`
`___________________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`______________
`
`AKAMAI TECHNOLOGIES, INC.,
`Petitioner,
`
`v.
`
`LIMELIGHT NETWORKS, INC.,
`Patent Owner.
`
`______________
`Case IPR2017-00348
`
`
`Patent 8,750,155
`
`______________
`
`
`
`DECLARATION OF KEVIN C. ALMEROTH, PH.D.
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`1
`
`
`
`
`
`TABLE OF CONTENTS
`
`I. BACKGROUND AND QUALIFICATIONS ................................................ 3
`II. MATERIAL CONSIDERED ..................................................................... 12
`III. OVERVIEW AND LEGAL STANDARDS .............................................. 13
`IV. DESCRIPTION OF THE RELEVANT FIELD AND THE
`RELEVANT TIME FRAME .............................................................................. 14
`V. PERSON OF ORDINARY SKILL IN THE ART ..................................... 15
`VI. CLAIM CONSTRUCTION ....................................................................... 15
`VII. RELATED IPR2016-01011 .................................................................... 17
`VIII. OPINIONS ABOUT THE ’155 PATENT .............................................. 17
`
`
`2
`
`
`
`
`
`I, Kevin Almeroth, Ph.D., declare as follows:
`
`I.
`
`
`1.
`
`BACKGROUND AND QUALIFICATIONS
`
`I have been retained by Greenberg Traurig, LLP, on behalf of Limelight
`
`Networks, Inc. (“Limelight”), as an expert in the above-captioned proceeding. I
`
`have been asked to render an opinion regarding the validity of claims 1, 8, and 13
`
`(“Challenged Claims”) of U.S. Patent No. 8,750,155 (“the ’155 Patent”; Ex. 1001).
`
`I am being compensated at a rate of $600.00 per hour for my study and testimony
`
`in this matter. I am also being reimbursed for reasonable and customary expenses
`
`associated with my work and testimony in this matter. My compensation is not
`
`contingent on the outcome of this matter or the specifics of my testimony.
`
`
`2.
`
`I submit this declaration based on my personal knowledge and in support of
`
`Limelight’s preliminary
`
`response
`
`(“Preliminary Response”)
`
`to Akamai
`
`Technology, Inc.’s (“Akamai’s”) inter partes review petition (“Petition”) regarding
`
`the ’155 Patent.
`
` My name is Kevin C. Almeroth. I am currently a Professor in the
`3.
`
`Department of Computer Science at the University of California, Santa Barbara. I
`
`also hold an appointment and am a founding member of the Computer Engineering
`
`(CE) Program. I am a founding member of the Media Arts and Technology
`
`(MAT) Program, and the Technology Management Program (TMP). I also served
`
`as the Associate Director of the Center for Information Technology and Society
`
`3
`
`
`
`
`
`(CITS) from 1999 to 2012. I have been a faculty member at UCSB since July
`
`1997.
`
`
`4.
`
`I hold three degrees from the Georgia Institute of Technology: (1) a
`
`Bachelor of Science degree in Information and Computer Science (with minors in
`
`Economics, Technical Communication, and American Literature) earned in June,
`
`1992; (2) a Master of Science degree in Computer Science (with specialization in
`
`Networking and Systems) earned in June, 1994; and (3) a Doctor of Philosophy
`
`(Ph.D.) degree in Computer Science (Dissertation Title: Networking and System
`
`Support for the Efficient, Scalable Delivery of Services in Interactive Multimedia
`
`System, minor in Telecommunications Public Policy) earned in June, 1997.
`
`
`5.
`
`One of the major themes of my research has been the delivery of multimedia
`
`content and data between computing devices and users. In my research I have
`
`looked at large-scale content delivery systems and the use of servers located in a
`
`variety of geographic locations to provide scalable delivery to hundreds, even
`
`thousands, of users simultaneously. I have also looked at smaller-scale content
`
`delivery systems in which content, including interactive communication like voice
`
`and video data, is exchanged between computers and portable computing devices.
`
`As a broad theme, my work has examined how to exchange content more
`
`efficiently across computer networks, including the devices that switch and route
`
`data traffic. More specific topics include the scalable delivery of content to many
`
`4
`
`
`
`
`
`users, mobile computing, satellite networking, delivering content to mobile
`
`devices, and network support for data delivery in wireless and sensor networks.
`
`
`6.
`
`Beginning in 1992, at the time I started graduate school, the initial focus of
`
`my research was the provision of interactive functions (e.g., VCR-style functions
`
`like pause, rewind, and fast-forward) for near video-on-demand systems in cable
`
`systems, in particular, how to aggregate requests for movies at a cable head-end
`
`and then how to satisfy a multitude of requests using one audio/video stream to
`
`broadcast to multiple receivers simultaneously. Continued evolution of this
`
`research has resulted in the development of new techniques to scalably deliver on-
`
`demand content, including audio, video, web documents, and other types of data,
`
`through the Internet and over other types of networks, including over cable
`
`systems, broadband telephone lines, and satellite links.
`
`
`7.
`
`An important component of my research from the very beginning has been
`
`investigating the challenges of communicating multimedia content between
`
`computers and across networks. Although the early Internet was used mostly for
`
`text-based non-real time applications, the interest in sharing multimedia content
`
`quickly developed. Multimedia-based applications ranged from downloading
`
`content to a device to streaming multimedia content to be instantly used. One of
`
`the challenges was that multimedia content is typically larger than text only
`
`content, but there are also opportunities to use different delivery techniques since
`
`5
`
`
`
`
`
`multimedia content is more resilient to errors. I have worked on a variety of
`
`research problems and used a number of systems that were developed to deliver
`
`multimedia content to users.
`
`
`8.
`
`In 1994, I began to research issues associated with the development and
`
`deployment of a one-to-many communication facility (called “multicast”) in the
`
`Internet (first deployed as the Multicast Backbone, a virtual overlay network
`
`supporting one-to-many communication). Some of my more recent research
`
`endeavors have looked at how to use the scalability offered by multicast to provide
`
`streaming media support for complex applications like distance learning,
`
`distributed
`
`collaboration,
`
`distributed
`
`games,
`
`and
`
`large-scale wireless
`
`communication. Multicast has also been used as the delivery mechanism in
`
`systems that perform local filtering (i.e., sending the same content to a large
`
`number of users and allowing them to filter locally content in which they are not
`
`interested).
`
`
`9.
`
`Starting in 1997, I worked on a project to integrate the streaming media
`
`capabilities of the Internet together with the interactivity of the web. I developed a
`
`project called the Interactive Multimedia Jukebox (IMJ). Users would visit a web
`
`page and select content to view. The content would then be scheduled on one of a
`
`number of channels, including delivery to students in Georgia Tech dorms
`
`delivered via the campus cable plant. The content of each channel was delivered
`
`6
`
`
`
`
`
`using multicast communication.
`
`
`10.
`
`In the IMJ, the number of channels varied depending on the capabilities of
`
`the server including the available bandwidth of its connection to the Internet. If
`
`one of the channels was idle, the requesting user would be able to watch their
`
`selection immediately. If all channels were streaming previously selected content,
`
`the user’s selection would be queued on the channel with the shortest wait time. In
`
`the meantime, the user would see what content was currently playing on other
`
`channels, and because of the use of multicast, would be able to join one of the
`
`existing channels and watch the content at the point it was currently being
`
`transmitted.
`
` The IMJ service combined the interactivity of the web with the streaming
`11.
`
`capabilities of the Internet to create a jukebox-like service. It supported true
`
`Video-on-Demand when capacity allowed, but scaled to any number of users based
`
`on queuing requested programs. As part of the project, we obtained permission
`
`from Turner Broadcasting to transmit cartoons and other short-subject content. We
`
`also attempted to connect the IMJ into the Georgia Tech campus cable television
`
`network so that students in their dorms could use the web to request content and
`
`then view that content on one of the campus’s public access channels.
`
` More recently, I have also studied issues concerning how users choose
`12.
`
`content, especially when considering the price of that content. My research has
`
`7
`
`
`
`
`
`examined how dynamic content pricing can be used to control system load. By
`
`raising prices when systems start to become overloaded (i.e., when all available
`
`resources are fully utilized) and reducing prices when system capacity is readily
`
`available, users’ capacity to pay as well as their willingness can be used as factors
`
`in stabilizing the response time of a system. This capability is particularly useful
`
`in systems where content is downloaded or streamed on-demand to users.
`
` As a parallel research theme, starting in 1997, I began researching issues
`13.
`
`related to wireless devices and sensors. In particular, I was interested in showing
`
`how to provide greater communication capability to “lightweight devices,” i.e.,
`
`small form-factor, resource-constrained (e.g., CPU, memory, networking, and
`
`power) devices. Starting by at least 2004, I researched techniques to wirelessly
`
`disseminate information, for example advertisements, between users using ad hoc
`
`networks. In the system, called Coupons, an incentive scheme is used to
`
`encourage users to relay information, including advertisements, to other nearby
`
`users.
`
` Starting in 1998, I published several papers on my work to develop a
`14.
`
`flexible, lightweight, battery-aware network protocol stack. The lightweight
`
`protocols we envisioned were similar in nature to protocols like Universal Plug and
`
`Play (UPnP) and Digital Living Network Alliance (DLNA).
`
` From this initial work, I have made wireless networking—including ad hoc,
`15.
`
`8
`
`
`
`
`
`mesh networks and wireless devices—one of the major themes of my research.
`
`One topic includes developing applications for mobile devices, for example, virally
`
`exchanging and
`
`tracking “coupons”
`
`through “opportunistic contact” (i.e.,
`
`communication with other devices coming into communication range with a user).
`
`Other topics include building network communication among a set of mobile
`
`devices unaided by any other kind of network infrastructure. Yet another theme is
`
`monitoring wireless networks, in particular different variants of IEEE 802.11
`
`compliant networks, to (1) understand the operation of the various protocols used
`
`in real-world deployments, (2) use these measurements to characterize use of the
`
`networks and identify protocol limitations and weaknesses, and (3) propose and
`
`evaluate solutions to these problems.
`
` Protecting networks, including their operation and content, has been an
`16.
`
`underlying theme of my research almost since the beginning. Starting in 2000, I
`
`have also been involved in several projects that specifically address security,
`
`network protection, and firewalls. After significant background work, a team on
`
`which I was a member successfully submitted a $4.3M grant proposal to the Army
`
`Research Office (ARO) at the Department of Defense to propose and develop a
`
`high-speed intrusion detection system. Once the grant was awarded, we spent
`
`several years developing and meeting the milestones of the project. I have also
`
`used firewalls in developing techniques for the classroom to ensure that students
`
`9
`
`
`
`
`
`are not distracted by online content.
`
` As an important component of my research program, I have been involved in
`17.
`
`the development of academic research into available technology in the market
`
`place. One aspect of this work is my involvement in the Internet Engineering Task
`
`Force (IETF) including many content delivery-related working groups like the
`
`Audio Video Transport (AVT) group, the MBone Deployment (MBONED) group,
`
`Source Specific Multicast (SSM) group, the Inter-Domain Multicast Routing
`
`(IDMR) group, the Reliable Multicast Transport (RMT) group, the Protocol
`
`Independent Multicast (PIM) group, etc. I have also served as a member of the
`
`Multicast Directorate (MADDOGS), which oversaw the standardization of all
`
`things related to multicast in the IETF. Finally, I was the Chair of the Internet2
`
`Multicast Working Group for seven years.
`
`
`18.
`
`I am an author or co-author of approximately 200 technical papers,
`
`published software systems, IETF Internet Drafts and IETF Request for Comments
`
`(RFCs). A list of these papers is included in my CV.
`
` My involvement in the research community extends to leadership positions
`19.
`
`for several journals and conferences. I am the co-chair of the Steering Committee
`
`for the ACM Network and System Support for Digital Audio and Video
`
`(NOSSDAV) workshop and on the Steering Committees for the International
`
`Conference on Network Protocols (ICNP), ACM Sigcomm Workshop on
`
`10
`
`
`
`
`
`Challenged Networks (CHANTS), and IEEE Global Internet (GI) Symposium. I
`
`have served or am serving on the editorial boards of IEEE/ACM Transactions on
`
`Networking, IEEE Transactions on Mobile Computing, IEEE Transactions on
`
`Networks and System Management, IEEE Network, ACM Computers in
`
`Entertainment, AACE Journal of Interactive Learning Research (JILR), and ACM
`
`Computer Communications Review.
`
` Furthermore, in the courses I teach, the class spends significant time
`20.
`
`covering all aspects of the Internet including each of the layers of the Open System
`
`Interconnect (OSI) protocol stack commonly used in the Internet. These layers
`
`include the physical and data link layers and their handling of signal modulation,
`
`error control, and data transmission. I also teach DOCSIS, DSL, and other
`
`standardized protocols for communicating across a variety of physical media
`
`including cable systems, telephone lines, wireless, and high-speed Local Area
`
`Networks (LANs). I teach the configuration and operation of switches, routers,
`
`and gateways including routing and forwarding and the numerous respective
`
`protocols as they are standardized and used throughout the Internet. Topics
`
`include a wide variety of standardized Internet protocols at the Network Layer
`
`(Layer 3), Transport Layer (Layer 4), and above.
`
`
`21.
`
`In addition, I co-founded a technology company called Santa Barbara Labs
`
`that was working under a sub-contract from the U.S. Air Force to develop very
`
`11
`
`
`
`
`
`accurate emulation systems for the military’s next generation internetwork. Santa
`
`Barbara Labs’ focus was in developing an emulation platform to test the
`
`performance characteristics of the network architecture in the variety of
`
`environments in which it was expected to operate, and in particular, for network
`
`services including IPv6, multicast, Quality of Service (QoS), satellite-based
`
`communication, and security. Applications for this emulation program included
`
`communication of a variety of multimedia-based services.
`
`
`22.
`
`In addition to having co-founded a technology company myself, I have
`
`worked for, consulted with, and collaborated with companies such as IBM, Hitachi
`
`Telecom, Digital Fountain, RealNetworks, Intel Research, Cisco Systems, and
`
`Lockheed Martin.
`
`
`23.
`
`I am a Member of the Association of Computing Machinery (ACM) and a
`
`Fellow of the Institute of Electrical and Electronics Engineers (IEEE).
`
` Additional information regarding my qualifications is set forth in my current
`24.
`
`curriculum vitae, which is attached hereto as Exhibit 2002.
`
`II. MATERIAL CONSIDERED
`
` The analysis provided in this Declaration is based on my education as well
`25.
`
`as my experience in the field. In addition to relying upon my knowledge based on
`
`written materials and other information that was known prior to March 26, 2009,
`
`which I have been told is the earliest possible priority date for the ’155 Patent, I
`
`12
`
`
`
`
`
`have considered the exhibits to the Petition (Exs. 1001-1011) and the materials
`
`listed in Exhibit 2003.
`
`III. OVERVIEW AND LEGAL STANDARDS
`
`
`26.
`
`I have been asked to provide opinions regarding the validity of claims of the
`
`’155 Patent in light of several prior art patents and publications.
`
`
`27.
`
`It is my understanding that a claimed invention is unpatentable under 35
`
`U.S.C. § 103 if the differences between the invention and the prior art are such that
`
`the subject matter as a whole would have been obvious at the time the alleged
`
`invention was made to a person of ordinary skill in the art to which the claimed
`
`invention pertains (a “POSITA”). This is sometimes described as “obviousness.”
`
`I understand that an obviousness analysis takes into account the level of ordinary
`
`skill in the art, the scope and content of the prior art, and the differences between
`
`the prior art and the claimed subject matter.
`
`
`28.
`
` I understand that a petitioner in an inter partes review bears the burden for
`
`showing the obviousness of the claimed invention. I further understand that this
`
`burden may not be met by mere conclusory statements and instead must be
`
`supported by specific reasoning and record evidence.
`
`
`29.
`
`It is my understanding that the Supreme Court has recognized several
`
`rationales for combining references or modifying a reference to show obviousness
`
`of the claimed subject matter. (E.g., KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398
`
`13
`
`
`
`
`
`(2007) and other cases.) Some of these rationales include the following:
`
`combining prior art elements according to known methods to yield predictable
`
`results; simple substitution of one known element for another to obtain predictable
`
`results; a predictable use of prior art elements according to their established
`
`functions; applying a known technique to a known device to yield predictable
`
`results; choosing from a finite number of identified, predictable solutions, with a
`
`reasonable expectation of success; and some teaching, suggestion, or motivation in
`
`the prior art that would have led a POSITA to modify the prior art or combine prior
`
`art teachings to arrive at the claimed invention. I also understand that a POSITA
`
`is a person of ordinary ingenuity and creativity, not an automaton.
`
`IV. DESCRIPTION OF THE RELEVANT FIELD AND THE RELEVANT
`TIME FRAME
`
`
`30.
`
`I have carefully reviewed the specification, drawings, and claims of the ’155
`
`Patent (Ex. 1001), as well as Exhibits 1002-1011 of the Petition and the materials
`
`listed in Exhibit 2003.
`
` Based on my review of these materials, I believe that the relevant field for
`31.
`
`purposes of the ’155 Patent is the control of connection protocols used to deliver
`
`content from an information processing system, such as a content delivery network
`
`(“CDN”). (Ex. 1001 at 1:16-20.) I have been informed that the relevant timeframe
`
`is prior to March 26, 2009.
`
` As described above and in my C.V., I have extensive experience in the
`32.
`
`14
`
`
`
`
`
`relevant technical field, including experience relating to content delivery using a
`
`CDN. Based on my experience and expertise in this field, I have an understanding
`
`of the relevant field in the relevant timeframe.
`
`V.
`
`PERSON OF ORDINARY SKILL IN THE ART
`
` To assess the level of ordinary skill in the art of the ’155 Patent, I have
`33.
`
`considered the type of problems encountered in the art, the prior solutions to those
`
`problems found in prior art references, the rapidity with which innovations are
`
`made, the sophistication of the technology, the level of education of active workers
`
`in the field, and my own experience working with those of skill in the art at the
`
`time of inventions. In my opinion a person of ordinary skill in the art of the ’155
`
`Patent would have a Bachelor’s degree in Computer Science, Computer
`
`Engineering, or the equivalent, and several years of experience in the field of
`
`distributed systems, name services, or Internet content delivery.
`
`
`34.
`
`I have reviewed the level of ordinary skill in the art identified by Dr.
`
`Bhattacharjee in his declaration (Exhibit 1002 at ¶48). I find Dr. Bhattacharjee’s
`
`definition to be generally consistent with my definition, and in any event, if I were
`
`to apply his definition in my analysis, it would not change any of my opinions or
`
`conclusions.
`
`VI. CLAIM CONSTRUCTION
`
`
`35.
`
`I understand that a claim subject to inter partes review must be given its
`
`15
`
`
`
`
`
`broadest reasonable construction that is consistent with the specification of the
`
`patent, which is different from the standard that applies in litigation.
`
`
`36.
`
`I understand that the meaning of a claim term is a meaning that the term
`
`would have to a POSITA at the time of the invention. I also understand that I am
`
`to consider a claim term in the context of the entire patent, including the
`
`specification in determining the meaning to a POSITA.
`
`
`37.
`
`I note that, on page 32 of the Petition, Petitioner listed the following terms
`
`and constructions from the Markman Order issued in the District Court Lawsuit:
`
`Claim Term
`
`Construction
`
`“a request for content” (claim 1)
`
`Plain meaning
`
`“a request” (claim 13)
`
`“using information from the request”
`(claim 1)
`
`“based on the request” (claim 13)
`
`“parameters relate / relating to
`utilization of available processing or
`memory capabilities of part or all of a
`system supporting the first connection”
`(claims 1 and 13)
`
`“parameters relate / relating to
`utilization of available processing or
`memory capabilities of part or all of a
`system supporting the first connection,
`but not those relate / relating to link
`capacity or the size or type of content”
`
`“the data source is configured to
`monitor a first connection for a
`
`“the data source is configured to
`monitor a first connection for one or
`
`16
`
`
`
`
`
`request” (claim 13)
`
`more requests”
`
`
`38.
`
`I further note that Petitioner stated:
`
`“Akamai does not disagree with the Court’s constructions and submits
`that the challenged claims would have been obvious over the prior art
`under the Court’s constructions or under the broadest reasonable
`interpretations (to the extent that they differ from the Court’s
`constructions).” (Petition at 33.)
`
`For the purposes of this Declaration, I do not dispute the above constructions.
`
`VII. RELATED IPR2016-01011
`
`
`39.
`
`In addition to the present IPR, I have been retained as an expert by
`
`Greenberg Traurig, LLP, on behalf of Limelight in related IPR2016-01011, which
`
`relates to U.S. Patent No. 7,715,324 (“the ’324 patent”). I understand that the ’155
`
`patent and the ’324 patent are related patent family members. I will refer to related
`
`IPR2016-01011 herein as “the Related IPR.”
`
`VIII. OPINIONS ABOUT THE ’155 PATENT
`
`
`40.
`
`I have been informed that a patent claim can be found unpatentable as
`
`obvious where the differences between the subject matter sought to be patented
`
`and the prior art are such that the subject matter as a whole would have been
`
`obvious at the time the invention was made to a person having ordinary skill in the
`
`art in the relevant field. I understand that an obviousness analysis involves a
`
`consideration of (1) the scope and content of the prior art; (2) the differences
`
`17
`
`
`
`
`
`between the claimed inventions and the prior art; (3) the level of ordinary skill in
`
`the pertinent art; and (4) secondary considerations of non-obviousness.
`
`
`41.
`
`I understand that the Petition lists one ground for unpatentability:
`
`“Claims 1, 8, and 13 would have been obvious under 35 U.S.C. §
`103(a) over Devanneaux, Chu, and Haverstock.” (Petition at 35.)
`
`
`42.
`
`I understand that Petitioner’s arguments for using Chu’s plug-in architecture
`
`in Devanneaux’s operating system are generally because both Devanneaux and
`
`Chu are in the same technical field and teach performing TCP connection
`
`optimization. (Petition at 25.) Petitioner also argues that Devanneaux has a kernel
`
`and Chu has a network stack. (Petition at 25-26.) Based on these basic
`
`similarities, Petitioner simply concludes a motivation would have existed. I
`
`disagree because, as set forth in more detail below, combining Chu with
`
`Devanneaux would not yield predictable results, would be beyond the skill of a
`
`POSITA, and the combination would disrupt normal edge server operations.
`
` Devanneaux relates to content-delivery networks (“CDN”), which are
`43.
`
`uncontrolled, highly unpredictable server-based networks configured to deliver
`
`internet content to end-users. (Ex. 1003 at ¶¶6-7.) For example, Devanneaux
`
`states:
`
`“As such, the capability and flexibility of the supporting Internet
`infrastructure for the Web site becomes mission-critical. In particular,
`the infrastructure must provide good performance for all end user
`
`18
`
`
`
`
`
`consumers, regardless of their location. The site must scale to
`handle high traffic load during peak usage periods. It must
`remain available 24x7, regardless of conditions on the Internet.
`When performance, reliability, or scalability problems do occur, Web
`site adoption and usage can be negatively impacted, resulting in
`greater costs, decreased revenue, and customer satisfaction issues.
`
`It is known in the prior art to off-load Web site content for delivery by
`a third party distributed computer system. One such distributed
`computer system is a ‘content delivery network’ or ‘CDN’ that is
`operated and managed by a service provider.” (Ex. 1004 at ¶¶6-7.)
`
`Thus, servers in a CDN must be capable of “provid[ing] good performance”
`
`regardless of the end user consumer’s location and regardless of the network traffic
`
`patterns, which may include spikes in network traffic or prolonged periods of high
`
`traffic during peak usage periods. (Id.) Accordingly, Devanneaux teaches that
`
`CDN traffic patterns are unpredictable and uncontrolled. Such a teaching is
`
`consistent with what a POSITA would have understood about one of the principle
`
`motivations for using a CDN: flash crowds. Flash crowds are large unanticipated
`
`spikes in requests, sometimes caused by external events of widespread interest.
`
`Such events create significant spikes in demand and are not predictable. Another
`
`patent from Akamai, U.S. Patent No. 6,108,703, highlights this unpredictability
`
`while describing the motivation for using a CDN as an alternative to a content
`
`provider using their own resources:
`
`19
`
`
`
`
`
`“Although mirroring and related load-balancing solutions do allow a
`Content Provider to distribute load across a collection of servers, the
`aggregate capacity of the servers must be sufficient to handle peak
`demands. This means that the Provider must purchase and maintain a
`level of resources commensurate with the anticipated peak load
`instead of the true average load. Given the highly variable and
`unpredictable nature of the Internet, such solutions are expensive and
`highly wasteful of resources.” (Ex. 2007 at 15:16-25.)
`
` Thus, CDN customers use a CDN to “provide good performance” to end
`44.
`
`users regardless of the network traffic patterns, which may include spikes in
`
`network traffic or prolonged periods of high traffic during peak usage periods.
`
`(Id.; see also ¶52, below.) CDNs use edge servers, such as the one provided in
`
`Devanneaux, to support the high volume of requests that generally accompany
`
`these high traffic periods and to otherwise improve the speed of delivery of
`
`content. (See Ex. 1003 at Abstract; ¶6.) Thus, edge servers are used to reduce
`
`latency when responding to content requests.
`
` At the other end of the spectrum is Chu, which focuses on a plug-in TCP
`45.
`
`architecture for end-user computer workstations or PCs, as opposed to content
`
`servers. Indeed, Chu uses the word “server” only six times and only in paragraphs
`
`54, 69, and 70. Of these, paragraphs 69 & 70 are instructive:
`
`“The plug-in approach also enables employing an aggressive,
`special-purpose technique in a controlled network environment. For
`
`20
`
`
`
`
`
`instance, a server in a data center with a well-controlled traffic
`pattern or well-tuned queuing model might deploy a non-compliant
`congestion control technique that allows packets to be sent without
`slow-start or any bandwidth throttling. This technique could be
`useful, for example, to eliminate the overhead of congestion control
`for connections that transfer data between two servers on a
`dedicated network link, or to expedite connections that exchange
`cluster membership heartbeat messages within the data center.
`Previously, such service variation either was not possible, or would
`require multiple servers.
`
`Finally, per-connection tuning can also be used to deploy and test
`experimental TCP behaviors on a limit set of TCP connections on a
`production server without exposing other, normal operations on the
`server to the riskier new behavior.” (Ex. 1004 at ¶¶69-70 (emphasis
`added).)
`
`Chu’s “well-controlled” network requirements for servers employing its “special-
`
`purpose technique” are in stark contrast to the CDN edge servers of Devanneaux
`
`which reside on the Internet because, as discussed above, the edge servers may be
`
`subject to a variety of network traffic patterns, including “spikes in network traffic
`
`or prolonged periods of high traffic during peak usage periods.” (Id.) In addition,
`
`the explanation in paragraph 70 shows that Chu’s architecture may result in “riskier
`
`new behavior.” Chu’s recognition that production servers needed to be protected
`
`21
`
`
`
`
`
`from “riskier new behavior” is completely at odds with the argument that
`
`combining Chu and Devanneaux would yield “predictable results.”
`
` Chu discloses a plug-in architecture for a network stack in an operating
`46.
`
`system, which includes changing the TCP behavior of a network connection. (Ex.
`
`1004 at Abstract at ¶¶0038-39.) Figure 3 of Chu, reproduced below, “presents a
`
`flow chart illustrating the process of changing the TCP behavior of a network
`
`connection.” (Ex. 1004 at ¶67.)
`
` Specifically, when the Chu system “determines or is notified of a need for
`47.
`
`changing the TCP behavior of a network connection (step 302)[,] the system
`
`
`
`22
`
`
`
`
`
`disables a relevant portion of the network stack in order to put the network
`
`connection into a quiescent state (step 304).” (Id. at ¶67.) “Then, the system
`
`changes the function pointer for the function associated with the TCP behavior to
`
`point to a new function with the desired behavior (step 306). Finally, the system
`
`re-enables the corresponding portion of the network stack to return the network
`
`connection to an active state (step 308).” (Id.)
`
` For the reasons set forth below, this mechanism is not suitable for the edge
`48.
`
`servers of Devanneaux because it would disrupt normal edge server operations and
`
`ultimately undermine the primary reason for using a CDN: improving the
`
`performance of web sites for end users.
`
` As described above in ¶¶43-44, above, Devanneaux relates to content-
`49.
`
`delivery networks (“CDN”), which are uncontrolled, highly unpredictable server-
`
`based networks configured to deliver internet content to end-users. (Ex. 1003 at
`
`¶¶6-7.) The edge servers of a CDN are used to support the high volume of requests
`
`that generally accompany high traffic periods and to otherwise improve the speed
`
`of delivery of content. (See Ex. 1003 at Abstract; ¶6.) Thus, edge servers are used
`
`to reduce latency when responding to content requests. Accordingly, a POSITA
`
`would not modify an edge server in a way that would increase latency or otherwise
`
`reduce the performance, particularly during high load and high traffic periods.
`
` As described above in ¶46, the mechanism of Chu relied on by Petitioner
`50.
`
`23
`
`
`
`
`
`includes Chu suspending and subsequently re-enabling a connection. Chu notes
`
`that since suspending and subsequently re-enabling a connection “occurs quickly
`
`enough, and the system typically has capacity to buffer packets, there is effectively
`
`no interruption of network service.” (Ex. 1004 at ¶67 (emphasis added).)
`
`However,