作者中文版序

当本书译者和审校者邀请我们为中文版写个简短的序时，我们首先想到的是要仔细回顾一下本书英文版出版以来的短短六个月中，SDN业界及相关技术领域发生的各种变化，以及取得的所有新进展。抱着这种想法，我们又看了一遍邀请邮件，发现他们只要求写“不到一页”！好吧，那就没事了，我们的序一定简明扼要。

本书英文版出版以来的短短六个月内，书中探讨的一些主题在现实中有了新的进展。

比如理想化的SDN框架这类概念，已经由OpenDaylight®项目给实例化了。越来越多的人开始抛弃专有控制器，因为他们已经意识到，专有控制器不过是另一种垄断形式而已。如今，人们已经转向公共、普适的SDN框架或基础设施，这样有助于各自的创新。

开源的日益重要和兴起不仅体现在OpenDaylight项目上，也体现在业内其他组织和人的努力上。OpenStack和OVS的快速发展及被人们所接受（部署）就是例证。

传统标准组织往往要花许多年才能让厂商的实现达成一致。开源项目则不同，它们由志愿者构成的用户社区共同维护，并基于可运行的代码创造出事实标准。这是两条完全不同的路线。认清这一点很重要，因为某些“伪标准”组织和论坛一开始也会打着“开源”的幌子，但运作方式仍然是封闭的。希望读者能够留心注意背后的区别。

我们俩现在都换了工作，我们希望自己服务的厂商能像OpenDaylight项目一样，积极遵循真正开源合作的理念（开放管理、精英主导、各尽所能、求同存异）。Thomas D. Nadeau现在博科公司领导一个大型团队，专注于构建基于OpenDaylight控制器的商业SDN/NFV应用程序，同时也非常注重对开源项目的贡献。Ken Gray现在思科公司的工作也差不多。

本书第1版涉及的其他主题同样有所进展，比如英特尔优化库（DPDK）已成为进一步扩大虚拟化网络服务与功能（及效果）的一种手段。这方面的进步不在控制平面，而在虚拟数据平面前所未有的性能提升上。某些虚拟数据平面在商用标准硬件上的性能，并不亚于在定制硬件上的性能。

同样，我们当初写作本书时刚刚进入共同探索第一年的NFV（网络功能虚拟化）也取得了进步。随着ETSI NFV研究组两年任期行将届满（并即将颁布体系结构指南），我们看到一些传统的服务提供商宣布愿意采用NFV，也看到了众多厂商宣布即将推出相关产品。

我们没有料到本书英文版获得了如此巨大的成功，我们要感谢那些购买它的读者。同时，我们也为本书能被翻译成中文版，与中文读者见面而感到兴奋。在此感谢本书中文版的译者和审校者，翻译可不是一件简单的事。

想了解我们最新的想法，请访问我们的博客：http://www.sdnprogrammability.net，或关注我们的推特@tdnjunisco和@graymatterken。

Thomas D. Nadeau

Ken Gray

2014年3月20日

英文原文

When Chao Zhou (who did the Chinese translation of our book) asked us to write a short Preface for this edition, the first thing I thought of was to delve into the details of all of the things in the SDN industry and technology space that have changed or rapidly evolved since we published this book just a short 6 months ago. Then I checked that thought when I read his email again that stated we could only have “less than a page” of space to do this in! Problem solved: we will keep this short and sweet!

In the short 6 months since this book came out, the industry has already moved forward with some of the topics we initiated in the original book.

Concepts like the Idealized SDN Framework were instantiated in the Hydrogen release of OpenDaylight®. There is now a growing shift away from proprietary controllers in the industry as providers realized they were just another means for vendor lock-in. We have movedtowards a common, ubiquitous framework or substrate on which we can focus our innovative efforts.

The growing importance and uptake of open source not just indicated by OpenDaylight-related efforts, but also by others working in this space. Examples of this also include the rapid evolution and acceptance (deployment) of Open Stack and OVS.

It seems that open source development is on a trajectory to leapfrog traditional standards organizations that have by definition taken many years to produce implementation agreements among vendors. Open source creates an avenue by which a community of interested users can create what are in effect de facto standards with accompanying running code. This observation is important in that the term “open source” is now being co-opted in the creation of other “pseudo standards” organizations and forums that work in decidedly closed ways. We hope our readers will take care to notice the differences behind the labels.

We have both also moved on to work for vendors we feel actively support the ideals of a truly open source collaboration in SDN that the OpenDaylight Project represents (open governance, meritocracy through contribution and a diverse community). Tom is now leading a large team at Brocade that is focused on building commercial SDN/NFV applications based on the ODL controller, as well as a strong focus on contributing to the open source project. For Ken's part, he is now working on similar SDN efforts at Cisco.

Other topics covered in the first edition that are making advances include the advancement of Intel's optimization library (DPDK) that others have latched onto as a means of further expanding the role (and efficacy) of virtualized network services and functions. The advances here are not in the control plane, but in the seemingly ever-increasing performance gains of virtualized data planes. At this point, there are real functional examples of virtualized data planes running on COTS hardware that rival some custom hardware.

In the same vein, Network Function Virtualization (which was just entering it's first year of concerted exploration when we wrote the book) is now moving forward. As the ETSI NFV Study Group nears the end of it's two-year charter (and publishes it's architecture guidelines), we see announcements from traditional Service Providers favoring adopting NFV and announcements of forthcoming products from numerous vendors.

Neither of us envisioned the wild success this book would have and we are both grateful to those who have purchased it already, and are also very excited to tap into an entirely different readership. We want to thank Chao (and other people's name) for embarking on what was likely not an easy process of translating our original writing.

To keep track of our latest thoughts and musings, please visit at our blog at http://www.sdnprogrammability.net or follow us on Ttwitter @tdnjunisco or @graymatterken.

Tom and Ken

March 20, 2014