Frontispiece

Provides the reader with comprehensive insight into the structural decisions that can to be made when architecting a content distribution system that uses IP-based networks

The narrative of this book draws on a wealth of real-world and practical experience that the author has accrued through two decades of coalface experience architecting and delivering large, mission critical live video, webcasts, and radio streaming online, over both the Internet and private IP networks.

From this loosely defined "tradeperson's" standpoint, rather than the often explored tightly academic or business-sales point of view, this book takes a broad, humored, and at times pencil-sucking look at the art of building content delivery workflows.

Topics Include

• Delivery of live, catch-up, scheduled, on-demand, TVOD and SVOD
• CDN topologies including edge-caching, stream-splitting, Pureplay, Operator, Satellite, and Hybrid
• Computation hosting and orchestration in models such as dedicated appliances and virtualization
• Format considerations and achieving adaptive, format resilient operator networks and backbone infrastructure
• General comments on market forces over cycles and eras of evolution of these technologies

This book aims to talk in backroom engineers' English about the challenges faced in the real world, and to stimulate the reader to think extremely broadly about the options and problem spaces, and how to ensure that delivery is always, at the least, "good enough" for the operator's and consumers' commercial objectives.

As we enter what the author calls the "third generation of CDN," architects who are new to the area can use this text to draw on the author's own practical experience over the first two generations.

The book will also be an interesting read for those who have themselves built large infrastructure, providing a moment to reflect on other ways around problems. It will be a useful quick-start tool for those who are trying to understand the complex challenges of large-scale content delivery.

Not one for hiding opinion, the author also throws a number of challenging "what if" scenarios into the discussion to highlight some possible long-term design architectures that today may be a little fantastical but tomorrow may evolve based on the clear demand that such architectures could reach, should the commercial model evolve in line.

This discussion zooms in on the recent evolution of software-defined networking and the changes that this schism will bring as capabilities for many players in the network stack become unlimited, and infrastructure allocated to a particular task can be repurposed at the flick of a bit.

About the Book

While content delivery network architecture texts typically focus on current and forthcoming best practice, few take a deep retrospective view and embrace the cycles in the sector. CDNs also typically comprise 20% of their engineering work on video despite its being 80% of their traffic overhead. The author has focused on live video and audio transmission because the problems span so many layers of the network stack. There are, of course, many application-specific challenges, with particularly gaming and conferencing and to a lesser extent dynamic website acceleration and small object or large file delivery. Some of these do cause network layer issues, but generally the traffic is not impacting to a network operator - it is impacting to the Software as a Service provider or the application user. There are many complex issues that can be explored, and many are touched on in this book; however, for the main part, the core focus of this book is on live (and to a lesser extent on-demand) video delivery - TV, radio, video, and live audio over IP networks.

Synposis

Starting in 1973, streaming audio and subsequently video have been baked into the IP protocols. With the web making the quick discovery of content near ubiquitous, the demand for not only huge volumes of text but also for web apps, and significantly for high-quality video, has exploded.

The likes of the BBC, YouTube, Netflix, and countless other online publishers, have lit up the information highway with literally inconceivable amounts of information conveyed in huge quantities of bytes. Those data have to be delivered to destinations by someone, and the dark art these people practice is called content delivery networking.

Over the past 20 years we have seen several trends emerge, and these exist at both the micro level, where we are encoding pixels of video into a streaming format, and the macro level, where millions of users are able to consume content from hundreds of thousands of servers, reliably and with a great deal of resilience.

Trends in GPUs are changing how encoding resources are deployed. Evolutions in distributed computing are bringing about a macro change in the architecture of these types of services.

This evolution promises greater service levels, more flexibility to meet the customers' exact requirements, and new security challenges as infrastructure becomes increasingly shared in multi-tenant public cloud models.

Telecoms network operators are now seeing IP services as a core part of their businesses, and their understanding of their own internal content delivery architecture requirements is a key driver for their rapid adoption of a software operating model. Soon operators will, at-will, be able to deliver the CDN as an SaaS model on their own infrastructure, and additionally offer other SaaS models in the same infrastructure, providing risk mitigation as they try to underpin services for an ever more divergent target market.

Unique Perspective

The book describes the historical context of the streaming media and content delivery market from the unique perspective of the author who is a true native to the sector. It draws heavily on personal experience and hands-on examples from 20 years of live webcast production through to public company infrastructure architecture. There are few in the industry who can boast such a rich and varied practical experience across the sector, and this unique insight is fundamental to the narrative.

Aside from the anecdotal and practical commentary, the book takes the implementer through a wide range of design considerations for different network topologies, starting with the author's own requirement filtration processes through to initial sketches, through to roles and responsibilities, and to the complexity of managing change in established teams, agile as opposed to waterfall considerations, in the context of large blue chips, security and commercial models, and value chain alignment.

This widely embracing viewpoint, supported by examples ranging from IETF discussions, regulatory considerations, policy formation, coders, hardware vendors network operators, and more, is rarely available from one author. The author draws on conversations with peers in the industry, and in the course of writing, he gathers their comments and input too.

While many books on these topic slice and dice these seemingly unrelated schools of thinking into their constituent parts of commercial, technical, operational (etc.), this book can help service designers embrace the worldview of influences that need to be considered when architecting a robust and high-quality content delivery service for today's online consumers and business users.

Market Need

Today's market is just about to fully enter what the author call its third generation.

• The first - which spanned until around 2005 - was the appliance era dedicated hardware and software
• The second - which spanned from 2005 until around 2014 - was the virtual machine era when software could be moved machine to machine
• The third - which started in 2014 - is the emerging container era characterized by software that is highly componentized and is deployed to the resource best suited to the task as the capability is required

As the SDN/NFV models stimulate understanding across the Telco sector, there is about to be a tech refresh like no other: all the hardware that has traditionally been dedicated to task is going to become software driven in entirety. The Telco operators who were about to deploy Gen2 CDNs are holding back to see how the underlying infrastructure is going to evolve, to then deploy their CDN as a gen3 model using the network's built in resources to deploy the CDN as an SaaS and when a client needs it.

That cycle is going to take a further three to five years.

As it happens, service architects are going to be planning more against customer requirement than against "productizability," and this requires a breadth of thinking at the COO / CTO level from every engineer and commercial participant too.

Designing a CDN for tomorrow is a broad challenge - and this book strives to get the reader thinking like a content delivery network designer.

Audience

Target: Streaming media readership, IP / cable/ satellite / Telco / mobile and TV operators, content producers, ISPs, policy and regulatory (net neutrality and content rights), and all stakeholders in networks that may deliver large quantities of video or audio (and data / applications too).

The book...