I recently finished reading “SDN: Software Defined Networks” and recommend it. The editing is not the best (“veru” instead of “very,” “Mangement,” etc.), but the breadth and depth of content far outweigh the minor blemishes.
“SDN: Software Defined Networks” published in August 2013 and is up-to-date as of the time of this writing, including discussions of Open Daylight, ETSI, OpenFlow 1.3 (though not 1.4, see http://reliablesdn.com/what-you-need-to-know-openflow-1-4/), etc.
The authors hail from Juniper and that background shows through, though not to an off-putting extent. The forewords by Brocade and Cisco CTOs also reinforce the notion that the book is from an incumbent equipment provider perspective, but again, not to an extent that significantly detracts from the value. In fact, I appreciated the incumbent viewpoint spelled out in a clear manner because their marketing messaging sometimes obscures the technical content.
The “idealized controller” architecture thread running through the book provided a helpful framework for the discussion. I might differ with the idealization (e.g., it includes a Java reference and concludes that the Cisco ONE is the best mapping of existing options to the ideal), but I appreciate the effort to plow new ground in framing the discussion.
Other aspects of the book I particularly liked include:
- Assumes minimal knowledge of networking, SDN and related fields
- P. 10 has a helpful figure and discussion on the controller continuum from centralized through fully distributed
- Chapter 3 has a good list of OpenFlow shortcomings, though not exhaustive (see http://reliablesdn.com/some-openflow-technical-issues-and-their-implications ) and without much discussion of the issues
- Chapter 4’s discussion of non-OpenFlow controllers, including virtual route reflectors, broadened my vision of a controller
- Overall, very interesting to see how incumbents are pulling their stacks apart to support SDN – even though there’s no common API
- The chapter 9 highlights of IETF efforts past and present around standardization, including I2RS (https://www.ietf.org/mailman/listinfo/i2rs), help provide context for ONF efforts
- From this reader’s perspective, the authors made good choices on where to go “wide” on a topic (e.g,. controllers) versus narrow (e.g., devops with Puppet). There are areas where they perhaps go deeper than the average reader needs (e.g., Google’s Protocol Buffers), but not a significant issue.
For a recent article on the book and interview with the authors see: http://searchsdn.techtarget.com/news/2240206451/New-SDN-book-looks-beyond-OpenFlow-and-central-control-Author-QA
The authors mention possible later editions. If/when they do revise it, here are a few suggestions of topics to add or expand on:
- VNF/application software engineering (see http://reliablesdn.com/challenges-and-opportunities-in-the-sdn-software-engineering-ecosystem/)
- More non-Juniper material
- More discussion of university research
- Impacts on mobile operators
- Survey of use cases and applications
- Business models and opportunities
I look forward to the second edition!