`Case 5:18—md-02834-BLF Document 592-12 Filed 03/20/20 Page 1 of 60
`
`
`
`
`
`
`EXHIBIT 11
`
`EXHIBIT 11
`
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 2 of 60
`
`Trials@uspto.gov
`571-272-7822
`
`
`
`
`
`
`
`
`
`
`
` Paper 64
`
`
`
` Entered: May 15, 2014
`
`
`UNITED STATES PATENT AND TRADEMARK OFFICE
`____________
`
`BEFORE THE PATENT TRIAL AND APPEAL BOARD
`____________
`
`EMC CORPORATION,
`Petitioner,
`
`v.
`
`PERSONALWEB TECHNOLOGIES, LLC and
`LEVEL 3 COMMUNICATIONS, LLC,
`Patent Owners.
`____________
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`____________
`
`
`
`
`Before KEVIN F. TURNER, JONI Y. CHANG, and
`MICHAEL R. ZECHER, Administrative Patent Judges.
`
`
`CHANG, Administrative Patent Judge.
`
`
`
`
`
`
`
`FINAL WRITTEN DECISION
`35 U.S.C. § 318(a) and 37 C.F.R. § 42.73
`
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 3 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`I.
`
`INTRODUCTION
`
`EMC Corporation (“EMC”) filed a petition on December 16, 2012,
`
`requesting an inter partes review of claim 1 of U.S. Patent No. 7,945,544 B2
`
`(“the ’544 patent”). Paper 3 (“Pet.”). PersonalWeb Technologies, LLC and
`
`Level 3 Communications, LLC (collectively, “PersonalWeb”) filed a patent
`
`owner preliminary response. Paper 9 (“Prelim. Resp.”). Taking into
`
`account the patent owner preliminary response, the Board determined that
`
`the information presented in the petition demonstrated that there was a
`
`reasonable likelihood that EMC would prevail with respect to claim 1.
`
`Pursuant to 35 U.S.C. § 314, the Board instituted this trial on May 17, 2013,
`
`as to claim 1 of the ’544 patent. Paper 14 (“Dec.”).
`
`After institution, PersonalWeb filed a patent owner response
`
`(Paper 33 (“PO Resp.”)), and EMC filed a reply to the patent owner
`
`response (Paper 40 (“Reply”)). Oral hearing was held on December 16,
`
`2013.1
`
`We have jurisdiction under 35 U.S.C. § 6(c). This final written
`
`decision is entered pursuant to 35 U.S.C. § 318(a). We hold that claim 1 of
`
`the ’544 patent is unpatentable under 35 U.S.C. §§ 102 and 103.
`
`
`
`1 This proceeding, as well as IPR2013-00082, IPR2013-00083, IPR2013-
`00085, IPR2013-00086, and IPR2013-00087, involve the same parties and
`similar issues. The oral arguments for all six inter partes reviews were
`merged and conducted at the same time. A transcript of the oral hearing is
`included in the record as Paper 63.
`
`2
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 4 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`A. Related Proceeding
`
`EMC indicates that the ’544 patent is the subject of litigation titled
`
`PersonalWeb Technologies LLC v. EMC Corporation and VMware, Inc.,
`
`No. 6:11-cv-00660-LED (E.D. Tex.). Pet. 1.
`
`B. The ’544 patent
`
`The ’544 patent relates to a method for identifying a data item
`
`(e.g., a data file or record) in a data processing system, by using an identifier
`
`that depends on all of the data in the data item and only on the data in the
`
`data item. Ex. 1001, 1:45-49; 3:53-56. Thus, the identity of a data item is
`
`said to be independent of its name, origin, location, and address. Id. at
`
`3:56-59. According to the ’544 patent, it is desirable to have a mechanism
`
`for identifying identical data items to reduce duplicate copies of a data item.
`
`Id. at 3:37-40. Figure 10(b) of the ’544 patent, reproduced below, is a flow
`
`chart for determining an identifier of a simple or compound data item.
`
`3
`
`
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 5 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`As shown in Figure 10(b) of the ’544 patent, for a simple data item
`
`(a data item whose size is less than a particular given size) (S216 and S218),
`
`a data identifier (True Name) is computed using a function (e.g., a message
`
`digest (“MD”) function, such as MD4 or MD5, or a secure hash algorithm
`
`(“SHA”) function). Id. at 12:18-49, 13:31-42; figs. 10(a) & 10(b). As a
`
`result, a data item that has an arbitrary length is reduced to a relatively small,
`
`fixed size identifier (True Name) that represents the data item. Id.
`
`If the data item is a compound data item (a data item whose size is
`
`greater than the particular given size), the system will partition the data item
`
`into segments (S220); assimilate each segment (S222); compute the True
`
`Name of the segment; create an indirect block consisting of the computed
`
`segment True Names (S224); assimilate the indirect block (S226); and
`
`replace the final 32 bits of the resulting True Name by the length modulo 32
`
`of the compound data item (S228). Id. at 13:43-61, fig. 10(b). The result is
`
`the True Name of the compound data item. Id.
`
`Figure 11 of the ’544 patent is reproduced below:
`
`4
`
`
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 6 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`Figure 11 of the ’544 patent depicts a mechanism for assimilating a
`
`data item into a file system. The purpose of this mechanism is to add a
`
`given data item to the True File registry. Id. at 14:4-11. If the data item
`
`already exists in the registry, the duplicate will be eliminated. Id.
`
`To assimilate a data item, the system will determine the True Name of
`
`the data item corresponding to the file (S230); look for an entry for the True
`
`Name in the True File Registry (S232); and determine whether a True Name
`
`entry exists in the True File Registry (S232). Id. at 14:4-27, fig. 11. If the
`
`entry record includes a corresponding True File ID (Step S237), the system
`
`will delete the file (Step S238). Otherwise the system will store the True
`
`File ID in the entry record (S239). Id. If there is no entry in the True File
`
`Registry for the True Name (S232), the system will create a new entry in the
`
`True File Registry for the True Name (S236). Id.
`
`C. Challenged Claim
`
`According to EMC, claim 1 essentially requires obtaining “values” for
`
`two data items, and then comparing these values to ascertain whether the
`
`two data items correspond to each other (e.g., whether they are the same).
`
`Pet. 16. Claim 1 recites the following:
`
`1. A computer-implemented method,
`
`comprising:
`
`the method
`
`(A) for a first data item comprising a first plurality of parts,
`
`(a1) applying a first function to each part of said first
`plurality of parts to obtain a corresponding part value for each
`part of said first plurality of parts,
`
`5
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 7 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`wherein each part of said first plurality of parts comprises
`a corresponding sequence of bits, and
`
`wherein the part value for each particular part of said first
`plurality of parts is based, at least in part, on the corresponding
`bits in the particular part, and
`
`wherein two identical parts will have the same part value
`as determined using said first function,
`
`wherein said first function comprises a first hash
`function; and
`
`(a2) obtaining a first value for the first data item, said
`
`first value obtained by applying a second function to the part
`values of said first plurality of parts of said first data item, said
`second function comprising a second hash function;
`
`(B) for a second data item comprising a second plurality of parts,
`
`(b1) applying said first function to each part of said
`second plurality of parts to obtain a corresponding part value
`for each part of said second plurality of parts,
`
`wherein each part of said second plurality of parts
`consists of a corresponding sequence of bits, and
`
`wherein the part value for each particular part of said
`second plurality of parts is based, at least in part, on the
`corresponding bits in the particular part of the second plurality
`of parts; and
`
`(b2) obtaining a second value for the second data item by
`applying said second function to the part values of said second
`plurality of parts of said second data item; and
`
`(C) ascertaining whether or not said first data item corresponds
`to said second data item based, at least in part, on said first
`value and said second value.
`
`Ex. 1001, 38:34-39:3 (emphases and indentions added).
`
`6
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 8 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`
`
`EMC relies upon the following prior art references:
`
`D. Prior Art Relied Upon
`
`July 15, 1997
`
`Woodhill US 5,649,1962
`
`Frederick W. Kantor, “FWKCS (TM) Contents-Signature System
`Version 1.22,” FWKCS122.REF (Aug. 10, 1993) (“Kantor,”
`Ex. 1004)
`
`
`(Ex. 1005)
`
`E. Grounds of Unpatentability
`
`The Board instituted the instant trial based on the following grounds
`
`of unpatentability:
`
`Claim
`
`Basis
`
`References
`
`1
`
`1
`
`1
`
`§ 102(e)
`
`Woodhill
`
`§ 102(b)
`
`Kantor
`
`§ 103(a)
`
`Kantor and Woodhill
`
`II. ANALYSIS
`
`A. Claim Construction
`
`We begin our analysis by determining the meaning of the claims.
`
`In an inter partes review, claim terms in an unexpired patent are given their
`
`broadest reasonable construction in light of the specification of the patent in
`
`which they appear. 37 C.F.R. § 42.100(b). Under the broadest reasonable
`
`construction standard, claim terms are given their ordinary and customary
`
`
`
`2 Woodhill claims the benefit of U.S. Patent Application No. 08/085,596,
`which was filed on July 1, 1993.
`
`7
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 9 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`meaning as would be understood by one of ordinary skill in the art in the
`
`context of the entire disclosure. In re Translogic Tech. Inc., 504 F.3d 1249,
`
`1257 (Fed. Cir. 2007). An inventor may rebut that presumption by
`
`providing a definition of the term in the specification with reasonable clarity,
`
`deliberateness, and precision. In re Paulsen, 30 F.3d 1475, 1480 (Fed. Cir.
`
`1994). In the absence of such a definition, limitations are not to be read
`
`from the specification into the claims. In re Van Geuns, 988 F.2d 1181,
`
`1184 (Fed. Cir. 1993).
`
`In the Decision on Institution, we construed the claim term “data
`
`item” to mean “sequence of bits,” and observed that in the context of the
`
`specification, the meaning also includes one of the following: (1) the
`
`contents of a file; (2) a portion of a file; (3) a page in memory; (4) an object
`
`in an object-oriented program; (5) a digital message; (6) a digital scanned
`
`image; (7) a part of a video or audio signal; (8) a directory; (9) a record in a
`
`database; (10) a location in memory or on a physical device or the like; and
`
`(11) any other entity which can be represented by a sequence of bits. Dec. 9.
`
`The parties agree with that claim construction. Pet. 6; PO Resp. 1. As noted
`
`in the Decision on Institution, that claim construction is consistent with the
`
`specification. Dec. 8-9 (citing Ex. 1001, 2:17-18 (“the terms ‘data’ and
`
`‘data item’ as used herein refer to sequences of bits.”); id. at 2:18-22, 27-32).
`
`We discern no reason to deviate from that claim construction for the
`
`purposes of this decision.
`
`8
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 10 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`B. Principles of Law
`
`To establish anticipation, each and every element in a claim, arranged
`
`as recited in the claim, must be found in a single prior art reference. Net
`
`MoneyIN, Inc. v. VeriSign, Inc., 545 F.3d 1359, 1369 (Fed. Cir. 2008);
`
`Karsten Mfg. Corp. v. Cleveland Golf Co., 242 F.3d 1376, 1383 (Fed. Cir.
`
`2001). We also recognize that prior art references must be “considered
`
`together with the knowledge of one of ordinary skill in the pertinent art.”
`
`Paulsen, 30 F.3d at 1480. Moreover, “it is proper to take into account not
`
`only specific teachings of the reference but also the inferences which one
`
`skilled in the art would reasonably be expected to draw therefrom.” In re
`
`Preda, 401 F.2d 825, 826 (CCPA 1968).
`
`A patent claim is unpatentable under 35 U.S.C. § 103(a) if the
`
`differences between the claimed subject matter 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 to which said
`
`subject matter pertains. KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 406
`
`(2007). The question of obviousness is resolved on the basis of underlying
`
`factual determinations, including: (1) the scope and content of the prior art;
`
`(2) any differences between the claimed subject matter and the prior art;
`
`(3) the level of skill in the art; and (4) objective evidence of nonobviousness.
`
`Graham v. John Deere Co. of Kansas City, 383 U.S. 1, 17-18 (1966).
`
`The level of ordinary skill in the art is reflected by the prior art of record.
`
`See Okajima v. Bourdeau, 261 F.3d 1350, 1355 (Fed. Cir. 2001);
`
`9
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 11 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`In re GPAC Inc., 57 F.3d 1573, 1579 (Fed. Cir. 1995); In re Oelrich,
`
`579 F.2d 86, 91 (CCPA 1978).
`
`We analyze the instituted grounds of unpatentability in accordance
`
`with the above-stated principles.
`
`C. Claim 1 – Anticipated by Woodhill
`
`EMC asserts that claim 1 is unpatentable under 35 U.S.C. § 102(e) as
`
`anticipated by Woodhill. Pet. 50-57. As support, EMC provides detailed
`
`explanations as to how each claim element, arranged as recited in the claim,
`
`is disclosed by Woodhill. Id. EMC also relies on the declaration of
`
`Dr. Douglas W. Clark. Ex. 1009 ¶¶ 43-49.
`
`PersonalWeb counters that Woodhill does not describe all of the
`
`limitations of claim 1. PO Resp. 3-15. Specifically, PersonalWeb contends
`
`that: (1) Woodhill fails to describe applying a second hash function to
`
`shadow files (id. at 5-11 (citing Ex. 2016 ¶¶ 25-35)); and (2) Woodhill does
`
`not describe binary object identifiers for the first data item and the second
`
`data item (id. at 11-15 (citing Ex. 2016 ¶¶ 36-40)). PersonalWeb also
`
`proffers a declaration of Dr. Robert B. K. Dewar. Ex. 2016 ¶¶ 20-41.
`
`Upon review of the parties’ arguments and evidence, we determine
`
`that EMC has demonstrated by a preponderance of the evidence that claim 1
`
`is unpatentable under 35 U.S.C. § 102(e) as being anticipated by Woodhill.
`
`Woodhill
`
`Woodhill discloses a system for distributed storage management on a
`
`computer network system using binary object identifiers. Ex. 1005, 1:11-17.
`
`10
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 12 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`The system includes a remote backup file server and a plurality of local area
`
`networks in communication with the remote backup file server. Id.
`
`Figure 1 of Woodhill, reproduced below, depicts a computer network
`
`system that includes a distributed storage management system:
`
`
`
`
`
`As illustrated in Figure 1 of Woodhill, remote backup file server 12
`
`communicates with wide area network 14, which communicates with a
`
`plurality of local area networks 16. Id. at 3:12-30. Each local area network
`
`16 includes multiple user workstations 18 and local computers 20. Id. at
`
`3:24-44. The storage space on each disk drive 19 on each local computer 20
`
`is allocated according to the hierarchy illustrated in Figure 2. Id. at 3:31-44.
`
`11
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 13 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`
`
`Woodhill’s system includes a Distributed Storage Manager (DSM)
`
`program for building and maintaining the file database. Id. at 3:44-49.
`
`The DSM program views a file as a collection of data streams, and divides
`
`each data stream into one or more binary objects. Id. at 4:13-23; 7:40-43;
`
`fig. 5A, item 132. Specifically, data streams represent regular data,
`
`extended attribute data, access control list data, etc. Id. at 7:44-47. If the
`
`size of the data stream is larger than the maximum binary object size, then
`
`the DSM program divides the data stream into multiple binary objects;
`
`otherwise, a single binary object represents the data stream. Id. at 4:23-30;
`
`7:47-59; fig. 5A, items 134 and 136. For each binary object being backed
`
`up, a binary object identification record is created in a file database and
`
`includes a Binary Object Identifier to identify a particular binary object
`
`uniquely. Id. at 7:60-8:1; 8:33-34.
`
`Binary object identifiers are calculated based on the contents of the
`
`data instead of from an external and arbitrary source so that the binary object
`
`identifier changes when the contents of the binary object changes. Id. at
`
`8:57-62; 8:40-42. Notably, the binary object identifier includes a binary
`
`object hash field that is calculated against the contents of the binary object
`
`taken one word (16 bits) at a time using a hash algorithm. Id. at 8:22-32.
`
`According to Woodhill, duplicate binary objects can be recognized from
`
`their identical binary object identifiers, even if the objects reside on different
`
`types of computers in a heterogeneous network. Id. at 8:62-65.
`
`For large database files on the network computer system, the DSM
`
`program utilizes a technique of subdividing the large database files into
`
`12
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 14 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`granules, and then tracks changes from the previous backup copy of the
`
`“granule” level. Id. at 14:53-65. This technique is used to reduce the
`
`amount of data that must be transmitted to the remote backup file server. Id.
`
`at 15:4-8. Figure 5G of Woodhill illustrates the “granularization” procedure
`
`and is reproduced below:
`
`
`
`As depicted in Figure 5G, if this is the first time that the binary object
`
`is being backed up using the “granularization” technique (step 402), the
`
`DSM program creates a shadow file, which contains a contents identifier for
`
`each granule in the binary object (step 404). Id. at 15:9-24. Each contents
`
`identifier includes a 32-bit hash number which is calculated against the
`
`contents of the granule. Id. at 15:24-30; Fig. 5A, step 138.
`
`13
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 15 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`Each time that the binary object is backed up, the DSM program
`
`calculates the contents identifier for each granule in the binary object, and
`
`then compares it to the contents identifier of the granule from the last time
`
`the binary object was backed up to determine if the granule has changed. Id.
`
`at 15:32-38. At step 406, the DSM program calculates a change identifier
`
`for each granule of the binary object and stores it in the shadow file for that
`
`binary object. Id. at 15:40-45.
`
`Applying a second hash function to shadow files
`
`Claim 1 requires “obtaining a first value for the first data item, said
`
`first value obtained by applying a second [hash] function to the part values
`
`of said first plurality of parts of said first data item” (i.e., “a hash of
`
`hashes”). In its petition, EMC asserts that Woodhill’s binary object
`
`identifiers for the shadow files meet this limitation. Pet. 53-56 (citing
`
`Ex. 1005, 5:62-63, 7:60-8:31; 9:6-28; 15:16-24; Ex. 1009 ¶¶ 43-49).
`
`PersonalWeb, however, argues that Woodhill’s granularization
`
`process does not disclose applying a second hash function to shadow files.
`
`PO Resp. 5, 7 (citing Ex. 2016 ¶¶ 29-35). In particular, PersonalWeb and its
`
`expert assert that “Binary Object identifiers 74 are not mentioned in
`
`connection with Woodhill’s ‘granularization’ procedure, and are not used
`
`therein.” PO Resp. 8 (citing Ex. 2016 ¶ 31). PersonalWeb also maintains
`
`that EMC’s reliance on Woodhill’s statement that “the default operation is to
`
`back up all files on all disk drives 19 on the local computer 20” (Ex. 1005,
`
`5:62-63) is incorrect because “Woodhill never describes shadow files as
`
`being stored on disk drives 19 of local computers 20.” Id. at 10 (citing
`14
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 16 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`Ex. 1005, 15:4-9; Ex. 2016 ¶ 34). Additionally, PersonalWeb, citing to its
`
`expert testimony, alleges that a binary object identifier is not created for a
`
`shadow file, because the granularization process, in which the shadow files
`
`are created, is not used for backing up copies of binary objects for storage on
`
`local computers. Id. at 8-9 (citing Ex. 2016 ¶¶ 31-35). PersonalWeb further
`
`contends that a shadow file will not be backed up by the DMS program, as a
`
`shadow file does not meet Woodhill’s definition of a “file” that requires at
`
`least two data streams. Id. at 11 (citing Ex. 1005, 4:14-15; Ex. 2016 ¶ 35).
`
`In its reply, EMC responds that Woodhill discloses “the application of
`
`a hash to the ‘contents identifiers’ in a shadow file.” Reply 1, n.1.
`
`Specifically, EMC alleges that Woodhill discloses calculating a binary
`
`object identifier for each shadow file when the DSM program backs up the
`
`file. Id. at 2. EMC also submits that the shadow file’s binary object
`
`identifier is for the associated underlying file or binary object. Id. at 6-7.
`
`We agree with EMC.
`
`
`
`PersonalWeb and its expert testimony narrowly focus on Woodhill’s
`
`granularization procedure. Notably, Woodhill specifically states that each of
`
`the functions performed by the DSM program operates in cooperation with
`
`the other functions to form a unitary computer program. Ex. 1005, 4:62-
`
`5:2; figs. 5a-5l. The disclosure of Woodhill merely divides the DSM
`
`program into several distinct functions for explanation purposes. Id.
`
`We agree with EMC that Woodhill’s “default operation is to back up
`
`all files on all disk drives 19 on the local computer 20” and each shadow file,
`
`like all files stored on disk drives 19, is divided into one or more binary
`
`15
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 17 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`objects to be backed up. Pet. 53 (citing Ex. 1005, 5:62-63); 55 (citing
`
`Ex. 1009 ¶¶ 46-48; Ex. 1005, 4:13-34; 5:61-63). As noted by EMC, in the
`
`process of backing up shadow files, Woodhill would obtain a first value by
`
`calculating a binary object identifier (i.e., applying a second hash function)
`
`for each shadow file binary object (i.e., the part values – the first hash).
`
`Pet. 55-56 (citing Ex. 1009 ¶¶ 45-48; Ex. 1005, 7:60-8:31; 15:16-24).
`
`EMC’s expert, Dr. Clark, testifies:
`
`the
`to backing up a binary object using
` Prior
`46.
`granularization technique for the first time, the local computer
`storing the binary object creates a “shadow file” containing the
`granule contents identifiers for each granule of that binary
`object. (Id. at col. 15, ll. 16-24; Ex 1005.) Woodhill also
`discloses claim portions [1c] and [1e]3 through his process of
`creating shadow files on local computers to store the latest
`granule contents identifiers for granularized binary objects,
`and then backup these shadow files. In particular, a shadow
`file, including each contents identifier for each granule of a
`binary object, like any file will be divided into one or more
`Binary Objects. In some cases, due to the concise nature of a
`shadow file, a shadow file may be backed up using a single
`binary object.
`
`47. As I have illustrated, each shadow file binary object, like
`all binary objects, has a corresponding Binary Object
`Identifier. Further, each Binary Object Identifier includes a hash
`of the contents of the Binary Object. Consequently, a Binary
`Object Identifier for a shadow file binary object satisfies these
`claim elements because it is a hash (second function) of the
`
`
`
`3 “Claim portions [1c] and [1e]” refer to steps (a2) and (b2) of claim 1.
`Ex. 1009 ¶ 16.
`
`16
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 18 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`contents identifiers, or granule hashes (i.e., “part values” of
`the plurality of parts [granules]).
`
`Ex. 1009 ¶¶ 46-47 (emphases added).
`
`Upon reviewing the evidence on record, we credit the testimony of
`
`Dr. Clark over that of Dr. Dewar. See Yorkey v. Diab, 601 F.3d 1279, 1284
`
`(Fed. Cir. 2010) (holding that Board has discretion to give more weight to
`
`one item of evidence over another “unless no reasonable trier of fact could
`
`have done so”). We find that Dr. Clark’s explanations are consistent with
`
`Woodhill. See, e.g., Ex. 1005, 4:13-34; 4:62-5:2; 5:61-63; 7:60-8:31;
`
`15:16-24; figs. 5a-5l. On the other hand, Dr. Dewar’s testimony (Ex. 2016
`
`¶ 34) that shadow files are not stored on the local computers contradicts the
`
`disclosure of Woodhill that shadow files are created by the DSM program
`
`and stored on the disk drives of the local computers. See, e.g., Ex. 1005,
`
`15:21-24 (The DSM program “creates a ‘shadow file’ which contains a
`
`‘contents identifier’ for each ‘granule’ in the binary object.”); 5:6-9 (The
`
`DSM program “operates in the same fashion on each local computer 20 on
`
`the network computer system 10.”); 5:7-9; fig. 2, item 24 (The DSM
`
`program resides on each disk drive 19 on each local computer 20.); 3:35-49;
`
`fig. 3 (The DSM program builds and maintains file database 25, which
`
`includes file identification record 34 and binary object identifier 74, on one
`
`of disk drives 19 on each local computer 20.); 14:62-65; 15:4-6 (The DSM
`
`utilizes the granularization procedure to subdivide large databases files into
`
`granules and then tracks changes from the previous backup copy at the
`
`17
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 19 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`granule level to reduce the amount of data that are being transmitted from
`
`the local computer to the remote backup file server.).
`
`To substantiate its position that shadow files are not stored on disk
`
`drives 19 on local computers 20, PersonalWeb also relies on Woodhill’s
`
`statement that the granularization “technique of subdividing files into
`
`‘granules’ . . . is not utilized in making backup copies of [database file]
`
`binary objects for storage on local computers.” PO Resp. 10 (citing Ex.
`
`1005, 15:4-9). However, such reliance is misplaced. As EMC notes,
`
`reading Woodhill’s statement in context, the statement merely confirms that,
`
`when backing up large database files using the granularization procedure,
`
`the system sends the backup copies of the database files to a remote server.
`
`Reply 3; see also Ex. 1005, 14:59-61 (“As a result, in most cases, the entire
`
`‘large’ database file would have to be backed up to the remote backup file
`
`server 12.”). PersonalWeb does not point out where the DSM program
`
`would execute the granularization procedure to create the shadow files. Nor
`
`does it explain sufficiently why the DSM program would not be executing
`
`the granularization procedure on the local computer. Given the disclosures
`
`of Woodhill noted above, we agree with EMC that the DSM program
`
`executes the granularization procedure to create shadow files on disk drive
`
`19 of local computer 20, and not on remote backup file server 12. Reply 3.
`
`We also are not persuaded by PersonalWeb’s argument and expert
`
`testimony that Woodhill sets forth a definition of the word “file” that
`
`requires at least two data streams, and that the DMS program would not
`
`backup a shadow file to create a binary object identifier, because a shadow
`
`18
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 20 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`file does not meet that alleged definition of the word “file.” See PO Resp.
`
`11; Ex. 2016 ¶¶ 27, 35. PersonalWeb’s argument and expert testimony are
`
`not consistent with the explicit disclosure of Woodhill. In particular, they
`
`ignore the fact that Woodhill specifically uses the word “file” in the term
`
`“shadow file.” They also do not provide sufficient explanation why a
`
`shadow file cannot have more than one data stream or more than one binary
`
`object. In fact, a shadow file is consistent with Woodhill’s description of a
`
`file. See Ex. 1005, 15:21-24 (the DSM program “creates a ‘shadow file’
`
`which contains a ‘contents identifier’ for each ‘granule’ in the binary
`
`object.”); id., 4:18-19 (“[A] file may contain its normal data and may also
`
`contain extended attribute data.”); id., 2:23-24 (“data files comprised of one
`
`or more binary objects”) (Emphases added.). As EMC notes, the actual text
`
`in Woodhill that PersonalWeb relies on is not a definition of the word “file,”
`
`and does not require a file to have at least two data streams. Reply 4 (citing
`
`Ex. 1005, 4:14-15). Indeed, Woodhill does not preclude a file from having
`
`only one data stream, or only one binary object. Ex. 1005, 2:23-24 (“storing
`
`data files comprised of one or more binary objects”); 4:21-23 (The DMS
`
`program “divides each data stream into one or more binary objects.”)
`
`(Emphasis added.).
`
`For the reasons stated above, EMC has demonstrated by a
`
`preponderance of the evidence that Woodhill describes applying a second
`
`hash function to shadow files (i.e., “a hash of hashes”).
`
`19
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 21 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`Shadow file identifiers are for the first and second data items
`
`Claim 1 requires “ascertaining whether or not said first data item
`
`corresponds to said second data item based, at least in part, on said first
`
`value and said second value.” In its petition, EMC takes the position that
`
`Woodhill meets this limitation because “by comparing binary objects of
`
`successive versions of shadow files, Woodhill by extension compares the
`
`binary objects underlying those shadow files.” Pet. 56 (citing Ex. 1005,
`
`9:5-28; Ex. 1009 ¶ 49). EMC further maintains that the comparison is
`
`“based, at least in part, on said first value” (the binary object identifier
`
`corresponding to a previous version of a shadow file) and “said second
`
`value” (the binary object identifier corresponding to the current version of
`
`the shadow file). Id. at 56-57 (citing Ex. 1005, 9:5-28; Ex. 1009 ¶ 49).
`
`PersonalWeb counters that Woodhill’s shadow file binary object
`
`identifiers are not “for the first data item” or “for the second data item.”
`
`PO Resp. 11-15 (citing Ex. 2016 ¶¶ 36-40). According to PersonalWeb,
`
`“it would be highly unlikely, if not impossible, for a single ‘shadow file’ to
`
`be separated from a data stream to form a single standalone ‘binary object,’”
`
`and that “the more likely scenario under this assumption would be that a
`
`‘binary object’ would be made up of many shadow files.” Id. at 11-12.
`
`In its reply, EMC responds that “PersonalWeb’s assumptions about
`
`Woodhill are directly contradictory to Woodhill’s explicit disclosure.”
`
`Reply 6 (citing Ex. 1005, 4:13-23; Ex. 1088 ¶¶ 14-15). We agree with
`
`EMC. Woodhill expressly discloses dividing files into one or more data
`
`streams, or one or more binary objects. Ex. 1005, 2:20-24 (“The present
`
`20
`
`
`
`Case 5:18-md-02834-BLF Document 592-12 Filed 03/20/20 Page 22 of 60
`
`Case IPR2013-00084
`Patent 7,945,544 B2
`
`invention is further directed to a method for the management of storage
`
`space . . . storing data files comprised of one or more binary objects.”);
`
`4:22-23 (The DSM program “further divides each data stream into one or
`
`more binary objects.”); 4:25-26 (A single binary object may represent a data
`
`stream.). Nothing in Woodhill suggests that a plurality of shadow files must
`
`be combined into a single binary object.
`
`We also agree with EMC that a binary object identifier for a shadow
`
`file is “a hash of hashes” for the underlying database binary object. Reply
`
`6-7. As Dr. Clark shows in his illustration (step 1), reproduced below, a
`
`binary object for a large database file (a first or second data item) is divided
`
`into a plurality of granules (a first or second plurality of parts) (Ex. 1088
`
`¶¶ 17-18; Ex. 1005, 14:53-15:16):
`
`As shown in step 2 of Dr. Clark’s illustration (Ex. 1088 ¶ 17),
`
`Woodhill’s DSM program calculates a contents identifier for each granule of
`
`the database binary object, using a hash function (first hash function), and
`
`
`
`21
`
`
`
`Case 5:18-md-02834-BLF Document 592