些缺点与它是公共云这一事实直接相关。与网格计算中的数据安全问题类似，即使公共云非常安全，人们仍然认为数据不像网络内部那样受到保护。云的优点之一是您可以为特定应用程序分配适量的内存、CPU 和磁盘空间，从而更好地利用并提高硬件利用率。因此，如果您想提高硬件利用率并且不必处理公共云的安全问题，您可能需要考虑运行自己的私有云。 ####Characteristics and Architecture of Clouds 云的特性和架构 At this point in the evolution of the cloud computing concept, there are some basiccharacteristics that all cloud implementations share. These characteristics have beenmentioned briefly before in this chapter, but it is time to understand them in moredetail. There are four general characteristics that we find in almost all public cloudimplementations, some of which do not apply to private clouds. These characteristicsare pay by usage, scale on demand, multiple tenants, and virtualization. Obviously,scaling on demand is an important characteristic when viewing the use of cloudsfrom a scalability perspective, but don’t dismiss the other characteristics as unimpor-tant. For cash strapped startup companies paying as you go instead of purchasinghardware or signing multiyear contracts, this could be the difference between surviv-ing long enough to be successful or not. 在云计算概念的演变过程中，所有云实现都具有一些共同的基本特征。本章之前已简要提及这些特征，但现在是更详细地了解它们的时候了。我们在几乎所有公共云实施中发现了四个一般特征，其中一些不适用于私有云。这些特征是按使用付费、按需扩展、多租户和虚拟化。显然，从可扩展性的角度来看云的使用时，按需扩展是一个重要特征，但不要认为其他特征不重要而忽视。对于现金短缺的初创公司来说，按使用量付费而不是购买硬件或签署多年合同，这可能是生存足够长的时间以取得成功与否的区别。 #####Pay By Usage 按用量付费 The idea of pay as you go or pay according to your usage is common of Software as aService (SaaS) and has been adopted by the cloud computing services. Before cloudcomputing was available, in order to grow your application and have enough capac-ity to scale, you had limited options. If you were a large enough organization, youprobably had servers that you owned or leased that were hosted in a data center orcollocation facility. This model requires lots of upfront capital expenditure as well asa pretty healthy monthly expense to continue paying bandwidth, power, space, andcooling costs. An alternative was a hosting service that provided the hardware foryou, and you paid them on either a long-term lease or high monthly cost for the useof the hardware. Both models are reasonable and have benefits as well as drawbacks.Many companies still use one or both of these models and will likely do so for manyyears to come. The cloud offers another alternative. Instead of long-term leases orhigh upfront capital outlays, the cloud model allows you to have no upfront cost forpurchasing hardware and pay by the utilization of either CPU, bandwidth, storage,or possibly all of them. 即用即付或按使用量付费的理念在软件即服务 (SaaS) 中很常见，并且已被云计算服务所采用。在云计算可用之前，为了扩展您的应用程序并拥有足够的扩展能力，您的选择有限。如果您是一个足够大的组织，您可能拥有或租用托管在数据中心或托管设施中的服务器。这种模式需要大量的前期资本支出以及相当健康的每月支出来继续支付带宽、电力、空间和冷却成本。另一种选择是为您提供硬件的托管服务，您可以通过长期租赁或每月支付高昂的硬件使用费用来向他们支付费用。这两种模型都是合理的，各有利弊。许多公司仍在使用其中一种或两种模型，并且可能会在未来很多年中这样做。云提供了另一种选择。云模型无需长期租赁或高额前期资本支出，而是允许您无需购买硬件的前期成本，并通过使用 CPU、带宽、存储或可能全部使用来付费。 #####Scale On Demand 按需扩展 Another characteristic of cloud computing is the ability to scale on demand. As a sub-scriber or client of a cloud, you have the theoretical ability of scaling as much as youneed. This implies that if you need terabytes of storage or gigahertz or more, computingthese would be available to you. There are of course practical limits to this that includehow much actual capacity the cloud provider has to offer, but with the larger clouds,it is reasonable that you can scale to the equivalent of several hundreds or thousandsof servers with no issues on most public clouds. Of course in a private cloud, thisbecomes your organization’s limitations on physical hardware. The time that it takesto do this is almost instant compared to the standard method of provisioning hard-ware in a data center. Let’s look at the typical process first, as if you were hostingyour site at a collocation facility and then as if you were running in a cloud environment. 云计算的另一个特点是能够按需扩展。作为云的订阅者或客户端，您理论上具有根据需要进行扩展的能力。这意味着，如果您需要 TB 的存储空间或千兆赫或更多的存储空间，您就可以进行计算。当然，这存在实际限制，包括云提供商必须提供多少实际容量，但对于更大的云，您可以合理地扩展到相当于数百或数千台服务器，而在大多数公共云上都没有问题。当然，在私有云中，这成为您的组织对物理硬件的限制。与在数据中心配置硬件的标准方法相比，执行此操作所需的时间几乎是即时的。让我们首先看一下典型的过程，就好像您在托管设施中托管站点，然后就好像您在云环境中运行一样。 Adding hundreds of servers in a collocation facility or data center can take days,weeks, or months, depending on an organization’s processes. For those who havenever worked in an organization that hosted with a collocation facility, this is a typi-cal scenario that you might encounter. Most organizations have budgeting andrequest processes regardless of where or how the site is hosted. After the budget orthe purchase order is approved, the process of provisioning a new server in a cloudand one in a collocation facility are almost completely different. For a collocationfacility, you need to ensure that you have available the space and power to accommo-date the new servers. This can entail going to a new cage in a collocation provider ifyou do not have any more space or power available in your current cage. If a newcage is required, contracts must be negotiated and signed for the lease of the newspace and cross connects are generally required to connect the cage’s networks. Afterthe space and power that are necessary are secured, purchase orders for the serverscan be placed. Of course, some companies will stockpile servers in anticipation ofcapacity demand. Others wait until the operations team alerts that capacity is at apoint that expanding the server pools is required. If your organization waits, order-ing and receiving the hardware can take weeks. After the hardware arrives at the col-location facility, it needs to be placed in the racks and powered up. After this isaccomplished, the operations team can get started ghosting, jumpstarting, or kickstarting the server depending on the operating system. And when all this is finallyaccomplished, the latest version of the software can be loaded and the server can beadded into the production pool. The total time for this process is at least days in themost efficient operations teams who already have hardware and space available. Inmost organizations, this process takes weeks or months. 在托管设施或数据中心添加数百台服务器可能需要几天、几周或几个月的时间，具体取决于组织的流程。对于那些从未在托管过托管设施的组织中工作过的人来说，这是您可能会遇到的典型场景。大多数组织都有预算和请求流程，无论站点在何处或如何托管。预算或采购订单获得批准后，在云中配置新服务器和在托管设施中配置新服务器的过程几乎完全不同。对于配置设施，您需要确保有足够的空间和电力来容纳新服务器。如果您当前的笼子中没有更多可用空间或电力，这可能需要转到配置提供商的新笼子。如果需要新的笼子，必须协商并签署新空间的租赁合同，并且通常需要交叉连接来连接笼子的网络。确保所需的空间和电源后，即可下达服务器采购订单。当然，一些公司会根据容量需求而储备服务器。其他人则等待，直到运营团队警告容量已达到需要扩展服务器池的程度。如果您的组织需要等待，订购和接收硬件可能需要数周时间。硬件到达并置设施后，需要将其放置在机架中并通电。完成此操作后，运营团队可以根据操作系统开始重影、快速启动或快速启动服务器。当这一切最终完成后，就可以加载最新版本的软件，并将服务器添加到生产池中。对于已经拥有可用硬件和空间的最高效的运营团队来说，此过程的总时间至少需要几天。在大多数组织中，此过程需要数周或数月的时间。 Let’s take a look at how this process might go if you were hosting your site in acloud environment and decided that you needed 10 or 20 more servers for a particu-lar pool. The process would start off similarly with the budget or purchase orderrequest to add to the monthly expense of the cloud services. After this was approved,the operations team would use the control panel of the cloud provider to simplyrequest the number of virtual servers, specifying the size and speed. Within a fewminutes, the systems would be available to load the machine image of choice and thelatest application code could be installed. The servers could likely be placed into pro-duction within a few hours. This ability to scale on demand is a common characteris-tic of cloud computing. 让我们看一下如果您在云环境中托管站点并决定为特定池需要 10 或 20 多个服务器，那么此过程可能会如何进行。该流程将以类似的预算或采购订单请求开始，以添加到云服务的每月费用中。获得批准后，运营团队将使用云提供商的控制面板来简单地请求虚拟服务器的数量，指定大小和速度。几分钟之内，系统就可以加载所选的机器映像，并可以安装最新的应用程序代码。这些服务器可能会在几个小时内投入生产。这种按需扩展的能力是云计算的共同特征。 #####Multiple Tenants 多个租户 Although the ability to scale on demand is enticing, all that capacity is not just wait-ing for you. Public clouds have many users running all sorts of applications on themand the capacity is being shared. This is known as multitenanting or having multipletenants existing on the same cloud. If all works as designed, these users never interactor impact each other. Data is not shared, access is not shared, and accounts are notshared. Each client has its own virtual environment that is walled off from other vir-tual environments. What you do share with other tenants in a cloud is the physicalservers, network, and storage. You might have a virtual dual processor server with32GB of RAM, but it is likely running on an eight processor with 128GB of RAMthat is being shared with several other tenants. Your traffic between servers and fromthe servers to the storage goes across common networking gear. There are no routers,switches, or firewalls dedicated to individual tenants. The same goes for the storage.Tenants share storage on virtual network-attached storage (NAS) or storage area net-work (SAN) devices making it appear as if they are the only ones using it. The realityis that there are multiple tenants using that same physical storage device. 尽管按需扩展的能力很诱人，但所有这些容量并不仅仅在等着您。公共云有许多用户在其上运行各种应用程序，并且容量是共享的。这称为多租户或在同一云上存在多个租户。如果一切按设计进行，这些用户永远不会相互影响。数据不共享，访问权限不共享，帐户不共享。每个客户端都有自己的虚拟环境，与其他虚拟环境隔离。您与云中的其他租户共享的是物理服务器、网络和存储。您可能有一个具有 32GB RAM 的虚拟双处理器服务器，但它可能运行在具有 128GB RAM 的八处理器上，并与其他几个租户共享。服务器之间以及从服务器到存储的流量通过常见的网络设备。没有专用于各个租户的路由器、交换机或防火墙。存储也是如此。租户在虚拟网络附加存储 (NAS) 或存储区域网络 (SAN) 设备上共享存储，使其看起来好像只有他们在使用该存储。现实情况是，有多个租户使用同一个物理存储设备。 #####Virtualization 虚拟化 All cloud computing offerings implement some form of a hypervisor on the serversthat provides the virtualization. This concept of virtualization is the fourth commoncloud characteristic and is really the core architectural principle behind clouds. Ahypervisor is either a hardware platform or software service that allows multipleoperating systems to run on a single host server. This is also known as a virtualmachine monitor (VMM). There are many vendors that offer hardware and softwaresolutions, such as VMware, Parallels, and Oracle VM. As we covered in the multi-tenant discussion, this virtualization is what allows multiple users to exist on com-mon hardware without knowing and hopefully without impacting each other. Thereare other virtualization, separation, or limitation techniques that are used to restrictaccess of cloud clients to only amounts of bandwidth and storage that they have pur-chased. The overall purpose of all of these techniques is to control access and provideas much as possible an environment that appears to be completely the client’s own.The better this is done the less likely that clients will notice each other’s presence. 所有云计算产品都在提供虚拟化的服务器上实现某种形式的虚拟机管理程序。虚拟化概念是公共云的第四个特征，也是云背后真正的核心架构原则。管理程序是一个硬件平台或软件服务，允许多个操作系统在单个主机服务器上运行。这也称为虚拟机监视器 (VMM)。有许多供应商提供硬件和软件解决方案，例如 VMware、Parallels 和 Oracle VM。正如我们在多租户讨论中所讨论的，这种虚拟化允许多个用户在不知情的情况下存在于通用硬件上，并且希望不会互相影响。还有其他虚拟化、分离或限制技术，用于将云客户端的访问仅限于他们购买的带宽和存储量。所有这些技术的总体目的是控制访问并尽可能提供一个完全属于客户自己的环境。做得越好，客户注意到彼此存在的可能性就越小。 #####Common Cloud Characteristics 常见的云特征 There are four basic common characteristics of cloud computing services: 云计算服务有四个基本共同特征 1.Pay as You Go. Users, subscribers, or clients only pay for the amount of bandwidth, stor-age, and processing that they consume. 1.随用随付。用户、订户或客户端只需为他们消耗的带宽、存储和处理量付费。 2.Scale on Demand. Cloud clients have access to theoretically unlimited capacity for scal-ing their applications by adding more bandwidth, servers, or storage. 2.按需扩展。云客户端理论上可以通过添加更多带宽、服务器或存储来扩展其应用程序，从而获得理论上无限的容量。 3.Multiple Tenants. Clouds service multiple, often many thousands, of clients. Clients share hardware, networks, bandwidth, and storage. Physical devices are not dedicated to clients. 3.多个租户。云为多个（通常是数千个）客户提供服务。客户端共享硬件、网络、带宽和存储。物理设备并非专用于客户端。 4.Virtualization. The way that multitenanting can be accomplished is through virtualization. This is the process by which hardware can have multiple operating systems running on it simultaneously. 4.虚拟化。实现多租户的方式是通过虚拟化。这是硬件可以同时运行多个操作系统的过程。 Some of these characteristics may or may not be present in private clouds. Let’stake a quick look at private cloud implementations. The concept of virtualization isat the core of all clouds regardless of whether it is public or private. The idea of usingfarms of physical servers in different virtual forms to achieve greater utilization, mul-titenancy, or any other various reasons is the basic premise behind the architecture ofa cloud. The ability to scale as necessary is likely to be a common characteristicregardless of whether it is a private or public cloud. There is a physical restriction onthe amount of scale that can occur. This physical restriction is dependent on howlarge the cloud is and how much extra capacity is built into it. 其中一些特征可能存在于私有云中，也可能不存在。让我们快速浏览一下私有云的实施。虚拟化的概念是所有云的核心，无论是公共云还是私有云。使用不同虚拟形式的物理服务器群来实现更高的利用率、多租户或任何其他各种原因的想法是云架构背后的基本前提。无论是私有云还是公共云，根据需要进行扩展的能力可能是一个共同特征。可能发生的水垢数量存在物理限制。这种物理限制取决于云有多大以及云中内置了多少额外容量。 Having multiple tenants is not required in a private cloud. Tenants in a privatecloud can mean different departments or different applications, not necessarily andprobably not different companies. Pay as you go is also not required but possible inprivate clouds. Depending on how cost centers or departments are charged in anorganization, each department might have to pay based on usage of the privatecloud. If there is a centralized operations team that is responsible for building andrunning its own profit and loss center, it may very well charge departments or divi-sions for the services provided. This can include the computational, bandwidth, andstorage in a private cloud. 私有云中不需要拥有多个租户。私有云中的租户可能意味着不同的部门或不同的应用程序，不一定也可能不是不同的公司。即用即付也不是必需的，但在私有云中是可能的。根据组织中成本中心或部门的收费方式，每个部门可能必须根据私有云的使用情况付费。如果有一个集中的运营团队负责建立和运营自己的损益中心，它很可能会向部门或部门提供的服务收费。这可以包括私有云中的计算、带宽和存储。 ####Differences Between Clouds and Grids 云和网格之间的差异 Now that we’ve covered some of the history and basic characteristics of clouds andgrids, we are going to conclude this chapter with a comparison of the two concepts.The terms cloud and grid are often confused and misused. We are going to cover afew of the differences and some of the similarities between the two to ensure that weare all clear on when each should be considered for use in our systems. 现在我们已经介绍了云和网格的一些历史和基本特征，我们将通过比较这两个概念来结束本章。云和网格这两个术语经常被混淆和误用。我们将介绍两者之间的一些差异和一些相似之处，以确保我们清楚何时应考虑在我们的系统中使用它们。 Clouds and grids serve different purposes. Clouds offer virtual environments forhosting user applications on one or many virtual servers. This makes clouds particu-larly compelling for applications that have unpredictable usage demands. When youare not sure if you need 5 or 50 servers over the next three months, a cloud can be anideal solution. We will cover more pros and cons in Chapter 30, Plugging in the Grid,as well as how to decide if a cloud is really the correct hosting solution for yourapplications. Clouds, again because of this virtualization architecture, allow users toshare the infrastructure. There can be many different users on the same physicalhardware consuming and sharing computational, network, and storage resources. 云和网格有不同的用途。云提供了在一台或多台虚拟服务器上托管用户应用程序的虚拟环境。这使得云对于具有不可预测的使用需求的应用程序特别有吸引力。当您不确定未来三个月内是否需要 5 台服务器还是 50 台服务器时，云可能是理想的解决方案。我们将在第 30 章“插入网格”中介绍更多优点和缺点，以及如何确定云是否确实是适合您的应用程序的正确托管解决方案。再次由于这种虚拟化架构，云允许用户共享基础设施。同一物理硬件上可能有许多不同的用户消耗和共享计算、网络和存储资源。 On the other hand, grids are infrastructures for dividing programs into small partsto be executed in parallel across two or more hosts. These environments are ideal foran application that needs computationally intensive environments. Grids are not nec-essarily great infrastructures to share with multiple tenants. You are likely running ona grid to parallelize and significantly increase the computational bandwidth for yourapplication; sharing the infrastructure simultaneously defeats that purpose. Sharingor multitenancy can occur serially, one after the other, in a grid environment whereeach application runs in isolation; when completed, the next job runs. This challengeof enabling multitenancy on grids is one of the core jobs of a grid operations team.Grids are also ideal only for applications that can be divisible into elements that canbe simultaneously executed. The throughput of a monolithic application cannot behelped by running on a grid. The same monolithic application can likely be replicatedonto many individual servers in a cloud, and the throughput can be scaled by thenumber of servers added. Stated as simply as we can, clouds allow you to expand andcontract your architecture; grids decompose work into parallelizable units. 另一方面，网格是将程序划分为小部分以便在两个或多个主机上并行执行的基础设施。这些环境非常适合需要计算密集型环境的应用程序。网格不一定是与多个租户共享的优秀基础设施。您可能在网格上运行以并行化并显着增加应用程序的计算带宽；同时共享基础设施违背了这一目的。共享或多租户可以在每个应用程序独立运行的网格环境中连续发生。完成后，下一个作业将运行。在网格上实现多租户的这一挑战是网格运营团队的核心工作之一。网格也仅适用于可分为可同时执行的元素的应用程序。在网格上运行并不能提高单体应用程序的吞吐量。相同的整体应用程序可能会复制到云中的许多单独服务器上，并且吞吐量可以根据添加的服务器数量进行扩展。简而言之，云允许您扩展和收缩您的架构；网格将工作分解为可并行的单元。 While serving different purposes, there are many crossovers and similarities. Thefirst major one is that some clouds run on top of a grid infrastructure. A good exampleof this is AppLogic from 3Tera, which is a grid operating system that is offered as soft-ware but is also used to power a cloud that is offered as a service. Other similaritiesbetween clouds and grids include on demand pricing models and scalable usage. If youneed 50 extra servers in a cloud, you can get them allocated quickly and you pay forthem for the time that you are using them. The same is the case in a grid environment.If you need 50 more nodes for improving the processing time of the application, youcan have this allocated rather quickly and you pay for only the nodes that you use.At this point, you should understand that clouds and grids are fundamentally dif-ferent concepts and serve different purposes but have similarities and share commoncharacteristics, and are sometimes intertwined in implementations. One last topic tofurther analyze these different implementations is the three different types of clouds. 虽然服务于不同的目的，但存在许多交叉和相似之处。第一个主要问题是一些云运行在网格基础设施之上。 3Tera 的 AppLogic 就是一个很好的例子，它是一个网格操作系统，作为软件提供，但也用于为作为服务提供的云提供支持。云和网格之间的其他相似之处包括按需定价模型和可扩展的使用。如果您在云中需要 50 台额外的服务器，您可以快速分配它们，并根据使用时间付费。网格环境中的情况也是如此。如果您需要 50 个以上的节点来提高应用程序的处理时间，您可以相当快地分配这些节点，并且只需为您使用的节点付费。此时，您应该了解这一点云和网格是根本不同的概念，服务于不同的目的，但具有相似之处并具有共同的特征，并且有时在实现中交织在一起。进一步分析这些不同实现的最后一个主题是三种不同类型的云。 #####Types of Clouds 云的类型 There really is no standard categorization for clouds. We like to use three categoriesto explain the differences between various providers and vendors. These three are ser-vice providers, backbones, and virtualization software. Of course, because there is nostandardization, many of the vendors cross between these categories. 云确实没有标准的分类。我们喜欢使用三个类别来解释不同提供商和供应商之间的差异。这三者是服务提供商、骨干网和虚拟化软件。当然，由于没有标准化，许多供应商在这些类别之间进行交叉。 Starting from the lowest level, the first category is the virtualization software pro-viders. This is software that allows for the hypervisor or the ability to run multipleoperating systems on a single server. There are many vendors that offer software inthis category. One such software is VMWare. VMWare is used to split physical serv-ers into multiple virtual servers that can be allocated memory, CPU cycles, and diskas if it were a completely separate physical server. Another virtualization software isAppLogic that is offered by 3Tera. AppLogic, as mentioned, is a grid operating sys-tem that allows a cloud service to be provided on top of the grid. 从最底层开始，第一类是虚拟化软件提供商。该软件允许虚拟机管理程序或在单个服务器上运行多个操作系统的能力。有许多供应商提供此类别的软件。 VMWare 就是这样的软件之一。 VMWare 用于将物理服务器拆分为多个虚拟服务器，这些虚拟服务器可以分配内存、CPU 周期和磁盘，就好像它是完全独立的物理服务器一样。另一种虚拟化软件是3Tera 提供的AppLogic。如前所述，AppLogic 是一个网格操作系统，允许在网格之上提供云服务。 The next category of clouds is the backbone providers. These are vendors that pro-vide basic virtualization services, typically on a very large scale. We consider vendors’offerings such as Amazon’s EC2, Rackspace’s cloud, Google Apps, and Microsoft’scloud to be in this category. These vendors utilize some form, either off-the-shelf, cus-tomized, or mostly home grown, virtualization software to provide their clouds. 下一类云是骨干提供商。这些供应商通常提供非常大规模的基本虚拟化服务。我们认为 Amazon 的 EC2、Rackspace 的云、Google Apps 和 Microsoft 的云等供应商的产品都属于此类。这些供应商利用某种形式的虚拟化软件（无论是现成的、定制的还是大部分是自制的）来提供云。 The last category of clouds is the service provider. Companies that fall in this cate-gory offer services on top of backbones. One such vendor is RightScale, who offersan improved management interface for Amazon’s EC2 and other services. Anothervendor is Mosso, who offers a service on top of Rackspace’s cloud. Both of these ser-vices have attempted to improve the deployment and management of the backboneofferings. 最后一类云是服务提供商。属于这一类的公司在骨干网之上提供服务。 RightScale 就是这样的供应商之一，它为亚马逊的 EC2 和其他服务提供了改进的管理界面。另一个供应商是 Mosso，它在 Rackspace 的云之上提供服务。这两项服务都试图改进主干产品的部署和管理。 Many vendors could make an argument that they belong in more than one of theseenvironments. Although a service provider might have a different management inter-face for a backbone, this doesn’t mean that the backbone provider itself does not alsohave a GUI management interface that is fully functional. Even though this categori-zation is often blurred in the real world, it does offer a basic framework to helpunderstand the multitude of vendors that have some sort of cloud related offering. 许多供应商可能会声称他们属于多个环境。尽管服务提供商可能有不同的主干管理界面，但这并不意味着主干提供商本身没有功能齐全的 GUI 管理界面。尽管这种分类在现实世界中常常很模糊，但它确实提供了一个基本框架来帮助了解拥有某种云相关产品的众多供应商。 ####Conclusion 结论 In this chapter, we started by covering the history that led up to the launch of cloudcomputing and included the background of grid computing. This history dates backseveral decades, and the concept of the modern-day cloud can be credited to IBM’smanifesto. However, the evolution of cloud computing into what it is today has beenmade possible by many different people and companies including one of the firstpublic cloud services, EC2.We covered many of the “as a Service” offerings includ-ing Software as a Service and Infrastructure as a Service. 在本章中，我们首先介绍了云计算推出的历史，并包括网格计算的背景。这段历史可以追溯到几十年前，现代云的概念可以归功于 IBM 的宣言。然而，云计算发展到今天的样子是由许多不同的人和公司实现的，其中包括第一个公共云服务 EC2。我们介绍了许多“即服务”产品，包括软件即服务和基础设施即服务。 In the history section, we discussed how the concept of grid computing had beenaround for almost two decades. Grid computing is used to describe the use of two ormore computers processing individual parts of an overall task. Tasks that are most wellsuited for a grid solution are ones that require intensive computations and those thatare divisible. The maximum amount of improvement that can be made by dividing anapplication up is limited by the amount of sequential processing that is required. 在历史部分，我们讨论了网格计算的概念是如何存在了近二十年的。网格计算用于描述使用两台或更多计算机处理整个任务的各个部分。最适合网格解决方案的任务是需要大量计算和可分割的任务。通过划分应用程序可以实现的最大改进量受到所需的顺序处理量的限制。 There are four common characteristics of public clouds. These are pay as you go,scale on demand, multiple tenants, and virtualization. We covered each of these indetail and also looked at what characteristics private clouds are likely to have. Weconcluded that virtualization and scale on demand are two characteristics that pri-vate clouds must have. Hosting multiple tenants and pay as you go are possible inprivate clouds but not necessary. 公共云有四个共同特征。这些是即用即付、按需扩展、多租户和虚拟化。我们详细介绍了这些内容，并研究了私有云可能具有的特征。我们的结论是，虚拟化和按需扩展是私有云必须具备的两个特征。在私有云中可以托管多个租户并按需付费，但这不是必需的。 We concluded the chapter with a comparison of grid and cloud computing. Westarted by stating that clouds and grids serve different purposes. Clouds offer virtualenvironments for hosting user applications on one or many virtual servers. Grids, onthe other hand, are infrastructures for dividing programs into small parts to be exe-cuted in parallel across two or more hosts. Clouds are compelling for applicationsthat have unpredictable usage demands, need to scale quickly, and can scale on inde-pendent servers. Grids are more suited for applications that need computationallyintensive environments. 我们通过比较网格和云计算来结束本章。我们首先指出云和网格有不同的用途。云提供了在一台或多台虚拟服务器上托管用户应用程序的虚拟环境。另一方面，网格是将程序划分为小部分以便在两个或更多主机上并行执行的基础设施。对于具有不可预测的使用需求、需要快速扩展并且可以在独立服务器上扩展的应用程序来说，云非常有吸引力。网格更适合需要计算密集型环境的应用程序。 The last topic covered under the comparison of grids and clouds was the categori-zation of cloud providers. We described a three-category hierarchy that started withvirtualization software providers, then moved up to backbone providers, and finallyended with service providers. We conceded that although this categorization helpsunderstand all the various offerings, it is excessively restrictive and many vendorscross over between these categories. 网格和云比较中涉及的最后一个主题是云提供商的分类。我们描述了一个三类层次结构，从虚拟化软件提供商开始，然后向上移动到骨干提供商，最后以服务提供商结束。我们承认，虽然这种分类有助于理解所有不同的产品，但它的限制性太大，并且许多供应商在这些类别之间进行交叉。 As we’ve pointed out repeatedly in this chapter, clouds and grids are importanttechnologies and concepts to scalability. With a thorough understanding of theircharacteristics and uses, they can become effective weapons in your arsenal to fightthe scalability war. 正如我们在本章中反复指出的那样，云和网格是可扩展性的重要技术和概念。彻底了解它们的特性和用途后，它们可以成为您的武器库中对抗可扩展性战争的有效武器。 #####Key Points 关键点 *The term cloud has been around for decades and was used primarily in networkdiagrams. * 云这个术语已经存在了几十年，主要用于网络图中。 * The idea of the modern cloud concept was put forth by IBM in its AutonomicComputing Manifesto. * 现代云概念的想法是由 IBM 在其 AutonomicComputing Manifesto 中提出的。 * Developing alongside the idea of cloud computing was the concept of Softwareas a Service, Infrastructure as a Service, and many more “as a Service” concepts. * Software as a Service refers to almost any form of software that is offered in apay as you use model. * 与云计算理念一起发展的是软件即服务、基础设施即服务以及更多“即服务”概念。 * 软件即服务是指以即付即用模式提供的几乎任何形式的软件。 * Infrastructure as a Service is the idea of offering infrastructure such as storage,servers, network, and bandwidth in a pay as you use model. * 基础设施即服务是指以按使用量付费的模式提供存储、服务器、网络和带宽等基础设施。 * Platform as a Service provides all the required components for developing anddeploying Web applications and services. * 平台即服务提供了开发和部署 Web 应用程序和服务所需的所有组件。 * Everything as a Service is the idea of being able to have small components thatcan be pieced together to provide a new service. * 一切即服务的理念是能够将小型组件组装在一起以提供新服务。 * Grid computing as a concept has been around for almost two decades. It is usedto describe the use of two or more computers processing individual parts of anoverall task. * 网格计算作为一个概念已经存在了近二十年。它用于描述使用两台或多台计算机处理整体任务的各个部分。 * There are three types of cloud vendors: service providers, backbones, and virtu-alization software providers. * 云供应商分为三种类型：服务提供商、骨干网和虚拟化软件提供商。
只有一条评论 (QwQ)
你的文章让我感受到了生活的美好，谢谢！ https://www.yonboz.com/video/88671.html