Advanced content delivery, streaming, and cloud services

ADVANCED CONTENT DELIVERY, STREAMING, AND CLOUD SERVICES... Library of Congress Cataloging-in-Publication Data: Advanced content delivery, streaming, and cloud services / editors, Mukadd

ADVANCED CONTENT DELIVERY, STREAMING, AND CLOUD SERVICES

ADVANCED CONTENT DELIVERY, STREAMING, AND CLOUD SERVICES

Edited by

Mukaddim PathanTelstra Corporation Ltd., Australia

Ramesh K SitaramanUniversity of Massachusetts, Amherst and

Akamai Technologies, USADom Robinsonid3as-company Ltd., UK

Copyright © 2014 by John Wiley & Sons, Inc All rights reserved.

Published by John Wiley & Sons, Inc., Hoboken, New Jersey.

Published simultaneously in Canada.

No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate per-copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, (978) 750-8400, fax (978) 750-4470, or on the web at www.copyright.com Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, (201) 748-6011, fax (201) 748-6008, or online at http://www.wiley.com/go/permission.

Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness

of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for

a particular purpose No warranty may be created or extended by sales representatives or written sales materials The advice and strategies contained herein may not be suitable for your situation You should consult with a professional where appropriate Neither the publisher nor author shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages.

For general information on our other products and services or for technical support, please contact our Customer Care Department within the United States at (800) 762-2974, outside the United States at (317) 572-3993 or fax (317) 572-4002.

Wiley also publishes its books in a variety of electronic formats Some content that appears in print may not

be available in electronic formats For more information about Wiley products, visit our web site at www.wiley.com.

Library of Congress Cataloging-in-Publication Data:

Advanced content delivery, streaming, and cloud services / editors, Mukaddim Pathan, Ramesh K.

Sitaraman, Dom Robinson.

To my wife Ziyuan for her inspiration, love, and support This book would not have been completed, if she did not single-handedly take care of everything, while I was too busy in writing and compilation!—Mukaddim

To my wife Vidya for her love and support And to our lovely children Anu and Siddu for reminding me that despite our best efforts the Internet still isn’t fast

enough for them!—Ramesh

To my wife Mariana and our wonderful kids Sofia and Zac—I am sure you will enjoy this book as a bedtime reading And to my parents (that funny “computer

thing” you bought me as a kid came in handy!)—Dom

2.2 Live Streaming Pre-Evolution 34

2.7 Platform Divergence and Codec Convergence 442.8 Adaptive Bitrate (ABR) Streaming 45

4 EFFICIENCY OF CACHING AND CONTENT DELIVERY IN

Gerhard Haßlinger

4.2 Options and Properties for Web Caching 734.3 Zipf Laws for Requests to Popular Content 754.4 Efficiency and Performance Modeling for Caches 764.5 Effect of Replacement Strategies on Cache Hit Rates 784.6 Replacement Methods Based on Request Statistics 814.7 Global CDN and P2P Overlays for Content Delivery 84

5 ANYCAST REQUEST ROUTING FOR CONTENT DELIVERY

Hussein A Alzoubi, Michael Rabinovich, Seungjoon Lee, Kobus Van Der

Merwe, and Oliver Spatscheck

CONTENTS ix

Tiago Cruz, Paulo Sim̃oes, and Edmundo Monteiro

6.2 Bringing Cloud Services to Home: State of the Art 1206.3 Virtualizing the Access Network Infrastructure 1236.4 Virtualization for Cloud Service Delivery to Home 130

Ram Lakshmi Narayanan, Yinghua Ye, Anuj Kaul, and Mili Shah

7.2 Mobile Broadband Architecture 142

7.4 Video Optimization Services 146

7.6 Cloud-Based Video Streaming 151

Benjamin Molina, Jaime Calvo, Carlos E Palau, and Manuel Esteve

12.6 Future Research Directions 256

13.2 Background and Related Work 260

13.4 From Content Delivery to Streaming 26313.5 Future Research Directions 267

15.2 Capacity Planning Process 29015.3 Undertaking the Capacity Planning Process 29515.4 CDN Capacity Planning Case Study 30015.5 Recent Developments and Challenges 302

Ramesh K Sitaraman, Mangesh Kasbekar, Woody Lichtenstein,

and Manish Jain

18 CONTENT DELIVERY IN CHINA: A ChinaCache PERSPECTIVE 353

Michael Talyansky, Alexei Tumarkin, Hunter Xu, and Ken Zhang

Mirosław Czyrnek, J ¸edrzej Jajor, Jerzy Jamro˙zy, Ewa Ku´smierek, Cezary

Mazurek, Maciej Stroi´nski, and Jan W ¸eglarz

Piotr Srebrny, Dag H.L Sørbø, Thomas Plagemann, Vera Goebel,

and Andreas Mauthe

21 CONTENT REPLICATION AND DELIVERY IN

Vasilis Sourlas, Paris Flegkas, Dimitrios Katsaros,

and Leandros Tassiulas

22 ROBUST CONTENT BROADCASTING IN VEHICULAR NETWORKS 431

Giancarlo Fortino, Carlos T Calafate, Juan C Cano, and Pietro Manzoni

The ever-evolving nature of the Internet brings new challenges in managing anddelivering content to end-users Content Delivery Networks (CDNs) improve Webaccess and streaming performance, in terms of response time and system through-put, while delivering content to Internet end-users through multiple, geographicallydistributed edge servers The CDN industry, that is, content delivery, consumption,and monetization, has been undergoing rapid changes The multidimensional surge

in content delivery from end-users has led to an explosion of new content, formats,and an exponential increase in the size and complexity of the digital content supplychain These changes have been accelerated by economic downturn in that the contentproviders are under increasing pressure to reduce costs while increasing revenue.The main value proposition for CDN services has shifted over time Initially, thefocus was on improving end-user-perceived experience by decreasing response time,especially when the customer website experiences unexpected traffic surges Nowadays,CDN services are treated by content providers as a way to use a shared infrastructure tohandle their peak capacity requirements, thus allowing reduced investment cost in theirown hosting infrastructure Moreover, recent trends in CDNs indicate a large paradigmshift toward a utility computing model, which allows customers to exploit advancedcontent delivery services, hosted on commodity hardware, without having to build adedicated infrastructure

From a market perspective, historically buyers based the bulk of demand and ing on “core” CDN products that facilitate the delivery of Web-based content services.Over the last few years, offering from video streaming and value-added services (VASs)peaked as the most demandable CDN products They formed the basis of most of thepresent-day CDNs’ offering, while strong demand for the basic CDN services still con-tinues Market research shows that on average, buyers reported 43% of total CDN spend-ing on core products, such as caching and content delivery, while 57% spending onVAS-based products

spend-While satisfying the market demands, CDN providers are more and more focusing

on higher margin, VAS offering in order to gain (or stabilize) overall profit margins.These VASs include mobile data acceleration, content protection, content management,application acceleration, mobile data delivery, and cloud-based storage While theseproducts currently have a reasonable market penetration, they represent even more sub-stantial near-term growth opportunities

In addition to the emergence of innovative CDN models, such as managed CDNs,licensed CDNs, and federated CDNs, Telco/operator CDNs are evolving into majormarket share holders Telcos/operators around the world have started building CDNplatform, technology, and support to aid content consumption, delivery, and rich mediaexperience by end-users The geographic expansion of Web-based content continues togrow and drive global CDN business requirements Many CDN players have started with

a regional focus and then expanded to offer services in new regions It is expected thatthese trends in the CDN industry will continue, as the definition and scope of a CDNgets broader

1.1 OVERVIEW AND SCOPE OF THE BOOK

The book entitled Advanced Content Delivery, Streaming, and Cloud Services presents

fundamental and trendy CDN technologies, with a comprehensive coverage of tion, current landscape, and future roadmap The book builds on academic and industrialresearch and developments, and case studies that are being carried out at different orga-nizations around the world In addition, the book identifies potential research directionsand technologies that will drive future innovations This book is aimed at a large audi-ence including systems architects, practitioners, product developers, and researchers Itcan be used as a reference/textbook for graduate students and a roadmap for academi-cians, who are starting to research in the field of content delivery We expect the readers

evolu-to have at least the basic knowledge about Web technologies and the Internet In ular, readers should be knowledgeable about Web caching, replication, Internet-basedservices and applications, and basic networking

partic-Upon reading this, book readers will perceive the following benefits:

1 Learn the state of the art in research and development on content management,delivery, and streaming technologies

2 Obtain a future roadmap by learning open research issues

3 Gather the background knowledge to tackle key problems, whose solutions willenhance the evolution of next-generation content networks

4 Use the book as a valuable reference and/or textbook

1.2 ORGANIZATION OF THE BOOK

This book is organized into three parts, namely, Part I: CDN and Media StreamingBasics; Part II: CDN Performance Management and Optimization; and Part III:Case Studies and Next-Generation CDNs Specifically, the topics of the book are thefollowing:

• CDN—Infrastructure, architecture, and technology for web content delivery,

content management services, and media streaming

PREFACE xvii

• Adaptive Bitrate Streaming (ABR) Techniques for multimedia streaming over

computer networks using the HTTP protocol

• Cloud-Based Content Delivery Integration of cloud computing with traditional

CDN model for content and Web application delivery

• Wide Area Network (WAN) Optimization Optimization algorithms to increase

data transfer efficiency in an end-to-end delivery path across WANs

• Mobile Acceleration Service Optimizing content and video streams to mobile

devices to meet dynamic and personalized content needs of mobile users

• Transparent Caching Carriers network caching technology to control over what

content to cache, when to cache, and how fast to accelerate the content delivery

• Request-Routing Techniques Known and advanced algorithms for redirecting

end-user requests, such as DNS-based routing, anycasting, and content-basedrouting

• CDN Performance, Availability, and Reliability SSL processing, network-based

personal video recorder (PVR), and measurement techniques

• Next-Generation CDNs Overview of managed/licensed CDN, Telco/carrier

CDNs, P2P CDN, and federated CDNs

• CDN Case Studies Overview of operational infrastructure and services from the

major CDNs

• CDN Business Intelligence Coverage of the CDN market trends, ongoing

plan-ning, and management

Part I of the book focuses on the basic ideas, techniques, and current practices related

to content delivery and media streaming Chapter 1 by Pathan presents an overview ofCDNs, operational models, and use cases It covers recent market and technology trends,

as well as identifies a few research issues in the CDN domain Robinson, in Chapters 2and 3, provides a comprehensive description of the live media streaming ecosystem anddemonstrates the practical configuration of live streaming using a few tools In Chapter 4,Haßlinger identifies key properties of caching and content delivery in broadband accessnetwork, and describes how efficiency can be achieved by configuration and performancetuning Alzoubi et al in Chapter 5 present mechanisms and algorithms to effectively redi-rect end-user requests in a CDN platform This chapter demonstrates the applicability

of IP anycasting for request redirection Basics of content delivery to cloud-based homeecosystem is covered in Chapter 6 by Cruz et al., highlighting key challenges, indus-try practices, and recent trends In Chapter 7, Narayanan et al describe the challenges

in delivering video in mobile networks and present various adaptation techniques formobile video streaming

Part II of the book provides a coverage of CDN performance measurement niques, tools, reporting, and analytics In Chapter 8, Siglin covers CDN analytics toolsand explores a variety of analytic practices and their implications in practical context,including new methods for analyzing adaptive bitrate (ABR) streaming technology.Mathematical modeling to optimize CDN services, such as video on demand (VoD)content delivery, is covered in Chapter 9 by Bekta¸s and Ercetin It makes the reader

tech-aware of fundamental optimization problems arising in content delivery and the ways

of effectively solving these problems Molina et al in Chapter 10 present a basicanalytical model to analyze the basic and advanced properties of a CDN Zhanikeev inChapter 11 describes a method for cloud-based multisource streaming and compares itsperformance over traditional methods In Chapter 12, Islam and Grégoire discuss on theintersection of CDN and cloud computing by exposing a number of trade-offs on thedeployment of multimedia processing functions inside the cloud and identify relevantperformance factors In Chapter 13, Yoshida describes the performance of a dynamicstreaming CDN, comprising techniques for dynamic network reorganization, and loaddistribution and balancing to realize dynamicity, as well as techniques for stream seg-mentation and reconstruction, and QoS assurance Cesario et al in Chapter 14 presentthe analysis of mining streaming data in a CDN, improving efficiency and effectiveness

of a CDN architecture A hybrid multidomain architecture is described that solvesthe problem of computing frequent items and frequent itemsets from distributed datastreams In Chapter 15, Davies and Pathan cover the capacity planning process that isinstrumental for the ongoing operation of a deployed CDN infrastructure It includes apractical application and workflow of the CDN capacity planning process

Part III, the final part of the book, consists of a handful of representative case studies

on present- and next-generation CDNs In Chapter 16, Sitaraman et al discuss ent network overlays that are crucial for meeting the needs for Internet-based services.Architecture and techniques of representative overlays are discussed, along with theirpractical usage and implications Chapter 17 by Pai provides coverage of a variety ofnext-generation CDNs and presents a case study of CoBlitz, a research CDN that became

differ-a commercidiffer-al licensed CDN In Chdiffer-apter 18, Tdiffer-alydiffer-ansky et differ-al describe the chdiffer-allenges ofcontent delivery in China, by drawing on experience from ChinaCache, a carrier-neutralCDN A brief coverage of content-aware network services offered by ChinaCache isprovided, along with future trends of content delivery within China Chapter 19, byCzyrnek et al., presents a case study of a high definition (HD) interactive TV platform,called PlatonTV In addition to describing the PlatonTV architecture, different aspects

of content delivery such as content ingest, content distribution, and management withinthe CDN are discussed In Chapter 20, Srebrny et al present CacheCast—a link layercaching system for single-source, multiple destination data transfers In this case study,CacheCast architecture, operational methodology, and deployment details are presented.Sourlas et al in Chapter 21 present a generic three-phase framework for content replica-tion in information centric networks (ICNs) Algorithms supporting efficient replication

in ICN are discussed and performance benefits are demonstrated Chapter 22 by Fortino

et al describes content delivery techniques in vehicular ad hoc networks (VANets) Acontent broadcasting methodology is presented, which improves content transfer timeand delivery efficiency in the radio network Finally, in Chapter 23, Kilanioti et al.discuss approaches to leverage information from online social networks (OSNs) for

PREFACE xix

rich media content delivery in CDNs Future research directions in this area, along with

a few commercial implications for CDNs, are also discussed

This book came into light because of the direct and indirect involvement of manyresearchers, academics, and industry practitioners We acknowledge and thank thecontributing authors; research institutions; and companies whose papers, reports,articles, notes, websites, and study materials have been referred to in this book Weare thankful to Professor Albert Zomaya, editor of the Wiley Series on Parallel andDistributed Computing, for his support in accepting the book proposal and guiding

us through Wiley’s publication process We express our gratitude to Simone Taylor,Director, Editorial Development of John Wiley & Sons, Inc., and Wiley’s publicationstaff, for handling the book project and ensuring a timely publication

All chapters were peer reviewed, and authors have updated their chapters addressingthe review comments Prior technical sources are acknowledged citing them at appro-priate places in the book In case of any errors, we would like to receive feedback so that

it could be taken into consideration in the next edition

We hope that this book will serve as a valuable text for students especially at ate level and a reference for researchers and practitioners working in the content deliverydomain

gradu-Mukaddim, Ramesh, and Dom

Hussein A Alzoubi Case Western Reserve University, Cleveland, OH, USA

Tolga Bekta¸s University of Southampton, Highfield, Southampton, UK

Carlos T Calafate Universitat Politècnica de València, Valencia, Spain

Jaime Calvo Universidad de Salamanca, Escuela Politecnica Superior de Zamora,

Zamora, Spain

Juan C Cano Universitat Politècnica de València, Valencia, Spain

Eugenio Cesario ICAR-CNR, Rende (CS), Italy

Tiago Cruz Faculdade de Ciências e Tecnologia da, Universidade de Coimbra,

Coimbra, Portugal

Mirosław Czyrnek Poznan Supercomputing and Networking Center, Pozna´n, Poland Phil Davie Telstra Corporation Limited, Melbourne, Victoria, Australia

Ozgur Ercetin Sabanc𝚤 University, ˙Istanbul, Turkey

Manuel Esteve Universitat Politecnica de Valencia, Valencia, Spain

Paris Flegkas University of Thessaly, Oktovriou, Volos, Greece

Giancarlo Fortino University of Calabria, Rende (CS), Italy

Chryssis Georgiou Department of Computer Science, University of Cyprus, Nicosia,

Cyprus

Vera Goebel University of Oslo, Oslo, Norway

Jean-Charles Grégoire INRS-EMT, Montréal, QC, Canada

Gerhard Haßlinger Deutsche Telekom Technik, Darmstadt, Germany

Salekul Islam United International University, Dhaka, Bangladesh

Manish Jain Akamai Technologies, Inc., Cambridge, MA, USA

J ¸edrzej Jajor Poznan Supercomputing and Networking Center, Pozna´n, Poland Jerzy Jamrȯzy Poznan Supercomputing and Networking Center, Pozna´n, Poland

Mangesh Kasbekar Akamai Technologies, Inc., Cambridge, MA, USA

Dimitrios Katsaros University of Thessaly, Oktovriou, Volos, Greece

Anuj Kaul Nokia Siemens Networks, Mountain View, CA, USA

Irene Kilanioti Department of Computer Science, University of Cyprus, Nicosia,

Cyprus

Ewa Ku´smierek Poznan Supercomputing and Networking Center, Pozna´n, Poland Seungjoon Lee AT&T Labs—Research, Florham Park, NJ, USA

Woody Lichtenstein Akamai Technologies, Inc., Cambridge, MA, USA

Pietro Manzoni Universitat Politècnica de València, Valencia, Spain

Carlo Mastroianni ICAR-CNR, Rende (CS), Italy

Andreas Mauthe InfoLab 21, Lancaster University, Lancaster, UK

Cezary Mazurek Poznan Supercomputing and Networking Center, Pozna´n, Poland Benjamin Molina Universitat Politecnica de Valencia, Valencia, Spain

Edmundo Monteiro Faculdade de Ciências e Tecnologia da, Universidade de Coimbra,

Coimbra, Portugal

Ram Lakshmi Narayanan Nokia Siemens Networks, Mountain View, CA, USA Vivek S Pai Princeton University, Princeton, NJ, USA

Carlos E Palau Universitat Politecnica de Valencia, Valencia, Spain

George Pallis Department of Computer Science, University of Cyprus, Nicosia, Cyprus Mukaddim Pathan Telstra Corporation Ltd., Melbourne, Victoria, Australia

Thomas Plagemann University of Oslo, Oslo, Norway

Michael Rabinovich Case Western Reserve University, Cleveland, OH, USA

Dom Robinson id3as-company Ltd., Rottingdean, Brighton, Sussex, UK

Mili Shah Nokia Siemens Networks, Mountain View, CA, USA

Timothy Siglin Braintrust Digital, Inc., Harriman, TN, USA

Paulo Simões Faculdade de Ciências e Tecnologia da, Universidade de Coimbra,

Coimbra, Portugal

Ramesh K Sitaraman University of Massachusetts, Amherst, and Akamai

Technolo-gies, MA, USA

Dag H.L Sørbø University of Oslo, Oslo, Norway

Vasilis Sourlas University of Thessaly, Oktovriou, Volos, Greece

Oliver Spatscheck AT&T Labs—Research, Florham Park, NJ, USA

Piotr Srebrny University of Oslo, Oslo, Norway

Maciej Stroi ´nski Poznan Supercomputing and Networking Center, Pozna´n, Poland Domenico Talia ICAR-CNR, Rende (CS), Italy; DIMES, University of Calabria,

Rende (CS), Italy

Michael Talyansky ChinaCache, Sunnyvale, CA, USA

Leandros Tassiulas University of Thessaly, Oktovriou, Volos, Greece

Alexei Tumarkin ChinaCache, Sunnyvale, CA, USA

Kobus Van Der Merwe University of Utah, Salt Lake City, UT, USA

Jan W ¸eglarz Poznan Supercomputing and Networking Center, Pozna´n, Poland Hunter Xu ChinaCache, Beijing, China

CONTRIBUTORS xxv

Yinghua Ye Nokia Siemens Networks, Mountain View, CA, USA

Norihiko Yoshida Information Technology Center, Saitama University, Saitama, Japan Ken Zhang ChinaCache, Beijing, China

Marat Zhanikeev Kyushu Institute of Technology, Iizuka, Japan

PART I CDN AND MEDIA STREAMING BASICS

CLOUD-BASED CONTENT DELIVERY AND STREAMING

Content delivery networks (CDNs) [1–4] improve websites, streaming, and load performance of Internet content by end-users while reducing the cost to serve forcontent providers A CDN is a collaborative collection of network elements spanningthe Internet, where content is replicated over mirrored Web servers (i.e., point of pres-

down-ence (PoP), edge or replica servers), located at the edge of the Internet service providers’

(ISPs’) networks to which end-users are connected As shown in Figure 1.1, content isserved into the CDN once, then content is delivered to end-users from the edge servers

Advanced Content Delivery, Streaming, and Cloud Services, First Edition.

Edited by Mukaddim Pathan, Ramesh K Sitaraman, and Dom Robinson.

© 2014 John Wiley & Sons, Inc Published 2014 by John Wiley & Sons, Inc.

ISP 2

ISP 1

rather than serving each individual request directly from the content provider’s originserver Nowadays, CDNs are a prerequisite for delivering quality online experience forlive, linear, and on-demand delivery of website and media content They often leveragecloud (compute and storage) infrastructure and services to provide scalability, resiliency,availability, and performance

The reason why a CDN may be used varies depending on the particularenterprise—providing live streaming coverage of major events, distributing trainingvideos to employees, providing fast and efficient software downloads to customers, orenhancing the performance of an e-commerce website A few of these practical usecases are the following:

• Accelerated Web Performance CDNs help to improve the delivery of website

content (static and dynamic) so that websites perform better, load faster, andgenerate more revenue for content providers CDNs not only cache static con-tent at the edge but also handle dynamic, transactional content from e-commerceproviders, online auction sites, by accelerating the data transfer to provide animproved experience to end-users

• Software Updates and Downloads CDNs enhance the automatic or on-demand

delivery of software or file downloads, including software patches For example,consumers can fast download the latest release of operating system or word pro-cessing software releases online instead of ordering and purchasing in-store

• Rich Media Content Streaming CDNs help deliver rich media content, that is,

interactive digital media such as audio and video files in different encoding mats (HyperText Transfer Protocol (HTTP) adaptive streaming with MicrosoftSmooth, Adobe HDS, and Apple HLS [5,6]), to specialized streaming clientsand devices of end-users Such streaming can be live or on-demand Live content

for-CDN OVERVIEW 5

streaming provides webcasts of corporate announcements, investor briefings, andonline coverage of events such as the Olympics On-demand streaming is per-formed with stored/archived content, based on end-user request to view it

• IPTV Use Case CDNs are used for Internet Pay TV (IPTV), such as catch-up TV

service for recent programs of broadcasters, as well as online live video channels.This type of service features delivery of most media types, digital rights man-agement (DRM), multiple bitrate streaming, and distribution of video content tomultiple regions IPTV services over CDN apply both linear content delivery andon-demand content delivery

• Managing and Delivering User-Generated Content CDNs are often used to

facil-itate the hosting and distribution of user-generated content YouTube is the primeexample where a CDN is used to deliver content uploaded by end-users In addi-tion, CDNs open the way for new applications such as Enterprise TV (e.g., Enter-prise YouTube where corporate staff can upload and access video content) forinteractions, training, and knowledge sharing

• End-to-End Online Video CDNs at its fullest use are often employed as the

glue in integrating online video and content management platform to create anend-to-end delivery chain, offering a real differentiation through high quality userexperience Such end-to-end video delivery is composed of content ingest overInternet Protocol (IP) or satellite; content management and repurposing usingspecialized software, corporate storage, and hosting; content delivery using aCDN; and analytics-based management of IP-connected devices of end-users.This chapter sets the context for the book, by providing an overview of CDN tech-nologies, use cases, trends, and research directions in this domain It presents how CDN

is positioned with respect to the video ecosystem, followed by operational details of aCDN for several use cases Coverage of recent technology trends in the CDN domain isprovided to assist practitioners in this field; in addition, a few research issues are listed

to demonstrate an R&D pathway to CDN researchers While introductory concepts onmany CDN terminologies, concepts, and technologies are covered in this chapter, read-ers should refer to later chapters in the book for detailed information and discussion onrelevant CDN technologies

1.2 CDN OVERVIEW

In addition to the growing popularity of news, sports, and e-commerce website tent, accelerating demand for online HD video content, coupled with improvements inbroadband technologies, and the increased availability of media-rich content are the keydrivers behind using a CDN by content providers At the most fundamental level, aCDN is about the efficient movement of digital content (video, software, audio, andfiles) across the Internet middle mile on a massive and ever-increasing scale Let us con-sider a specific example (Figure 1.2) on how a CDN can be useful for large-scale videostreaming Good video QoE requires high network bandwidth and low network loading

con-Local video streaming Remote video streaming

Embedded viewing window

Media client

Ethernet network

1−n viewers

n −m

viewers

m−z viewers

Buffering

Aggregation

Aggregation Aggregation

Aggregation IP network

core

to avoid contention Moreover, the streaming server needs to be close to the end-user,characterized as local video streaming When significant network infrastructure (i.e.,Internet middle mile) is present between end-users and the originating media server,

in the remote video streaming scenario, satisfactory performance may be obtained at asmall scale Nevertheless, as demand grows, end-user viewing experience gets increas-ingly worse (e.g., video buffering, jitter) Centralized media server suffers performanceand reliability problems, and significant cost is incurred to deliver video files acrosslarge network distances A CDN can be used to tackle such scenario where it helps tominimize the cost of video content delivery, ensures that network resources are utilizedefficiently, and optimizes end-user experience

Without a CDN, bandwidth is overused as each request for the same content isretransmitted over and over Moreover, a “brute force best effort” solution means thatmedia content is not optimized (linear streaming vs video on demand (VoD)) for optimaledge delivery, as required for real-time video streaming With a CDN, bandwidth usage

of the content provider is optimized, as if a single end-user has requested unique tent only once No additional investment is required by the content provider to increasebandwidth capacity, as media content is packaged for delivery by the CDN infrastructure

con-1.2.1 CDN Types

The CDN market is highly dynamic, comprising a myriad of players, with different ing and targeted market segments Table 1.1 lists the different CDN types

offer-1.2.2 Market and Product Segments

A CDN is composed of content distribution and management, content routing,content edge delivery, content switching, intelligent network services, origin server

CDN OVERVIEW 7

TABLE 1.1 Types of CDNs

content without the ISPs being involved in the control anddistribution of the content itself Pure-play CDNs delivercontent over ISPs’ network (e.g., Akamai) or own infrastructure(e.g., Limelight Networks, Level 3)

AT&T, that provide content delivery as a means to reduce thedemands on the network backbone and reduce infrastructureinvestment, using hardware and software from vendors, forexample, Cisco, Juniper, or Alcatel-Lucent

component of the carrier’s network with their professionalservices group, for example, Limelight Deploy This approachleverages the expertise, infrastructure, and software of apure-play CDN

testing, and deployment on the carrier’s infrastructure Althoughintegration assistance is initially available from the pure-playCDN provider, licensed CDN is managed by the networkoperator itself For example, EdgeCast, Highwinds, and AkamaiAura provide licensed CDN product

directly against pure-play CDNs It is interesting for a contentprovider willing to deliver its content to the aggregated audience

of the federation [7,8] Cisco and Highwinds are workingtoward federated CDNs

integration, self-service portal, professional services, and core networking From aproduct perspective, a CDN provider can bundle content management, storage, andcustomer-onboarding (portal) services, in addition to the core CDN functionalitiesand value-added services (VAS) Figure 1.3 shows an example product portfolio of aCDN, showing the five constituents Although earlier CDNs have been feature-rich incore functionalities, recent market trends indicate that traditional CDN services (e.g.,caching, routing, delivery) are running the risk of being commoditized and sold mainly

on price Therefore, CDN providers are adding new services on top of their basicoffering to generate revenue from running the business There is a growing demand ofVAS (e.g., website/application acceleration, Front-End Optimization (FEO), applicationsecurity, rights protection, and analytics) and ancillary offering (e.g., origin storage,encoding/transcoding, content management, and ingest) It may also be required by

a CDN to offer personalized service and custom solutions for individual enterprisecustomer needs Details on the representative value-added and ancillary CDN servicescan be found in the later chapters of the book

Content management services,

for example, publish, store, manage

CDN portal

Origin storage services for

encoding, lngest, playout

Core CDN services, for example, caching, content delivery

Value-added services, for example, FEO, App acceleration

There are a number of driving forces in the CDN marketplace that influence theamount of revenue a CDN provider can make [9,10] A representative list of these marketdrivers is provided in Table 1.2

From a market perspective, a CDN provider offers its services to the followingmarket segments:

• E-commerce and Consumer Products/Retail CDNs offer whole-site and dynamic

content delivery services to website owners A large portion of such websites

is composed of applications and dynamic content, such as e-commerce/onlineretailers (eBay, Amazon), auction website (graysonline), and consumer productswebsite (Walmart, JC Penny) These websites are heavily used during sales andspecial events (e.g., Thanksgiving, Christmas), requiring a CDN to tackle theheavy traffic demand for them

• Media and Entertainment CDNs offer progressive download, linear, VoD, and

live streaming services to digital media companies such as content broadcasters(CNN, BBC), Internet-based publishing (Fairfax Media), and experimental dig-ital media (DXARTS, ACMI) These services are predominantly based on videocontent, with limited website static object caching Users access the VoD or livestreaming in mobile devices, set-top box, and online CDNs help to deliver thevideo content at scale as well as tackle bursty traffic [11] for special events, such

as the Cricket/Football World Cup and the Olympics

• Hospitality, Travel, and Leisure This market segment features websites and

ser-vices used by end-users for accommodation booking, holiday booking, air ticketpurchase, and so on Similarly, in e-commerce websites, the websites Zuji, Expe-dia, and HotelClub have embedded applications and dynamic content, but differ

in terms of back-end search and computation functionalities This market ment is aided by CDNs by taking care of the high traffic to websites, as well asapplication and database query execution at scale, during airfare/accommodationsale times

seg-CDN OVERVIEW 9

TABLE 1.2 CDN Market Drivers and Impacts

Uptime/availability of

service

The most crucial market driver to set up a CDN’s reputation

as a reliable provider (and hereby to attractrevenue-generating customers) is its service availability

throughput performance, as the key driver for choosing aCDN

Network performance:

first byte delivery

A CDN’s network performance, in terms of first bytedelivery, provides the very first impression of the serviceprovided by it and also gives the customer confidence onits chosen CDN

commoditized level CDNs are now focusing onvalue-added services (VAS) to generate revenue Thereare six underlying cost models that appear frequently,namely, Gbps billing, GB transfer billing, GB storagebilling, pay-as-you-go, monthly commit, and capped(percentile-based pricing)

defined by the moment the service fails and when it is toocomplex to initialize When things go wrong,

communication barriers only add to the frustration.Therefore, the customer service staff should have fluency

to communicate in the language spoken byregional/global customer

“heartbeat” reporting, analytics, and live alerts are to beprovided proactively, in order to cater for failureresolution

Customer portal/online

account management

An online account management system allows fastercustomer onboarding and allows the customer to trackthe service activation process, as well as ongoing accountand/or service management

Provider’s financial

stability

The CDN marketplace is very dynamic and has experiencednumber of acquisition and/or mergers, as well as closeoperations A CDN provider has to demonstrate that it ismaking stable revenue from the market to gain

portfolio and the VAS it provides A set of innovative andtrendy product feature set characterizes a CDN’s revenuepotential

• Banking, Financial Services, and Insurance CDNs help this market segment by

delivering a large amount of static website content, as well as dynamic contentand applications, such as calender, currency exchange, financial projections, andloan calculators Traffic profiles to these websites remain stable around the year,with high peaks only during important financial times, for example, end of thefinancial year and beginning of the year

• Public Sector/Government and Enterprise CDNs also help the Government

(such as the White House website) to deliver heavy websites of secure and staticcontent Traffic profiles to these websites are constant, featuring limited flashcrowds, except during the online release of important legislations or Governmentchanges

1.2.3 Video Ecosystem and CDN

Increased level of Internet activity, the popularity of online video, the use of richmedia for marketing and advertising, and an increase in the types and volumes ofInternet-connected devices (smart phones, iPhone, iPad, tablets, connected smart TVs,and gaming consoles) are driving significant growth in the volume of online data beingconsumed Toward this end, a number of technologies and evolving business model,centered on video and changing consumer lifestyle, are propelling the rise of onlinevideo and fundamentally transforming media distribution, TV, advertising, and contentdelivery methods

Video content by far is the significant driver for the increased utilization and vance of CDNs In fact, over 48% of global CDN revenue is associated with video con-tent delivery [12] When video and online content consumption is broadly distributed,the QoE of end-users becomes significantly important The further the content has totravel, the greater the latency in the content delivery and the poorer is the end-user expe-rience A CDN with its rich technical features, capabilities, and VAS is at the core of thevideo ecosystem, enhancing the end-user experience by accelerating the content deliverymethod (download/streaming speed) across geographically distributed end-users.Figure 1.4 shows the position of a CDN with respect to the end-to-end video ecosys-tem supply chain Video content is ingested from the source, packaged into the rightcontainer format by encoding platform, and is indexed in the storage by the contentmanagement system CDN as the delivery vehicle distributes the content among the edgeservers for delivery to IP-connected devices User profiling through reporting and ana-lytics ensures that further revenue generation opportunity is created through ad insertion,based on end-user behavioral profile Such advertisements are shown as embedded in thevideo streams, on Web pages, or within the electronic program guide (EPG) of set-topboxes

rele-1.3 WORKINGS OF A CDN

As discussed earlier, a CDN delivers static, dynamic, and streaming content On the basis

of operational functionalities, a CDN can be described from four perspectives, namely,

Connected devices

Video content

source

Content systems

Distribution over

IP network

Access networks

Edge servers User

profiling

Origin server

Ad manager

Ad selection

Video store Content

TABLE 1.3 CDN Functional Attributes

• Servers—origin and replication server

• Relationship—client-to-edge-to-origin, networkelement-to-caching proxy, and interproxy

• Interaction protocols—network elements and intercacheinteractions

• Content/service types—static, dynamic, streaming content andvalue-added services

• Content sourcing—cooperative push-based, non-cooperativepull-based, and cooperative pull-based

• Cache organization and management—caching techniques:intra- and intercluster caching; cache update: periodic,on-demand, propagation, and invalidation

• Request-routing mechanisms—DNS-based: NS and CNAMEredirection, application level, and IP anycasting; transport layer:usually combined with DNS-based techniques; application layer:HTTP redirection, URL rewriting using centralized directory,distributed hash table (DHT), or document routing

Performance

measurement

• Internal and external measurement—network statistics tion: network probing, traffic monitoring, surrogate feedback tomeasure performance metrics such as geographical proximity,latency, packet loss, average bandwidth, downtime, startup time,frame rate, server load, and simulation-based performance mea-surement

acquisi-CDN composition, content distribution and management, request routing, and mance measurement [13] Table 1.3 summarizes the functional attributes of a CDN.CDN functional attributes are closely related to each other The structure of a CDNvaries depending on the content/services it provides to its customers Within a CDNstructure, a set of edge servers is used to build the content delivery component, somecombinations of relationships and mechanisms are used for redirecting end-user requests