SanDisk Enterprise IP LLC

Patent Trials and Appeals Board

Mar 12, 2020

14597174 - (D) (P.T.A.B. Mar. 12, 2020)

UNITED STATES PATENT AND TRADEMARK OFFICE
UNITED STATES DEPARTMENT OF COMMERCE
United States Patent and Trademark Office
Address: COMMISSIONER FOR PATENTS
P.O. Box 1450
Alexandria, Virginia 22313-1450
www.uspto.gov
APPLICATION NO. FILING DATE FIRST NAMED INVENTOR ATTORNEY DOCKET NO. CONFIRMATION NO.
14/597,174 01/14/2015 Frederic H. Tudor 058752-01-5213-US 1097
82474 7590 03/12/2020
Morgan, Lewis & Bockius LLP (PH)(SanDisk)
1701 Market Street
Philadelphia, PA 19103-2921
EXAMINER
KHAN, MASUD K
ART UNIT PAPER NUMBER
2131
NOTIFICATION DATE DELIVERY MODE
03/12/2020 ELECTRONIC
Please find below and/or attached an Office communication concerning this application or proceeding.
The time period for reply, if any, is set in the attached communication.
Notice of the Office communication was sent electronically on above-indicated "Notification Date" to the
following e-mail address(es):
judith.troilo@morganlewis.com
phpatentcorrespondence@morganlewis.com
PTOL-90A (Rev. 04/07)
UNITED STATES PATENT AND TRADEMARK OFFICE
________________

BEFORE THE PATENT TRIAL AND APPEAL BOARD
________________

Ex parte FREDERIC H. TUDOR, HARIHARA KADAYAM,
BRIAN W. O’KRAFKA, and JOHANN GEORGE
________________

Appeal 2019-001307
Application 14/597,174
Technology Center 2100
________________

Before JOHNNY A. KUMAR, SCOTT B. HOWARD, and
MICHAEL T. CYGAN, Administrative Patent Judges.

KUMAR, Administrative Patent Judge.

DECISION ON APPEAL

Appeal 2019-001307
Application 14/597,174

2
STATEMENT OF CASE
Introduction
Pursuant to 35 U.S.C. § 134(a), Appellant1 appeals the Final Rejection
of claims 1–25. We have jurisdiction under 35 U.S.C. § 6(b).
We affirm.

Exemplary Claim
Exemplary claim 1 under appeal reads as follows:
1. A method of managing a data storage system that includes: a
memory controller with one or more processors; a non-volatile
memory with a datastore and a log stream; and a key-map,
wherein the key-map stores datastore location information for a
plurality of keys, the plurality of keys corresponding to data
objects in one or more tiered data structures stored in the
datastore; the method comprising:

at the memory controller operatively coupled with the
non-volatile memory and the key-map:
in response to receiving a request, from a requestor, to
perform a transaction including a set of two or more memory
operations to be performed on the one or more tiered data
structures, wherein a first memory operation in the set of two or
more memory operations is associated with a first key:

writing a start transaction record to the log
stream including a unique transaction identifier corresponding
to the transaction; and
performing the two or more memory
operations, including:
for each of the memory operations, upon
completion of the memory operation, writing an operation

1 We use the word “Appellant” to refer to “applicant” as defined in
37 C.F.R. § 1.42. According to Appellant, SanDisk Technologies LLC is the
real party in interest. Appeal Br. 3.
Appeal 2019-001307
Application 14/597,174

3
commit record to the log stream, wherein the operation commit
record guarantees a memory operation level of atomicity across
the one or more tiered data structures; and
for the first memory operation in the set of
two or more memory operations:

writing a new data object in the
datastore;
assigning, in the key-map, a location
of the new data object in the datastore to the first key; and
maintaining an old data object in the
datastore, wherein a location of the old data object was
previously assigned, in the key-map, to the first key;
wherein the operation commit record
for the first memory operation includes the first key, the unique
transaction identifier, a pointer to the location of the new data
object in the datastore, and a pointer to the location of the old
data object in the datastore; and

in accordance with a determination that the transaction,
including the set of two or more memory operations in the
transaction, is complete, writing a transaction commit record to
the log stream including the unique transaction identifier
corresponding to the transaction, wherein the transaction
commit record guarantees a transaction level of atomicity
across the data storage system.
See Appeal Br. 21 (emphases added).

REJECTIONS
Claims 1–13, 21–23, and 25 stand rejected under 35 U.S.C. § 103 as
being unpatentable over Adl-Tabatabai et al. (US 2011/0145512 A1; June
16, 2011) (“Adl-Tabatabai”), Marathe et al. (US 2011/0252067 A1; October
13, 2011) (“Marathe”), and Lomet et al. (US 2012/0179645 A1; July 12,
2012) (“Lomet”). Final Act. 3; Ans. 3.
Appeal 2019-001307
Application 14/597,174

4
Claims 14–20 and 24 stand rejected under 35 U.S.C. § 103 as being
unpatentable over Adl-Tabatabai, Marathe, Walsh et al. (US 9,170,938 B1;
October 27, 2015) (“Walsh”), and Lomet. Final Act. 26; Ans. 3.

ANALYSIS
We have only considered those arguments that Appellant actually
raised in the Briefs.2 Arguments Appellant could have made, but chose not
to make, in the Briefs have not been considered and are deemed to be
waived. See 37 C.F.R. § 41.37(c)(1)(iv) (2019).

A. Whether the Examiner erred in finding that Lomet teaches “writing a
start transaction record to the log stream”
The Examiner finds that Lomet teaches, inter alia, a transaction
component that writes an operation, its log sequence number, and
information from a reply message to its log. Final Act. 7 (citing Lomet at
Abstract). The Examiner finds, in particular, that “every operation of a
transaction is associated with a new log sequence number.” Ans. 5.
The Appellant contends that the writing a transaction record to the log
stream of Lomet, contrary to the Examiner’s finding, is not a “start”
transaction. Appeal Br. 13–14. The step of writing to the log, Appellant
argues, occurs after completion of the operations and after completion of the
transaction. Id. at 14; Reply Br. 5. More specifically, Appellant asserts,
Lomet at figure 3 “teaches a ‘write operation and log sequence number to

2 Claims 2 and 7–24 are not argued separately from claim 1 in either of
Appellant’s briefs (Appeal Br. 12–20; Reply Br. 5–11), and will not be
separately addressed.
Appeal 2019-001307
Application 14/597,174

5
log’ at step 334 . . . that occur[s] after completion of the transaction.” Reply
Br. 5.
Our reviewing court has held that “[t]here is a ‘heavy presumption’
that the terms used in claims ‘mean what they say and have the ordinary
meaning that would be attributed to those words by persons skilled in the
relevant art.’” SuperGuide Corp. v. DirecTV Enters., Inc., 358 F.3d 870,
874–75 (Fed. Cir. 2004) (citing Tex. Digital Sys., Inc. v. Telegenix, Inc., 308
F.3d 1193, 1202 (Fed. Cir. 2002)). Although claim terms are interpreted in
light of the specification, we do not read limitations from the specification
into the claims. See Constant v. Advanced Micro-Devices, Inc., 848 F.2d
1560, 1571–72 (Fed. Cir. 1988) (various limitations on which appellant
relied were not stated in the claims; the specification did not provide
evidence indicating these limitations must be read into the claims to give
meaning to the disputed terms); see also In re Am. Acad. Of Sci. Tech Ctr.,
367 F.3d 1359, 1364 (Fed. Cir. 2004). In addition, proper claim construction
requires a broadest reasonable interpretation consistent with the
specification. In re Bond, 910 F.2d 831, 833 (Fed. Cir. 1990); see also
Phillips v. AWH Corp., 415 F.3d 1303, 1317 (Fed. Cir. 2005).
Per the guidance above, we find that Appellant’s argument is not
commensurate with the scope of claim 1. Applying a broad but reasonable
interpretation of the claim language in view of Appellant’s Specification, we
disagree with the Appellant that the limitation at issue—“writing a start
transaction record to the log stream”—requires that the start transaction
record must be written to the log stream prior to completion of the
transaction.
Appeal 2019-001307
Application 14/597,174

6
Appellant’s Specification describes “start transaction record” as being
a record that “includ[es] a unique transaction identifier corresponding to the
transaction.” Spec. ¶ 19. A person having ordinary skill in the art at the time
of Appellant’s invention would not, under broadest reasonable
interpretation, interpret Appellant’s “start transaction record” as being
required to be written either at the start or prior to completion of the
transaction.

Moreover, Lomet teaches a step 322 of “determine log sequence
number,” which occurs for each operation of a transaction. See Lomet fig. 3.
Lomet teaches that in this step 322, “[a] log sequence number may be
determined in block 322 for the transaction.” Lomet ¶ 67. The Examiner
indicates that this issuance of a “log sequence number” is the start
transaction record of Lomet. Ans. 5-6. According to figure 3 of Lomet, step
322 occurs at the start of the transaction. See Lomet, fig. 3. Therefore, even
were we to construe the language of claim 1 to require the transaction record
to be at the start of the transaction, we would find such a teaching in Lomet.

B. Whether Lomet teaches “a unique transaction identifier
corresponding to the transaction” that is included in the “start
transaction record”
The Examiner finds that Lomet teaches “a unique transaction
identifier corresponding to the transaction,” in the form of a new log
sequence number. Final Act. 7 (emphasis omitted). The Examiner explains:
[w]hen a new transaction starts, according to ¶0079, it starts
with a new sequence number starting after the end of previous
transaction. For example, if first transaction uses sequence
Appeal 2019-001307
Application 14/597,174

7
numbers 1-5, the second transaction will start with sequence
number 6 and so on. Therefore, when a transaction ends and it
writes its last sequence number which is unique compared to
subsequent transactions’ sequence number to be written in the
log.
Id.
Appellant disagrees with the Examiner’s finding, arguing that the “log
sequence number” of Lomet is not a “unique transaction identifier
corresponding to the transaction,” since a “log sequence number” is
determined for each operation, rather than for each transaction. Appeal Br.
14–15. The Appellant points to the Abstract of Lomet, and argues: “The
transaction component may manage a transaction using a log sequence
number for each operation. . . . The transaction component may then write
the operation, its log sequence number, and information from the reply
message to its log.” Id. (emphases omitted); see also Reply Br. 6–8.
Similar to Appellant’s arguments presented supra, at section A, we
find that the Appellant’s arguments are not commensurate with the scope of
the claims.
A broadest reasonable interpretation of “a unique transaction identifier
corresponding to the transaction” does not preclude multiple unique
transaction identifiers corresponding to each transaction. The Examiner,
citing to the Abstract and paragraph 12 of Lomet for teaching of
“monotonically increasing log sequence [number]” for each operation, finds
that
if a transaction had 5 operations, the log sequence numbers for
those 5 operations can be 10, 11, 12, 13[,] and 14. Consider, the
next transaction have [sic] three operations. Log sequence
numbers of those operations will be 15, 16[,] and 17. So, we see
Appeal 2019-001307
Application 14/597,174

8
First transaction starts with a log sequence number 10 and the
second transaction starts with a log sequence number 15.
Ans. 5 (emphasis added). We agree with the Examiner that each transaction
has a unique transaction identifier. Nor does the Appellant apprise us of any
Examiner error in finding that the unique transaction identifier is included in
the “start transaction record.”
Accordingly, we sustain the Examiner’s obviousness rejections of
claims 1, 2, and 7–24.

C. Whether Marathe teaches “allocating a first slab” of a predefined
size3 to write the new data4, per claim 3
The Examiner finds that Marathe teaches, inter alia, “each slab…has
a respective size, of a set of predefined sizes,” because Marathe teaches
detecting a sector size of a persistent storage, and ensuring that each node “is
not larger than the sector size of the persistent storage.” Final Act. 9 (citing
Marathe ¶ 3). The Examiner further finds that Marathe teaches the limitation
“prior to writing the new data object in the datastore, allocating the first slab
in the datastore for the new data object,” because Marathe teaches “[a]t
operation 408, the sector containing the newly inserted entry can be flushed
to persistent storage.” Final Act. 10 (emphasis omitted, citing Marathe ¶ 48).

3 A limitation “each slab of the plurality of slabs has a respective size, of a
set of predefined sizes,” appears in claim 2 (Appeal Br. 22), from which
claim 3 depends. Id.
4 Claims 4–6 are not argued separately from claim 3 in either of Appellant’s
briefs (Appeal Br. 12–20; Reply Br. 5–11), and will not be separately
addressed.
Appeal 2019-001307
Application 14/597,174

9
The Examiner finds, in particular, that memory must be allocated to flush
data in a storage. Id.
Appellant argues “Marathe merely teaches ‘receiving a key . . . [and]
searching a B+ data structure using the key to find a leaf node.” Appeal Br.
17 (alteration in original, no citation to Marathe provided). Appellant
further argues that Marathe teaches only that each respective B+ tree node
may have a smaller or similar size as the sector size, and thus Marathe does
not teach whether each B+ tree node is of a predefined size. Reply Br. 9.
Furthermore, the Appellant asserts that the Examiner’s finding regarding
flushing a newly inserted entry is inconsistent with the Examiner’s finding
that the nodes of Marathe are the claimed slabs; and that the flushing to
persistent storage occurs after new data has been written in the node. Appeal
Br. 18. Appellant additionally disputes the Examiner’s finding that memory
must be allocated to flush data in a storage. Id.

1. Whether Marathe teaches “each slab . . . has a respective size, of a set
of predefined sizes”
We disagree with Appellant’s argument that Marathe does not teach
whether each B+ tree node is of a predetermined size.
In addition to the roadmap discussed above, at § A, our reviewing
courts have held that a broadest reasonable interpretation of the claim
language may involve extrinsic evidence such as “the appropriate use of
dictionaries” for “assist[ing] in understanding the commonly understood
meaning of words.” Phillips, 415 F.3d at 1322 (citing Exhibit Supply Co. v.
Ace Patents Corp., 315 U.S. 126, 134 (1942) (relying on dictionaries to
construe the claim term “embedded”); Weber Elec. Co. v. E.H. Freeman
Appeal 2019-001307
Application 14/597,174

10
Elec. Co., 256 U.S. 668, 678 (1921) (approving circuit court’s use of
dictionary definitions to define claim terms); Renishaw PLC v. Marposs
Societa’ per Azioni, 158 F.3d 1243, 1247–53 (Fed. Cir. 1998) (approving the
use of dictionaries with proper respect for the role of intrinsic evidence)).
The Phillips court held that, unless intrinsic evidence such as the
specification or prosecution history provides otherwise, or the claim
language is a term of art, dictionary use to define claim terms is permissible.
Phillips, 415 F.3d at 1321–22.
Here, since Appellant’s Specification does not provide a special
definition for “predetermined,” and the term is not one of art, we turn to
dictionaries to provide a definition for “predetermined.” Merriam-Webster
Dictionary defines “predetermine” as “to determine beforehand” or “to
impose a direction or tendency on beforehand.” Predetermine, MERRIAM-
WEBSTER, https://www.merriam-webster.com/dictionary/predetermine (last
visited Mar. 9, 2020). The Cambridge English Dictionary provides a similar
“decided or arranged at an earlier time” as the definition of “predetermined.”
Predetermined, CAMBRIDGE DICTIONARY, https://dictionary.cambridge.org/us/
dictionary/english/predetermined (last visited Mar. 9, 2020).
As such, we find that the cited portions of Marathe, on which the
Examiner relies, teach “each slab of the plurality of slabs has a respective
size, of a set of predefined sizes,” per the language of claim 2. Appeal Br.
22. Specifically, we agree with the Examiner that Marathe’s teaching that
each node “is not larger than (or is the same size as) the sector size of the
persistent storage” (Ans. 6 (emphasis omitted, citing Marathe ¶ 63)) is a
teaching for a set of sizes that are determined beforehand, i.e.
“predetermined.”
Appeal 2019-001307
Application 14/597,174

11

2. Whether Marathe teaches “prior to writing the new data object in the
datastore, allocating the first slab . . . for the new data object”
We are not persuaded by Appellant’s arguments that the Examiner
erred in finding that Marathe teaches the above limitations.
In response to the Appellant’s first argument that the Examiner’s
discussion “concerning flushing a newly inserted entry to persistent storage
. . . is not consistent with the Examiner’s assertion on page 9 of the [final]
office action that Marathe’s nodes are the equivalent of slabs” (Appeal
Br. 18), the Examiner elucidates:
Examiner refers to ¶0048 of Marathe: “At operation 408, the
sector containing the newly inserted entry can be flushed to
persistent storage.”, and ¶0052 “For example in an
embodiment, a flush to disk of the free space management
structure can be initiated after allocation of eight blocks”
Each sector containing a node of the B+ tree is synonymous to
a slab of predefined size. Those sectors are flashed [sic] in the
persistent storage/disk.
Ans. 6. Appellant’s Reply Brief does not repeat the assertion that the
flushing of a newly inserted entry is inconsistent with the Examiner’s
finding that the nodes of Marathe are analogous to the claimed slabs.
We do not find inconsistency with the Examiner’s determination that
Marathe teaches each sector containing a node, and the sector can be flushed
to persistent storage, based on the support the Examiner provides for
Marathe teaching each sector, which contains a node, being flushed to
persistent storage. Id.
Regarding Appellant’s additional arguments: the Examiner’s findings
that “[t]o flush the data in storage, memory need [sic] to be allocated” (Final
Appeal 2019-001307
Application 14/597,174

12
Act. 10); and “allocating the first slab” occurs “prior to writing the new data
object” each amount to an assertion of inherency, without providing
sufficient support (Appeal Br. 18 (emphasis omitted)), we disagree for
reasons explained below.
The Examiner cites to Marathe paragraph 52 for teaching “that
flushing of data in disk requires allocation of free disk space.” Ans. 6.
Moreover, Marathe at paragraph 52 teaches the claimed allocating occurring
prior to the claimed writing. Id. Specifically, we agree with the Examiner
that “a flush to disk . . . can be initiated after allocation” (Marathe ¶ 52) is a
teaching of the claim limitation “prior to writing the new data object in the
datastore, allocating the first slab . . . for the new data object.” Appeal Br.
22. In other words, Marathe teaches allocating the first slab, then a flush to
disk (Marathe ¶ 52), and finally writing the new data object in the datastore.
Id. and Marathe ¶ 48 (“[i]n this fashion a new entry can be inserted into a
B+ tree data structure” (emphasis added)).
Therefore, we sustain the Examiner’s obviousness rejections of claims
3–6.

Appeal 2019-001307
Application 14/597,174

13
D. Whether Lomet teaches “identifying an incomplete transaction,”
wherein “the incomplete transaction includes a start transaction
record having a unique transaction identifier in the log stream,” and
“performing a recovery process for the incomplete transaction,
wherein . . . [the] incomplete transaction . . . includes a start
transaction record having a lowest sequence number in the log
stream”5
The Examiner finds that Lomet, at paragraph 56, teaches the
limitations of claim 25. Final Act. 25–26.
Appellant argues that Lomet does not teach a “start transaction
commit record,” and thus cannot teach either “the incomplete transaction
includes a start transaction record having a unique transaction identifier in
the log stream,” or “a start transaction commit record . . . without a
corresponding transaction commit record,” per the language of claim 25.
Appeal Br. 20. The Appellant further argues that, since Lomet lacks a start
transaction commit record without a corresponding transaction commit
record, Lomet does not teach “identifying an incomplete transaction” and
“performing a recovery process” based on the start transaction commit
record as claimed. Id. Appellant further argues that, contrary to the claimed
features, Lomet “merely performs a recovery from a ‘redo scan start point,’
to redo ‘all of the operations with log sequence numbers greater than the
redo scan start point.’” Reply Br. 11 (citing Lomet ¶ 86); see also Appeal
Br. 20.

5 This limitation appears in claim 25. Appeal Br. 30.
Appeal 2019-001307
Application 14/597,174

14
Appellant’s arguments challenging the Examiner’s finding that Lomet
teaches “a start transaction record having a unique transaction identifier in
the log stream” have already been discussed, supra, at section A, and will
not be further addressed.
In response to the Appellant’s argument that Lomet does not teach
“identifying an incomplete transaction” and “performing a recovery
process,” the Examiner finds:
Fig. 5 of Lomet . . . clearly shows the recovery form [sic] an
erroneous (i.e., incomplete) transection [sic] using the log
sequence number. ¶0086 to ¶0088 explains the steps of the
recovery from an incomplete transaction. Examiner cites from
¶0086 to clarify the recovery [redo] operation: “Embodiment
500 illustrates a method that may be performed to recover from
an error with a transaction component. The error may cause the
transaction component to be restarted, which may lose any
information that has not been saved. In order to effectively
resume, a redo scan start point may be used to begin recovery
from a log. . . . So, when an error occurs in a transaction, to
recover from the error, a redo scan is done thought [sic] the
log to find out the start of the transaction and those incomplete
operations are re-sent.
Ans. 7 (original emphases removed; emphasis added).
We agree with the Examiner that figure 5 and the cited portions of
Lomet teach the claim language at issue. Specifically, Lomet at step 502
teaches “identifying an incomplete transaction in the log stream.” See Lomet
at figure 5; Ans. 7. Lomet at paragraphs 56 and 86 teaches “wherein the
incomplete transaction . . . [is] without a corresponding transaction commit
record”; more specifically, we agree with the Examiner that Lomet teaches
that the incomplete transaction has not been added to the “stable log.” Final
Act. 25–26. We further agree with the Examiner’s finding that Lomet shows
Appeal 2019-001307
Application 14/597,174

15
“performing a recovery process” in at least figure 5. Ans. 7. Further, we
agree with the Examiner that Lomet teaches “wherein the earliest incomplete
transaction . . . includes a start transaction record having a lowest sequence
number in the log stream,” at least in part because the Examiner finds that
Lomet teaches “monotonically increasing log sequence numbers.” Lomet
¶ 12; Final Act. 25–26; Ans. 6–7. Therefore, upon detection of an error,
Lomet teaches a redo to a previous, lower-number log sequence number.
Lomet ¶ 12; see also Lomet fig. 5.
Accordingly, we sustain the Examiner’s obviousness rejection of
claim 25.

Appeal 2019-001307
Application 14/597,174

16
DECISION
For at least the reasons described supra:
We affirm the Examiner’s decision rejecting claims 1–13, 21–23, and
25 under 35 U.S.C. § 103 as being unpatentable over Adl-Tabatabai,
Marathe, and Lomet.
We affirm the Examiner’s decision rejecting claims 14–20 and 24
under 35 U.S.C. § 103 as being unpatentable over Adl-Tabatabai, Marathe,
Walsh, and Lomet.

CONCLUSION

No time period for taking any subsequent action in connection with
this appeal may be extended under 37 C.F.R. § 1.136(a)(1)(iv). See
37 C.F.R. § 1.136(a)(1)(iv).

AFFIRMED
Claim(s)
Rejected
35
U.S.C. §
Reference(s)/Basis Affirmed Reversed
1–13, 21–
23, 25
103 Adl-Tabatabai,
Marathe, Lomet
1–13, 21–
23, 25

14–20, 24 103 Adl-Tabatabai,
Marathe, Walsh,
Lomet
14–20, 24
Overall
Outcome
1–25