`U.S. Patent No. 8,671,132
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`
`MICROSOFT CORPORATION,
`Petitioner,
`
`v.
`
`DAEDALUS BLUE, LLC,
`Patent Owner.
`
`Case IPR2021-00831
`Patent No. 8,671,132
`
`DECLARATION OF DR. CHRISTOPHER JULES WHITE
`IN SUPPORT OF PATENT OWNER DAEDALUS BLUE, LLC’S
`PRELIMINARY RESPONSE
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 1 of 56
`
`
`
`
`
`
`I.
`
`INTRODUCTION
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`1. My name is Dr. Christopher Jules White, and I am currently an
`
`Associate Professor of Computer Science in the Department of Electrical
`
`Engineering and Computer Science, as well as the Associate Dean for Strategic
`
`Learning Programs at Vanderbilt University.
`
`2.
`
`I have personal knowledge of the facts and opinions set forth in this
`
`declaration, and, if called upon to do so, I would testify competently thereto.
`
`3.
`
`I have been asked by Patent Owner Daedalus Blue, LLC (“Daedalus”)
`
`to provide my opinions and analysis responsive to certain issues raised by the
`
`Petition for Inter Partes Review of U.S. Patent No. 8,671,132 (“Petition”) by
`
`Petitioner Microsoft Corporation (“Microsoft” or “Petitioner”) and the supporting
`
`Declaration of Dr. Erez Zadok (“Zadok Declaration”). For this work I am being
`
`compensated at my normal hourly rate of $450 per hour plus reasonable expenses.
`
`The amount of my compensation is not dependent upon the substance of my
`
`opinions or upon the outcome of this matter. I am working as a private consultant
`
`on this matter and the opinions presented here are my own.
`
`4.
`
`I have prepared this declaration at the request of Daedalus and its
`
`counsel. For purposes of this declaration, I have been asked to provide my
`
`opinions on certain aspects of Petitioner’s theories regarding patentability of U.S.
`
`Patent No. 8,671,132 (“the ’132 patent”) as it relates to this IPR proceeding. If an
`
`1
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 2 of 56
`
`
`
`
`
`
`inter partes review is instituted, I reserve the right to provide additional opinions
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`on the state of the art, the cited references in the Petition, the obviousness
`
`arguments presented in the Petition, the validity of the ’132 Patent claims in light
`
`of the cited references under 35 U.S.C. §§102 and/or 103, claim construction, and
`
`any other issues arising in connection with the Petition and these proceedings.
`
`5.
`
`In forming my opinions, I have relied on my knowledge and
`
`experience in computer science, hardware and software design, and wireless
`
`communication systems, and on the documents and information described below. I
`
`have also relied on my work experience with Ziiio, Optio Labs, Cloudpoint/PAR
`
`Works, and IBM. I may also review other materials throughout this case,
`
`including other documents or testimony that may emerge in this case. Those
`
`materials may affect my opinions in this matter.
`
`6.
`
`I reserve the right to modify and supplement my analysis and
`
`conclusions set forth in this declaration based upon any additional evidence,
`
`briefing, or decisions submitted in these proceedings. I also reserve the right to
`
`modify and supplement my analysis and conclusions set forth in this declaration
`
`based upon any change to any of the applicable legal standards explained to me by
`
`Daedalus’ counsel.
`
`2
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 3 of 56
`
`
`
`
`
`
`II. QUALIFICATIONS AND BACKGROUND
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`7.
`
`I have over 20 years of experience in computer science and computer
`
`engineering. I received a Bachelor’s degree in Computer Science from Brown
`
`University, and a Master of Science and Ph.D. in Computer Science from
`
`Vanderbilt University. I am currently an Associate Professor of Computer Science
`
`in the Department of Computer Science at Vanderbilt University. I was also
`
`previously an Assistant Professor in Computer Engineering at Virginia Tech.
`
`8.
`
`I have also held a variety of positions within industry and in affiliation
`
`with Federally-supported research labs. Proceeding in reverse chronological order:
`
`From 2012-2015, I was Chief Technology Officer of Optio Labs, Inc., a company
`
`that I co-founded. From 2012-2016, I was the Chief Scientist of Cloudpoint Labs, a
`
`company that I co-founded. From 2011-2012, I was a Visiting Research Scientist
`
`at the Carnegie Mellon Software Engineering Institute, a Federally-funded research
`
`and development center. From 2001-2002, I worked as a software engineer at IBM.
`
`9.
`
` I have published over 130 papers on topics ranging from mobile
`
`computing to cyber-security to cloud computing. According to Google Scholar,
`
`there are 5,351 citations to these papers. In addition to publishing on a variety of
`
`research topics, I have received several important distinctions. I received a
`
`National Science Foundation (NSF) CAREER Award, which is one of the NSF’s
`
`“most prestigious awards in support of early-career faculty who have the potential
`
`3
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 4 of 56
`
`
`
`
`
`
`to serve as academic role models in research and education and to lead advances in
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`the mission of their department or organization.” I have also received several Best
`
`Paper Awards for research work in domains ranging from Software Product-lines
`
`to Civil Engineering.
`
`10. As a faculty member, I run the Mobile Application computinG,
`
`optimizatoN, and secUrity Methods (MAGNUM) Group at Vanderbilt University.
`
`As a Principal Investigator or Co-Principal Investigator, I have received over
`
`$17,000,000 in grant support, both from Federal entities, such as the National
`
`Science Foundation (NSF), Defense Advanced Research Projects Agency
`
`(DARPA), Air Force Research Labs (AFRL), and Office of Naval Research
`
`(ONR), and private entities, such as Varian Medical Systems and NOAH
`
`Basketball. My research focuses on securing, optimizing, and leveraging data from
`
`mobile cyber-physical systems. My mobile cyber- physical systems research spans
`
`four key focus areas: (1) mobile, cloud, and CPS security and data collection, (2)
`
`high-precision mobile understanding of the world through indoor positioning and
`
`augmented reality, (3) CPS, mobile device and cloud infrastructure power and
`
`configuration optimization, and (4) applications of mobile cyber-physical systems
`
`in multi-disciplinary domains, including manufacturing, energy-optimized cloud
`
`computing, smart grid systems, healthcare/manufacturing security, next-generation
`
`construction technologies, citizen science, and augmented reality.
`
`4
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 5 of 56
`
`
`
`
`
`
`
`11. My research work has been transitioned to industry where it has both
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`won patent protection and led to venture-backed startup companies. My work on
`
`securing mobile devices was licensed into Optio Labs, a company that I co-
`
`founded. Optio Labs received over $10,000,000 in venture capital backing. My
`
`work on mobile augmented reality was licensed into PAR Works / Cloudpoint
`
`Labs, a company that I co-founded. The company received $1,000,000 in venture
`
`backing and won awards, such as being an Innovation Award Honoree at CES and
`
`a finalist for Technical Achievement Award at SXSW. I am an inventor on
`
`multiple patents.
`
`12. Attached hereto as Appendix A is a true and correct copy of my
`
`Curriculum Vitae describing my background and experience.
`
`III.
`
`INFORMATION CONSIDERED
`
`13.
`
`In forming my opinions, I have reviewed and considered, in addition
`
`to my knowledge and experience noted above, the following materials:
`
`a. Microsoft Corporation Petition for Inter Partes Review of U.S.
`
`Patent No. 8,671,132 (“Petition”) & Exhibits Thereto;
`
`b. U.S. Patent No. 8,671,132 (“the ’132 patent”);
`
`c. U.S. Patent No. 8,671,132 File History;
`
`d. Declaration of Dr. Erez Zadok (“Zadok Declaration”);
`
`14.
`
`I have also reviewed and relied upon the materials cited herein.
`
`5
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 6 of 56
`
`
`
`
`
`
`IV. ONE OF ORDINARY SKILL
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`15.
`
`I understand that the Petition proposed a level of skill for a person of
`
`ordinary skill in the art (“POSITA”). I understand that the Petition indicates that a
`
`POSITA “would have had at least a bachelor’s degree in computer science,
`
`electrical engineering, computer engineering, or a similar discipline, along with
`
`two years of experience in the design, operation, or control of data management
`
`systems, storage systems, and/or distributed storage systems” and that “[a] more
`
`advanced degree can substitute for work experience and vice versa.” Petition, 13.
`
`16. For the purpose of this Declaration, I have applied this level of skill to
`
`evaluate the Petition Grounds.
`
`V. UNDERSTANDINGS OF LAW
`
`17.
`
`I am informed and understand that a claim is unpatentable if it is
`
`anticipated by the prior art. I am further informed and understand that a claim is
`
`anticipated only if each and every element as set forth in the claim is found
`
`expressly or inherently described, in a single art reference. I also understand that a
`
`single reference cannot merely disclose each element of a claim to be found to
`
`anticipate. Rather it must disclose all of the elements as arranged in the claim.
`
`18.
`
`I am further informed and understand that anticipation by a
`
`preponderance of the evidence requires a showing that a person could make and
`
`use the claimed invention by looking at one reference. I also understand that a
`
`6
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 7 of 56
`
`
`
`
`
`
`requirement of a claim that is missing from a prior art reference may be disclosed
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`inherently if that missing requirement is necessarily present in the reference. I am
`
`further informed and understand that the reference to be considered anticipatory
`
`must enable and describe the claimed invention sufficiently to have placed it in the
`
`possession of a person of ordinary skill in the art.
`
`19.
`
`I am informed and understand that evaluating obviousness requires
`
`assessing if the differences between the claimed invention and the prior art are
`
`such that the claimed invention as a whole would have been obvious in light of
`
`prior art. I am further informed and understand that a number of factors are to be
`
`considered in that determination, including determining the scope and content of
`
`the prior art, ascertaining the differences between the claimed invention and the
`
`prior art, and determining the level of ordinary skill in the pertinent art. I am
`
`further informed and understand that in evaluating obviousness, a variety of
`
`rationales may support an obviousness determination. For example, I understand
`
`that obviousness may be supported by rationales such as combining known prior
`
`art elements according to known methods to yield predictable results, or simple
`
`substitution of one known element for another to obtain predictable results. I also
`
`understand that obviousness can be supported where there was a teaching,
`
`suggestion, or motivation in the prior art that would have led a POSITA to modify
`
`a prior art reference or combine prior art reference teachings to arrive at the
`
`7
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 8 of 56
`
`
`
`
`
`
`claimed invention. I am further informed and understand that there must be a
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`motivation to make the combination and a reasonable expectation that such a
`
`combination would be successful in order to support a conclusion of obviousness.
`
`I further understand that a reference may “teach away” from a claimed invention,
`
`and that “teaching away” cuts against arguments that a claimed invention would
`
`have been obvious and supports non-obviousness. I understand that a reference
`
`“teaches away” from a claimed invention when a POSITA would have been
`
`discouraged from following the reference’s path, or would be led in a direction
`
`different from the path taken in the claimed invention.
`
`20.
`
`I am informed and understand that in order for a reference to be relied
`
`on to demonstrate obviousness, it must be “analogous art” to the invention argued
`
`to be obvious. I further am informed and understand that a reference can be
`
`“analogous art” in two ways. The first way is if the reference is “from the same
`
`field of endeavor” as the patent. In that case, a reference can be analogous art even
`
`if it addresses a different problem, as long as it is in the same field. The second
`
`way is if the reference is “reasonably pertinent to the problem faced by the
`
`inventor.” In that second case, the reference need not be in the same field of
`
`endeavor as the claimed invention that is argued to render obvious. I am further
`
`informed and understand that in order to be “reasonably pertinent,” a reference
`
`“logically would have commended itself to an inventor’s attention in considering
`
`8
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 9 of 56
`
`
`
`
`
`
`his problem.” I am also informed and understand that, in evaluating whether a
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`reference would have been reasonably pertinent to the problem the invention seeks
`
`to solve, the “purposes of both the invention and the prior art are important,” and if
`
`the prior art is directed to a different purpose than the claimed invention, an
`
`inventor would “have had less motivation or occasion to consider it.”
`
`VI. THE PRIOR ART
`
`A. U.S. Patent No. 5,018,060 (“Gelb”)
`21. U.S. Patent No. 5,018,060 (“Gelb”) is titled “Allocating Data Storage
`
`Space of Peripheral Data Storage Devices Using Implied Allocation Based on User
`
`Parameters.” EX1005, 1. According to the information on the face of the patent,
`
`Gelb was filed on January 26, 1989, issued on May 21, 1991, and assigned to IBM.
`
`22. Based on my review of the specification, Gelb is directed to an IBM
`
`mainframe with data storage devices directly attached to the mainframe. EX1005,
`
`1:8-14, 1:22-30. Job Control Language (JCL), which is discussed at 1:49-59 and
`
`elsewhere, is the name of scripting languages used on IBM mainframe operating
`
`systems at the time of Gelb’s filing. IBM patents, papers, and products are cited
`
`throughout the Background of the Invention. EX1005, 2:5-31, 2:52-3:29, 3:46-53,
`
`3:65-4:11. IBM products are also used to refer to exemplary components, user
`
`interfaces, and code excerpts. Id., 6:8-40, 10:25-39, 11:51-57, 15:64-16:59. Figures
`
`4-5 are also an illustration of an IBM mainframe system. Id., Figs. 4-5.
`
`9
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 10 of 56
`
`
`
`
`
`
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`“Tivoli Storage Manager: A Technical Introduction” (“Tivoli”)
`B.
`23. Tivoli is a document entitled “Tivoli Storage Manager: A Technical
`
`Introduction” (“Tivoli”). EX1006, 1. I understand that Petitioner contends that
`
`Tivoli “was published and publicly accessible at least by June 15, 2001 and no
`
`later than November 28, 2001.” Petition, 16-17. Tivoli states that it is a “high-
`
`level technical introduction to Tivoli Storage Manager.” EX1006, 4.
`
`24. Tivoli purports to describe a distributed system comprised of multiple
`
`client systems, “including laptop computers, PCs, workstations, or server systems,”
`
`connected by a network to a Tivoli Storage Manager server, Tivoli Storage
`
`Manager database, and storage repositories. See e.g., EX1006, 7-12, 14-16.
`
`C. U.S. Patent No. 7,269,612 (“Devarakonda”)
`25. U.S. Patent No. 7,269,612 (“Devarakonda”) is titled “Method,
`
`System, and Program for a Policy Based Storage Manager.” EX1008, 1.
`
`According to the information on the face of the patent, Devarakonda was filed on
`
`May 31, 2002, issued on September 11, 2007, and assigned to IBM.
`
`VII. PETITIONER’S GROUNDS 1 AND 2: GELB IN VIEW OF TIVOLI
`
`26.
`
`I understand that Petitioner challenges claims 15-21 and 23-25 as
`
`obvious over Gelb in view of Tivoli and the knowledge of a POSITA (Ground 1),
`
`and further challenges claim 22 as obvious over Gelb in view of Tivoli and
`
`Callaghan and the knowledge of a POSITA (Ground 2). See Petition, 20, 41.
`
`10
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 11 of 56
`
`
`
`
`
`
`
`27.
`
`I further understand that Petitioner relies on Gelb combined with
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`Tivoli to teach certain limitations of the independent challenged claims, including
`
`“receiving one or more attributes of a file from one of a plurality of clients, the
`
`clients comprising at least two different computing platforms” in Claims 15, 23.
`
`A. Motivation to Combine Gelb and Tivoli
`28. Petitioner states that “Gelb and Tivoli are a natural combination”
`
`(Petition, 35) because they both “describe storage management systems that have
`
`similar structures, purposes, and functions” (Petition, 37). I disagree. Gelb and
`
`Tivoli represent entirely different types of computing systems with very different
`
`requirements and would not make sense to combine. In my opinion, at the time of
`
`the invention, a POSITA would not have chosen to modify Gelb’s single-machine
`
`architecture with specific aspects of Tivoli’s architecture for backing up data.
`
`29. Gelb is focused on Job Control Language (JCL) programs for a
`
`mainframe—a single machine architecture with a single operating system, and
`
`directly attached peripheral storage. Mainframes are large computers that are
`
`typically used in systems where the system designer is willing to incur extra
`
`hardware costs to run the system on a single machine and avoid the complexities of
`
`using a multiple machines in a distributed architecture. Mainframes at the time
`
`generally had significant processing power, memory, cache, and I/O bandwidth to
`
`communicate with storage devices directly-connected to their buses. See EX1005,
`
`11
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 12 of 56
`
`
`
`
`
`
`1:8-14, 5:19-21, 5:31-37, 15:52-17:15, Figs. 4-5. The name is derived from ‘main
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`frame,’ the cabinet traditionally used to house such computers. EX2011, 3.
`
`30. Tivoli, in contrast, is a backup/archive system for heterogeneous
`
`enterprise network environments that include multiple operating systems. See
`
`EX1006, 4, 7. Tivoli’s architecture is based upon using network links with limited
`
`bandwidth to connect a set of client workstations with different operating systems,
`
`storage management servers, and storage devices. Id. Separate application clients
`
`are used on each workstation to connect to the storage management server. Id., 13.
`
`31. Gelb and Tivoli employ two completely different types of computing
`
`systems: Gelb is a homogeneous single-machine mainframe system; Tivoli is a
`
`heterogeneous distributed system built on a more resource-constrained network. A
`
`POSITA would consider these to be very different systems that are architecturally
`
`incompatible.
`
`32. Gelb and Tivoli are designed to solve two entirely different computing
`
`tasks. Gelb is designed for high-capacity high-performance computing with the
`
`actively used copy of data. Tivoli is designed for off-hours backup, archive, and
`
`disaster recovery. Id., 7. The task of backing up data is inherently a low-priority
`
`task without strict performance needs. A backup copy of a file is not the actively
`
`accessed copy and does not have many requirements, other than being available
`
`when needed in a disaster. Id., 16-17. The backup copy is not actively accessed and
`
`12
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 13 of 56
`
`
`
`
`
`
`instead needs to be “restored” before it can used by a client workstation. Id. Thus,
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`Tivoli is meant to solve at an entirely different problem than Gelb. A POSITA
`
`would not consider them to be “similar.”
`
`33. Petitioner claims that both Gelb and Tivoli “have an administrative
`
`component and a client component” (Petition, 37). This statement is incorrect.
`
`Gelb does not discuss the “clients” described in the patent. The “clients” in the
`
`patent are part of a distributed system where “client” refers to computing elements
`
`that are talking to the system through a network and capable of running on
`
`different computing platforms. As described above, Gelb is built on a single-
`
`machine mainframe architecture that is not distributed. Since Gelb does not have a
`
`distributed system architecture, it cannot be referring to the same “clients.”
`
`34.
`
`In my opinion, a POSITA would not have been motivated to modify
`
`Gelb based on the teachings in Tivoli because these fundamental differences in
`
`system architecture and purpose are incompatible with each other. Gelb does not
`
`describe key elements of Tivoli’s system architecture (distributed, heterogeneous,
`
`networked). Further, there is nothing in Gelb that suggests looking toward Tivoli
`
`for teachings, which makes sense since it is a distributed system architecture.
`
`35. Petitioner argues that Gelb and Tivoli are “sufficiently related and
`
`complementary to yield a successful combination of elements.” Petition, 37. I
`
`disagree. For example, Tivoli’s client applications (i.e., the applications on the
`
`13
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 14 of 56
`
`
`
`
`
`
`client workstations) are designed to backup data, which is (1) inherently a low
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`priority task without strict performance or prioritization requirements, (2) usually
`
`done on an incremental and/or whole disk basis (i.e., not on a file basis), and (3)
`
`has nothing to do with the actively used copy of data used for computing tasks on
`
`the clients. A POSITA would recognize that Tivoli would lack the performance,
`
`prioritization, and careful distributed resource management needed to be successful
`
`in supporting high performance applications like those in the mainframe in Gelb.
`
`36.
`
`In addition, Gelb’s high-performance applications are designed for a
`
`homogeneous computing platform with a single operating system, and thus the
`
`“administration component” is not designed to facilitate cross-platform operations.
`
`Gelb does not teach how this “administration component” would manage cross-
`
`platform differences or nuances. A POSITA would not have considered Tivoli’s
`
`disclosure of client workstations of different operating systems to have utility in
`
`Gelb or that Tivoli provide needed information to modify Gelb’s “administration
`
`component” to handle cross-platform issues.
`
`37. Petitioner attempts to draw parallels between the policies and classes
`
`in Gelb and Tivoli. But the only classes shared are management classes. Tivoli
`
`does not have (or disclose) data classes or storage classes like Gelb. In addition,
`
`there is no concept of service in Tivoli, particularly regarding performance goals,
`
`which typically weren’t a major concern of backup systems at the time. It would be
`
`14
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 15 of 56
`
`
`
`
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`
`non-obvious to a POSITA to combine such dissimilar systems, one system
`
`designed to meet performance and other needs of storing and accessing “primary”
`
`data on a single system and computing platform with a client-server backup
`
`solution that is fundamentally not designed for performance or even dealing with
`
`access to the primary data (e.g., concerned with managing the “backup” secondary
`
`copy).
`
`38. To build a distributed client system that supports high-performance
`
`applications like those to which Gelb is directed, a POSITA would have
`
`recognized that the client must be capable of managing allocation across multiple
`
`computing platforms and network components. The ’132 patent explicitly
`
`discusses how clients on multiple platforms participate in resource allocation of a
`
`diverse set of resources, ranging from cache, to I/O, to network traffic
`
`prioritization. In contrast, a POSITA considering Tivoli would have understood
`
`that Tivoli does not sufficiently address or describe these issues, because Tivoli is
`
`a document addressed to a backup system, and is not directed to supporting high
`
`performance applications across multiple computing platforms and network
`
`components. In other words, Tivoli would not enable a POSITA to successfully
`
`implement a client capable of supporting Gelb’s applications across multiple
`
`computing platforms and network components.
`
`15
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 16 of 56
`
`
`
`
`
`
`
`39. A POSITA would have considered that, contrary to what the Petition
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`states (Petition, 37), applying a client architecture not designed for performance
`
`(Tivoli) to a system that supports high-performance data access (Gelb) would
`
`likely yield a worse system. When systems with fundamentally diverging
`
`requirements are combined, the result is often a worse performing system that
`
`meets none of the needs of either individual system. A POSITA would have
`
`recognized the fundamental importance of only combining systems with
`
`compatible requirements and goals.
`
`40. Petitioner argues that a POSITA would have been motivated to
`
`transform Gelb’s system in view of Tivoli by (1) moving the ‘application
`
`programs’ in Gelb to “a client computer system instead of in the same system as
`
`the storage management software” (Petition, 38); (2) modify it further so that “the
`
`client computers may use different operating systems,” which would require
`
`modifying the application programs so that they can work on different operating
`
`systems (Petition, 39); and (3) transforming Gelb’s storage manager into an open
`
`systems model so that it can work with diverse storage resources (Petition, 40). In
`
`other words, Petitioner proposes changing the entire system architecture of Gelb.
`
`41. Petitioner (and its expert) argue that these changes would have been
`
`simple, straightforward, “standard practice,” and “easily accomplished.” Petition,
`
`39-41. I completely disagree. For example, it is incorrect to state that “[a] POSITA
`
`16
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 17 of 56
`
`
`
`
`
`
`would have easily created versions of the client software to accommodate multiple
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`operating systems.” Petition, 41. A software solution that works on multiple
`
`operating systems is typically described as “cross-platform.” In the early-2000s,
`
`creating solutions that were cross-platform was extremely difficult. The assertion
`
`by Petitioner fails to properly account for the vast complexity of attempting to
`
`develop software for multiple computing platforms, each of which has different
`
`resources, application programming interfaces, data formats, and many other
`
`distinctions. Especially for performance sensitive applications like in Gelb, these
`
`differences are critical and commonly necessitate writing multiple independent
`
`code bases, which vastly increases cost and complexity, while significantly
`
`reducing the chance of success. Tivoli does not provide the information that a
`
`POSITA would have needed to navigate these complexities and successfully build
`
`a cross-platform client solution for Gelb.
`
`42. Petitioner claims a POSITA would have had a reasonable chance of
`
`success in modifying Gelb in view of Tivoli in this way. I disagree. Gelb describes
`
`a system designed for a single machine. It is rarely, if ever, possible to directly
`
`translate a single machine solution into a distributed system without solving a
`
`number of complex issues, ranging from timing to synchronization to messaging
`
`unreliability. These issues are particularly important (and difficult) when trying to
`
`17
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 18 of 56
`
`
`
`
`
`
`design a distributed system to meet the performance requirements of application
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`programs like those that run on Gelb. These issues include, but are not limited to:
`
`a.
`
`Unlike single-machine systems, distributed systems have
`
`unreliable messaging between elements of the system, so a POSITA cannot
`
`directly apply single-machine solutions to distributed systems or architectures.
`
`Whereas in a single machine system, messages sent between system components,
`
`such as processes, are highly reliable, messages in distributed systems can and do
`
`fail to be delivered. Distributed systems that are not designed to account for lost
`
`messages, for example by resending messages, fail to operate properly.
`
`b.
`
`Unlike single-machine systems, distributed systems often have
`
`clients on different computing platforms, which in turn have completely different
`
`mechanisms for controlling resources like RAM, I/O capacity, networking, that
`
`would make a single machine solution inappropriate. Heterogeneous distributed
`
`system architectures must provide mechanisms for mapping from high-level goals
`
`down to the individual capabilities of clients. System architectures that do not
`
`account for differences in resource handling and capability do not perform well.
`
`c.
`
`Unlike single-machine systems, the networks underlying
`
`distributed systems have more limited bandwidth than the internal buses on a
`
`single machine. The bandwidth constraints in a heterogeneous distributed system
`
`typically necessitate optimization of message size, frequency, as well as bandwidth
`
`18
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 19 of 56
`
`
`
`
`
`
`allocation to individual clients, otherwise network links quickly become saturated
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`and degrade overall performance. Single machine architectures that do not account
`
`for limitations in network bandwidth often fail in heterogeneous distributed
`
`systems.
`
`d.
`
`Unlike single-machine systems, the networks underlying
`
`distributed systems often have higher latency than internal bus communication on a
`
`single machine. Latency creates timing delays that can significantly slow down a
`
`computational process if not considered properly in the protocols supporting a
`
`storage system. For example, communication protocols that rely on lots of back
`
`and forth communication may introduce little overhead on a single machine but
`
`yield dramatic slowdowns when between machines over a high latency network.
`
`e.
`
`Unlike single-machine systems, scheduling in distributed
`
`systems, especially those with diverse operating systems, is more difficult. A
`
`single machine can easily see the state of the workloads and plan / optimize
`
`scheduling decisions to meet performance goals. In a heterogeneous distributed
`
`system, the workload state needed to support decision making for the system is
`
`spread across numerous computing systems and can be difficult to capture.
`
`f.
`
`Unlike single-machine systems, timing of resource management
`
`operations are much more difficult on a distributed system and require advancing
`
`planning and synchronization across clients. Because of the unreliable messaging,
`
`19
`
`IPR2021-00831
`
`Daedalus EX2001
`Page 20 of 56
`
`
`
`
`
`
`latency, and challenges capturing overall state, making sure actions have been
`
`IPR2021-00831
`U.S. Patent No. 8,671,132
`
`communicated and completed in a distributed system can be very difficult.
`
`g.
`
`Unlike single-machine systems where there is no need for
`
`consensus, in a distributed system where the actors all have to arrive at a consensus
`
`about what should be done, different parts of the system may think that the system
`
`is in a different state due to communication issues, delays, etc. Single machine
`
`approaches do not consider these types of issues.
`
`h.
`
`Unlike in a single-machine system, in a distributed system,
`
`geographic, network, and placement location matters a great deal. For example, if
`
`two parts of the system are connected via a high-bandwidth low-latency computing
`
`link, they are likely to perform much better than if separated by a low-bandwidth,
`
`high-latency link. For computing elements that do not directly communicate often,
`
`it may be fine to use the lower quality link so that it can be reserved for portions of
`
`the system that have greater communication demands. Moreover, the geographic
`
`distance between the machines that the elements are placed on can impact the
`
`communication latency, message loss rate, and overall reliability of the system.
`
`i.
`
`Unlike in a single-machine system, release and cleanup of
`
`resources is more complex in distributed systems. For example, limited network
`
`traffic from a client may make it look like the client has finished with the resources
`
`allocated to it, but th