Introduction
Three models of cloud computing, the forum has been introduced many times, see: three models of cloud computing: public cloud, private cloud and hybrid cloud , and three types of cloud computing services (SaaS, PaaS, IaaS) . When the user selects a different cloud computing deployment model, you should understand the differences between these different modes:
In the past period of time, the use of private cloud business has grown significantly. This is because the private cloud can offer some of the advantages of the public cloud, but to address the potential disadvantages of deploying public clouds. However, the introduction of private cloud is a very complex project, may require large-scale investments in technology, processes and technical staff. This article will explore several important steps in the implementation of a private cloud, and analyzes the impact of cloud network deployment.
More information
Cloud implementation phase :
Important embodiments of cloud four stages, as follows:
Integration:
Organizations through a variety of virtualization technology to consolidate its infrastructure to dramatically improve network efficiency. Virtualization is the first step in the migration to cloud computing institution, because it can create a seamless logical storage resources and increase the use of technology assets. Virtualization separates applications and hardware infrastructure, resulting in higher service flexibility. Virtualization properly implemented, it will be able to re-allocate resources more efficiently, improve flexibility and reduce operating costs. Currently, server virtualization is the most well known type of virtualization, users should also be considered in addition virtualized storage systems, applications, network infrastructure and endpoints, such as desktop computers, which will help to integrate the work.
Optimization :
Integrated IT infrastructure can increase efficiency by increasing asset utilization. However, integration must be optimized storage, network and service management, to support virtualized infrastructure, and achieve the same common needs. There are reports that the network infrastructure needs through three key changes to help data center network to better implement virtualization and cloud computing.
Protocol enhancements: First, companies must find new protocol, such as TRILL (Transparent Interconnection of Lots of Links) to replace STP (Spanning Tree Protocol). The widespread use of TRILL achieve multipath second layer, to improve the availability of bandwidth.
重新设计架构：现有的数据中心大部分都是基于三层体系架构，涵盖了接入、汇聚及核心层。这种架构在虚拟环境中的运行效率低下，无法优化服务器到服务器和服务器到存储所产生的流量。网络用量不断流过不同层次，这更造成了延迟，也影响了实时应用的性能。近年来，交换机的性能已显著改善，这包括了通过迁移到第二层体系结构，来消除汇聚层。此外，消除多层次也能减少数据中心网络所需的交换机和电缆的数量，从而降低运营成本以及资本消耗。
采用开放式架构：随着数据中心网络的异质性上升，客户应依据其行业标准，选择采用开放式的解决方案。这项投资在未来将能证明其正确性。
网络优化通过使用以太网架构，确保网络以最高的效率运行。它创建了一个更弹性、更灵活的网络基础设施，该设施可提供无缝的可扩展性以支持动态业务需求，并允许管理员维持商定的服务水平协议（SLA）。
自动化：
企业必须将工作负载自动化，才能真正实现云技术的承诺–即业务敏捷性和按需重新配置资源。自动化可将大量繁复及耗时的手工流程转换到无缝的工作流程，从而减少重复过程。下面的图表概述了云自动化的各个阶段：
网络自动化是整个云自动化的重要组成部分。它必须支持按需重新配置资源和工作负载自动化。在选择网络自动化基础设施时，应考虑的关键因素包括：虚拟机感知零接触式虚拟化支持、最低配置的网络基础设施、执行自动化策略、预先定义的配置模板和服务配置。
管理：
最重要的一点，云服务供应商和企业内部的IT团队必须维持私有云的高可用性和服务水平。由于核心基础设施资源（计算、存储和网络功能）的虚拟化性质有不稳定的状态，这导致私有云的监控系统相当复杂。
另外重要的一点是，在业务运营被影响之前，通过积极主动的监控和预警管理来减少潜在的性能瓶颈。早期预警系统可以加快故障排除，并允许企业进一步改善和微调监控系统。尽管监控是管理方面很重要的一点，客户在考虑实现网络进化过程中，也应考虑系统是否有能力隔离、分析和报告流量模式以简化网络运营。
参考
Brocade技术白皮书
《Think Flat with Ethernet Fabric - Importance of a Flat Network Architecture in Cloud Implementation》
应用于
云计算