SAP SE

Patent Trials and Appeals Board

Feb 9, 2022

2020006790 (P.T.A.B. Feb. 9, 2022)

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. 15/213,816 07/19/2016 Ahmad Hassan 22135-0940001/150605US01 8530 32864 7590 02/09/2022 FISH & RICHARDSON, P.C. (SAP) PO BOX 1022 MINNEAPOLIS, MN 55440-1022 EXAMINER EYERS, DUSTIN D ART UNIT PAPER NUMBER 2164 NOTIFICATION DATE DELIVERY MODE 02/09/2022 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): PATDOCTC@fr.com PTOL-90A (Rev. 04/07) UNITED STATES PATENT AND TRADEMARK OFFICE ____________ BEFORE THE PATENT TRIAL AND APPEAL BOARD ____________ Ex parte AHMAD HASSAN ___________ Appeal 2020-006790 Application 15/213,816 Technology Center 2100 ____________ Before CAROLYN D. THOMAS, JAMES B. ARPIN, and MICHAEL J. ENGLE, Administrative Patent Judges. ARPIN, Administrative Patent Judge. DECISION ON APPEAL Appellant1 appeals under 35 U.S.C. § 134(a) from the Examiner’s decision rejecting claims 1-4, 6-11, 13-18, and 20, all of the pending claims. Final Act. 2.2 Claims 5, 12, and 19 are cancelled. Id. We have jurisdiction under 35 U.S.C. § 6(b). We affirm. 1 “Appellant” refers to “applicant” as defined in 37 C.F.R. § 1.42 (2012). Appellant identifies the real party-in-interest as SAP SE. Appeal Br. 4. 2 In this Decision, we refer to Appellant’s Appeal Brief (“Appeal Br.,” filed May 1, 2020) and Reply Brief (“Reply Br.,” filed September 29, 2020); the Final Office Action (“Final Act.,” mailed January 2, 2020) and the Examiner’s Answer (“Ans.,” mailed August 5, 2020); and the Specification (“Spec.,” filed July 19, 2016). Rather than repeat the Examiner’s findings and Appellant’s contentions in their entirety, we refer to these documents. Appeal 2020-006790 Application 15/213,816 2 STATEMENT OF THE CASE The claimed methods, media, and systems relate to computer-implemented methods for using modified query execution plans in in-memory database systems. In some implementations, methods include actions of receiving a query from an application, processing a query execution plan (QEP) of the query using a cache simulator to simulate queries to an in- memory database in a hybrid memory system, providing a miss- curve based on the QEP, the miss-curve relating miss-ratios to memory sizes, and determining relative sizes of a first type of memory and a second type of memory in the hybrid memory system at least partially based on the miss-curve. Spec. ¶ 3. As depicted in the Specification’s Figure 1, “a hybrid main memory system 104 may include any number of instances, or cells, of [dynamic random access memory (DRAM)] and [non-volatile memory (NVM)], to provide any amount of memory for use by the [central processing unit(s) (CPU(s))] 102.” Id. ¶ 23. Appeal 2020-006790 Application 15/213,816 3 The Specification’s Figure 3A, depicting miss-curves in accordance with implementations of the claimed methods, media, and systems and including our annotations, is reproduced below. See Spec. ¶ 11. Figure 3A depicts simulated memory sizes ranging from 1KB to 2GB . . . on the horizontal axis (x-axis), and memory miss percentages (miss- ratios) are provided on the vertical axis (y-axis). In some examples, the miss percentage indicates the percentage of memory that is not served. The point on the horizontal axis of miss curves, after which there is no decrease, or a significantly slowed decrease in the miss ratio, indicates the size for the faster access memory. In a hybrid memory system, DRAM is the faster access memory (relative to NVM), but less energy efficient Appeal 2020-006790 Application 15/213,816 4 (relative to NVM). With particular reference to FIG. 3A, the size of the DRAM can be selected as 1024MB, which, no matter what the size of main memory is, the miss ratio remains constant at approximately 50%. This value can be used as the delineation point between DRAM sizing and NVM sizing in a hybrid memory system. Id. ¶ 33 (emphases added). The Specification’s Figure 3B, depicting miss-curves in accordance with implementations of the claimed methods, media, and systems including “mitosis” to provide fragmented relations and including our annotations, is reproduced below. Figure 3B depicts a miss-curve, where a partition of 10 fragments (i.e., M = 10) is used. Spec. ¶ 35. In this example, and as compared to Figure 3A, the miss-curve “indicates that the size of DRAM can be reduced to 150MB, Appeal 2020-006790 Application 15/213,816 5 while still achieving a miss rate of up to 50%.” Id. The Specification discloses: In mitosis, a relation (table) is broken down into M fragments. This approach better uses multi-core parallelism, and improves the application cache locality. The goal of mitosis is to apply horizontal partitioning, and run analysis on smaller subsets of data. In some examples, the granularity of partitioning is important, because too many fragments can cause contention in the caches, where multiple threads will be scheduled to [a] single core and the data will be fetched into the caches for each of the fragments. The tool of the present disclosure, described above, enables selection of the number of fragments M, and provides respective miss-curves for each configuration. In this manner, the best suitable value for M for a given workload (e.g., TPC-H) can be determined. In experiments performed using implementations of the present disclosure, an example of M = 10 was selected, and each function call of an original QEP was broken into 10 smaller fragments. Id. ¶ 34 (emphases added). As noted above, claims 1-4, 6-11, 13-18, and 20 are pending. Claims 1, 8, and 15 are independent. Appeal Br. 20-21 (claim 1), 22-23 (claim 8), 24-25 (claim 15 (Claims App.). Claims 2-4, 6, and 7 depend directly or indirectly from claim 1; claims 9-11, 13, and 14 depend directly or indirectly from claim 8; and claims 16-18 and 20 depend directly or indirectly from claim 15. Id. at 20-26. Claim 1, reproduced below with disputed limitations emphasized, is representative. 1. A computer-implemented method for determining relative sizes of different types of memory in an in-memory database system having hybrid memory, the method being executed by one or more processors and comprising: receiving, by the one or more processors, a query from an application for querying an in-memory database system over one Appeal 2020-006790 Application 15/213,816 6 or more fragmented relations, the in-memory database system being instrumented with code that is executed to collect miss data; processing, by the one or more processors, a query execution plan (QEP) of the query using a cache simulator to simulate queries to the in-memory database, the cache simulator simulating multiple main memory configurations, each main memory configuration comprising respective sizes of each of a first type of memory and a second type of memory bound between a lower bound and an upper bound; providing, by the one or more processors, a set of miss- curves based on the QEP, each miss-curve relating miss-ratios to memory sizes for a main memory configuration and a respective number of fragments and being provided at least partially by fragmenting the QEP into a plurality of fragment QEPs equal to the respective number of fragments, each fragment QEP corresponding to a respective fragmented relation provided by using mitosis to apply horizontal partitioning to one or more relational tables to provide multiple fragmented relations, and executing the fragment QEPs over the respective fragmented relations within the hybrid memory system; and determining the relative sizes of a first type of memory and a second type of memory in the in-memory system at least partially based on a miss-curve of the set of miss-curves. Appeal Br. 20-21 (Claims App.) (emphasis added). Each of independent claims 8 and 15 recites limitations corresponding to the disputed limitations of claim 1. Id. at 23, 25. Appeal 2020-006790 Application 15/213,816 7 REFERENCES AND REJECTIONS The Examiner relies upon the following references: Name3 Reference Published Filed Li US 2007/0050328 A1 Mar. 1, 2007 Aug. 29, 2005 Kim US 2011/0078340 A1 Mar. 31, 2011 Sept. 25, 2009 Simon US 2011/0131199 A1 June 2, 2011 Nov. 30, 2009 Bivens US 2012/0124318 A1 May 17, 2012 Nov. 11, 2010 Waldspurger US 2014/0310462 A1 Oct. 16, 2014 June 26, 2014 Thorat US 2015/0309789 A1 Oct. 29, 2015 Apr. 28, 2014 The Examiner rejects: (1) claims 1, 3, 4, 6, 8, 10, 11, 13, 15, 17, 18, and 20 under 35 U.S.C. § 103 as obvious over the combined teachings of Li, Waldspurger, Bivens, and Simon (Final Act. 3-21); (2) claims 2, 9, and 16 under 35 U.S.C. § 103 as obvious over the combined teachings of Li, Waldspurger, Bivens, Simon, and Kim (id. at 21- 24); and (3) claims 7 and 14 under 35 U.S.C. § 103 as obvious over the combined teachings of Li, Waldspurger, Bivens, Simon, and Thorat (id. at 25-27). We review the appealed rejections for error based upon the issues identified by Appellant, and in light of the contentions and evidence produced thereon. Ex parte Frye, 94 USPQ2d 1072, 1075 (BPAI 2010) (precedential). The Examiner and Appellant focus their findings and contentions, respectively, on claims 1, 8, and 15; so we do as well. See 3 All reference citations are to the first named inventor only. Appeal 2020-006790 Application 15/213,816 8 Appeal Br. 16, 17, 18; Ans. 3-6; Reply Br. 1. Arguments not made are forfeited.4 Unless otherwise indicated, we adopt the Examiner’s findings in the Final Office Action and the Answer as our own and add any additional findings of fact for emphasis. We address the rejections below. ANALYSIS A. Obviousness over Li, Waldspurger, Bivens, and Simon As noted above, the Examiner rejects claims 1, 8, and 15 as obvious over the combined teachings of Li, Waldspurger, Bivens, and Simon. Final Act. 3-8, 9-14, 15-19. In particular, with respect to claim 1, the Examiner finds Li teaches or suggests many, if not the majority, of the limitations of claim 1. Id. at 3-4 (citing Li ¶¶ 4, 7, 29, 32, 38, 53). The Examiner finds, however, Li does not expressly disclose: A computer-implemented method for determining relative sizes of different types of memory in an in-memory database system having [hybrid]5 memory, the method being executed by one or more processors and comprising: the in-memory database system being instrumented with code that is executed to collect miss data; 4 See In re Google Tech. Holdings LLC, 980 F.3d 858, 863 (Fed. Cir. 2020) (“Because Google failed to present these claim construction arguments to the Board, Google forfeited both arguments.”); 37 C.F.R. § 41.37(c)(1)(iv) (2019) (“Except as provided for in §§ 41.41, 41.47 and 41.52, any arguments or authorities not included in the appeal brief will be refused consideration by the Board for purposes of the present appeal.”). 5 Although the Examiner struck through the word “hybrid,” the Examiner relies on Bivens, not Li, to teach or suggest a “hybrid” memory. Final Act. 7. Appeal 2020-006790 Application 15/213,816 9 the cache simulator simulating multiple main memory configurations, each main memory configuration comprising respective sizes of each of a first type of memory and a second type of memory bound between a lower bound and an upper bound; providing, by the one or more processors, a set of miss- curves based on the QEP,6 each miss curve relating miss-ratios to memory sizes for a main memory configuration and a respective number of fragments and being provided . . . determining the relative sizes of a first type of memory and a second type of memory in the in-memory system at least partially based on a miss-curve of the set of miss-curves. See id. at 4-5. Nevertheless, the Examiner finds Waldspurger teaches or suggests the majority of the limitations missing from Li. Final Act. 5-6 (citing Waldspurger ¶¶ 15, 48, 56, 61, 64, 104, Fig. 2). Further, the Examiner finds a person of ordinary skill in the relevant art would have had reason to combine the teachings of Waldspurger with those of Li “to obtain the predictable result of improving cache/memory management.” and, thereby, to achieve the missing limitations of claim 1. Id. at 6. The Examiner also finds, “Li does not expressly disclose: hybrid memory system [and] first type of memory and a second type of memory in the hybrid memory system.” Final Act. 6. However, the Examiner finds Bivens teaches or suggests these limitations. Id. at 7 (citing Bivens ¶ 9); see supra note 5. Moreover, the Examiner finds a person of ordinary skill in the relevant art would have had reason to combine the teachings of Bivens with 6 The Examiner strikes through “EP” of “QEP.” Because the Examiner relies on Li to teach or suggest query execution plans (QEPs), we understand this partial strikethrough to be a harmless error. Final Act. 3 (citing Li ¶ 4). Appeal 2020-006790 Application 15/213,816 10 those of Li and Waldspurger “to obtain the predictable result of improving storage systems.” and, thereby, to achieve the missing limitations of claim 1. Id. at 7. Finally, the Examiner finds, “Li does not expressly disclose: using mitosis to apply horizontal partitioning to one or more relational tables.” Final Act. 7 (emphasis added); see id. at 4 (“each fragment QEP corresponding to a respective fragmented relation provided by using mitosis to apply horizontal partitioning to one or more relational tables to provide multiple fragmented relations”). However, the Examiner finds Simon teaches or suggests this limitation. Id. at 7 (citing Simon ¶¶ 39, 46); see Ans. 6 (citing Simon ¶¶ 33, 39). Moreover, the Examiner finds a person of ordinary skill in the relevant art would have had reason to combine the teachings of Simon with those of Li, Waldspurger, and Bivens “to obtain the predictable result of improving storage” and, thereby, to achieve the methods of claim 1. Id. at 7-8. Appellant contends the Examiner errs for three reasons. Appeal Br. 15-18. As discussed below, these reasons do not persuade us of Examiner error. 1. Li’s Alleged Deficiencies As noted above, the Examiner finds Li teaches or suggests, “at least partially by fragmenting the QEP into a plurality of fragment QEPs equal to the respective number of fragments . . ., each fragment QEP corresponding to a respective fragmented relation provided by using mitosis to apply horizontal partitioning to one or more relational table to provide multiple fragmented relations to provide multiple fragmented relations.” Final Act. 4. Appellant contends: Appeal 2020-006790 Application 15/213,816 11 For example, Li is wholly silent as to “mitosis.” Further, and as discussed above, Li explicitly provides that “[e]ach query plan includes a plurality of query fragment plans and each query fragment plan is associated with a specific data source within the federated information system.” Li, ¶ [0007] (emphasis added). Consequently, Li does not disclose or render obvious “each fragment QEP corresponding to a respective fragmented relation,” and instead provides that “each query fragment plan is associated with a specific data source.” Here, the term “fragmented relation” recited in each of claims 1, 8, and 15 is specific to “relational tables,” also recited in each of claims 1, 8, and 15. In an effort to substantiate Li as relevant, the final Office action dissects the explicit claim language by striking through “relational tables,” and then ignoring the term “fragmented relation.” Appeal Br. 15-16 (italics added); see Reply Br. 2. We disagree with Appellant. First, as noted above, the Specification discloses, “[i]n mitosis, a relation (table) is broken down into M fragments. This approach better uses multi-core parallelism, and improves the application cache locality. The goal of mitosis is to apply horizontal partitioning, and run analysis on smaller subsets of data.” Spec. ¶ 34; see Final Act. 28 (“Applicant’s specification paragraph [0034] states that in mitosis, a relation table is broken into M fragments where the goal is to apply horizontal partitioning.”). “Mitosis” is a term borrowed from biology. We are not aware that it has an accepted meaning in the field of this application, and Appellant does not identify any such meaning. Consistent with Appellant’s Specification, we understand “mitosis” to broadly mean breaking down a table into fragments. The Specification further discloses advantages that may be obtained using “mitosis” in the context of the claimed methods, media, and systems, e.g., “horizontal Appeal 2020-006790 Application 15/213,816 12 partitioning” and “analysis on smaller subsets of data.” See Spec. ¶¶ 34 (“In some examples, the smaller relation fragments (components) for data processing are ideal to fit into the caches of enterprise systems, and exploit the locality with in caches.” (emphasis added)), 35 (“FIG. 3B depicts an example miss-curve for TPC-H Q #3, where a partition of 10 fragments (i.e., M = 10) is used.”); Fig. 3B (depicting horizontal partitioning). Li discloses, “[t]he simulator is further adapted to determine cost estimates for each query plan (i.e., first query cost estimates) and separate cost estimates for each query fragment contained within each query plan (i.e., first query fragment cost estimates).” Li ¶ 7 (emphases added). The Examiner finds, “[t]his shows that Li teaches a plurality of query plan fragments.” Ans. 4. The Examiner need not show that the applied references use the precise terminology recited in the rejected claim in order to show obviousness. Here, it is sufficient for the Examiner to show that “mitosis” as broadly used by Appellant is taught or suggested by Li. Final Act. 4 (citing Li ¶¶ 7, 32), 28 (citing Li ¶ 7); Ans. 3-4 (citing Li ¶ 7). We are persuaded that the Examiner shows Li teaches or suggests the recited “mitosis.” Second, contrary to Appellant’s contention, the Examiner does not ignore the recited “fragmented relations.” The Examiner acknowledges, “Li was not shown to disclose horizontal partitioning on relational tables.” Ans. 4. However, the Examiner finds: The final rejection showed that Simon discloses horizontal partitioning on one or more tables (Paragraph [0039]). Simon further teaches that the tables can have metadata relational schema (Paragraph [0033]). This shows that Simon teaches horizontal partitioning on relation tables. Li teaches the fragmented query plans while Simon teaches horizontal partitioning on relational tables. The combination of the Appeal 2020-006790 Application 15/213,816 13 references would allow for one of ordinary skill in the art to add the result of the horizontal partitioning to the query fragments in order to obtain the predicable result of improving the execution of the query plans. Ans. 4 (emphasis added). In particular, Simon discloses, “[a]n [Enterprise Information Integration (EII)] system may be implemented as middleware that includes wrappers and a query engine. Wrappers expose the metadata relational schema of the data sources called source tables.” Simon ¶ 33 (emphasis added). Further, Simon discloses, “[h]orizontal partitioning may occur because the source tables have been partitioned into multiple databases.” Id. ¶ 39 (emphasis added); see id. ¶ 46 (“All the data about the boxes and their contents is horizontally partitioned over about 80 different databases within each region.”). Thus, the Examiner finds Li’s and Simon’s combined teachings teach or suggest, “each fragment QEP corresponding to a respective fragmented relation provided by using mitosis to apply horizontal partitioning to one or more relational tables to provide multiple fragmented relations.” Ans. 4; Appeal Br. 20 (Claims App.). Appellant cannot show nonobviousness by attacking references individually when the rejection is based on combined teachings of the references. See In re Merck & Co., Inc., 800 F.2d 1091, 1097 (Fed. Cir. 1986); In re Keller, 642 F.2d 413, 426 (CCPA 1981). Here the Examiner relies on the combined teachings of Li and Simon. Final Act. 27-28; Ans. 3-4. Consequently, we are not persuaded of Examiner error for this first reason. 2. Waldspurger’s Alleged Deficiencies As noted above, the Examiner finds Waldspurger teaches or suggests, Appeal 2020-006790 Application 15/213,816 14 providing, by the one or more processors, a set of miss- curves based on the QEP, each miss-curve relating miss-ratios to memory sizes for a main memory configuration and a respective number of fragments and being provided . . .; and determining the relative sizes of a first type of memory and a second type of memory in the in-memory system at least partially based on a miss-curve of the set of miss-curves. Final Act. 6 (citing Waldspurger ¶¶ 15, 48, 64, 104, Fig. 2). Appellant contends, however, [Waldspurger] is directed to a “hash-based spatial sampling method that requires no prior knowledge of the system or its input workload” for miss ratio curve (MRC) construction. [Waldspurger], ¶ [0027]. In further detail, [Waldspurger] discusses a “cache analysis system” that “stores the location identifiers L submitted by any or all of the clients that one desires to construct an MRC,” and notes that “the references (submission of storage location identifiers L) of more than one, or even all, of the clients, for example, all of the [virtual machines (VMs)] on a single host, may be considered as a whole for analysis,” while, “[i]n other cases, however, cache analysis is preferably done per- client so as to be able to construct a separate cache utility curve CUC(Ci) for each client Ci.” Waldspurger, ¶ [0047]. Waldspurger provides that “FIG. 2 is a MRC plot illustrating the results achieved by a prototype of the” hash-based sampling method, where “FIG. 2 plots representative miss ratio curves (MRCs) obtained without sampling, and with 5% and 1% sampling rates, for a 24-hour trace of disk block accesses from a real-world VM workload.” Waldspurger, ¶¶ [0023], [0034]. Appeal Br. 16-17 (emphasis added); see Reply Br. 2. Again, Appellant does not persuade us of Examiner error. Initially, we note the independent claims do not specify how the miss ratio curves are generated. E.g., Appeal Br. 20 (Claims App.) (“the in- memory database system being instrumented with code that is executed to collect miss data” (emphasis added)). Moreover, Appellant does not show Appeal 2020-006790 Application 15/213,816 15 where the claims or the Specification discuss(es) sampling techniques or where the Specification describes particular code to collect miss data. See, e.g., Spec. ¶ 30 (“In some examples, the executable code 208 is provided as bit-code (e.g., machine-readable code) and is executed by the compile-time instrumentation framework 204 to provide a miss-curves file 210, as described in further detail herein.”). Thus, this contention “fail[s] from the outset because . . . [it is] not based on limitations appearing in the claims.” In re Self, 671 F.2d 1344, 1348 (CCPA 1982). Nevertheless, the Examiner finds, “Waldspurger teaches a set of miss- curves for memory.” Ans. 4 (citing Waldspurger, Fig. 2). Waldspurger’s Appeal 2020-006790 Application 15/213,816 16 Figure 2, depicting a graph of cache miss ratio versus cache size, is reproduced below. Figure 2 depicts a miss ration curve (MRC) plot “illustrating the results achieved by a prototype of spatial sampling embodiment.” Waldspurger ¶ 24; see id. ¶ 38 (describing Fig. 2). Waldspurger discloses, however, “the core concept of the invention-hash-based sampling-does not presuppose any particular method for cache simulation.” Id. ¶ 61. Similar to the Specification’s Figure 3A, reproduced above, Waldspurger’s miss curve shows the miss ratio leveling off between a cache size of about 1000MB and about 2000MB. Appeal 2020-006790 Application 15/213,816 17 Although the Examiner acknowledges that Waldspurger’s “miss- curves are not based on a query execution plan,” Li “discloses the query execution plan. Li teaches this by disclosing a query execution plan and selecting an execution plan with the lowest execution cost.” Ans. 4 (citing Li ¶ 4); see also Waldspurger ¶ 15 (“FIG. 1 illustrates qualitatively a typical ‘miss ratio curve’ (MRC) which is often used to represent cache performance. By convention, an MRC is plotted with the cache size on the X-axis, and the cache miss ratio (i.e., misses/(hits+misses)) on the Y-axis.”). Thus, the Examiner relies on Li’s and Waldspurger’s combined teachings to teach or suggest, “providing, by the one or more processors, a set of miss- curves based on the QEP, each miss-curve relating miss-ratios to memory sizes for a main memory configuration and a respective number of fragments.” As we noted above, Appellant cannot show nonobviousness by attacking references individually when the rejection is based on combined teachings of the references. See Merck, 800 F.2d at 1097; Keller, 642 F.2d at 426. With regard to the second limitation, i.e., “determining the relative sizes of a first type of memory and a second type of memory in the in- memory system at least partially based on a miss-curve of the set of miss- curves,” the Examiner finds: Waldspurger teaches the set of miss curves which is shown in the graph comparing cache miss ratio to cache size (Fig. 2). This figure shows multiple miss curves along with multiple cache sizes. The claim does not appear to require that the first and second types of memory be different but merely that there are two sets of memory that each have a type. Therefore, using the graph of figure 2 in Waldspurger, one of ordinary skill in the art would see how multiple sizes of memory could be determined. Appeal 2020-006790 Application 15/213,816 18 Ans. 5 (emphasis added). Waldspurger discloses, “the slope of the MRC is used to help determine the optimal partitioning and allocation.” Waldspurger ¶ 15; see id. ¶ 61 (“In an online system that makes fine-grained dynamic cache allocation decisions, the per-client MRCs may be updated incrementally in an online manner; allocation decisions may then be made periodically using the current set of per-client MRCs.”). Thus, the Examiner finds Waldspurger teaches or suggests this limitation of the independent claims. Ans. 5. Consequently, we are not persuaded of Examiner error for this second reason. 3. Simon’s Alleged Deficiencies As noted above, the Examiner finds Simon teaches or suggests, “using mitosis to apply horizontal partitioning to one or more relational tables.” Final Act. 7 (citing Simon ¶¶ 39, 46); Ans. 5 (citing Simon ¶¶ 33, 39); see supra Section A.1. The Examiner further finds, “[w]hile Simon does not mention mitosis, the goal of mitosis is to horizontally partition relation tables (Appellant’s Specification Paragraph [0034]). Since Simon discloses horizontally partitioning relation tables, one of ordinary skill in the art would recognize that this could be considered mitosis based on the Appellant’s specification.” Ans. 5. Appellant contends, however, Nowhere does the instant application state or suggest that horizontal partitioning is only achievable using mitosis. In fact, the specification of the instant application at least implies the opposite in noting that “[t]his approach,” referencing mitosis of the immediately preceding sentence, “better uses multicore parallelism, and improves the application cache locality,” as an explicit comparison to other approaches. Accordingly, the Appeal 2020-006790 Application 15/213,816 19 specification does not provide support for the Examiner’s conclusion that horizontal partitioning necessarily requires using mitosis and, in fact, provides the opposite. Reply Br. 4 (emphases added); see id. at 5 (“Accordingly, Simon does not provide support for the Examiner’s conclusion that horizontal partitioning necessarily implies mitosis.”). We are not persuaded of error by Appellant’s argument. Appellant misunderstands the Examiner’s findings. As we noted above, the Examiner finds the Specification broadly defines mitosis as partitioning that is capable of achieving certain advantages (see Spec. ¶ 34), Li teaches or suggests such horizontal partitioning, such as mitosis (see Li ¶ 7), and Simon teaches or suggests applying such horizontal partitioning to relational tables to provide multiple fragmented relations (see Simon ¶¶ 33, 39). Ans. 5-6. The Examiner does not find that horizontal partitioning requires mitosis, but, instead, that, given the Specification’s broad definition of mitosis, “mitosis” encompasses horizontal partitioning. See Final Act. 28 (“Applicant’s specification paragraph [0034] states that in mitosis, a relation table is broken into M fragments where the goal is to apply horizontal partitioning.”). Thus, the Examiner finds, “the combination of the references would allow for horizontal partitioning using mitosis.” Id.; see Ans. 5 (“Since Simon discloses horizontally partitioning relation tables, one of ordinary skill in the art would recognize that this could be considered mitosis based on the Appellant’s specification.”). We agree with the Examiner. Consequently, we are not persuaded of Examiner error for this third reason. Appeal 2020-006790 Application 15/213,816 20 We are not persuaded the Examiner errs in rejecting independent claim 1, as well as independent claims 8 and 15, as obvious over the combined teachings of Li, Waldspurger, Bivens, and Simon, and we sustain the Examiner’s rejection of those claims. Appellant does not challenge this rejection of dependent claims 3, 4, 6, 10, 11, 13, 17, 18, and 20 separately from the challenge to the rejection of the independent claims. Appeal Br. 18; see Ans. 6. Therefore, we also sustain the rejection of those dependent claims. B. Obviousness over Li, Waldspurger, Bivens, Simon, and Kim or Thorat As noted above, the Examiner also rejects: (1) claims 2, 9, and 16 under 35 U.S.C. § 103 as obvious over the combined teachings of Li, Waldspurger, Bivens, Simon, and Kim; and (2) claims 7 and 14 under 35 U.S.C. § 103 as obvious over the combined teachings of Li, Waldspurger, Bivens, Simon, and Thorat. Final Act. 21-27. Each of claims 2, 7, 9, 14, and 16 depends from one of claims 1, 8, and 15. Appeal Br. 21-26 (Claims App.). Appellant contends that, because neither Kim nor Thorat cures the alleged deficiencies in the combined teachings of Li, Waldspurger, Bivens, and Simon with respect to their base claims, each of these dependent claims is patentable for at least the same reasons as its base claim. Appeal Br. 18. Appellant does not otherwise challenge the rejections of these dependent claims separately. Because we are not persuaded the Examiner errs in rejecting claim 1, 8, or 15 (see supra Section A), on this record, we also are not persuaded the Examiner errs in rejecting dependent claim 2, 7, 9, 14, or 16. Consequently, we sustain the rejections of those dependent claims. Appeal 2020-006790 Application 15/213,816 21 DECISION 1. The Examiner does not err in rejecting: a. claims 1, 3, 4, 6, 8, 10, 11, 13, 15, 17, 18, and 20 under 35 U.S.C. § 103 as obvious over the combined teachings of Li, Waldspurger, Bivens, and Simon; b. claims 2, 9, and 16 under 35 U.S.C. § 103 as obvious over the combined teachings of Li, Waldspurger, Bivens, Simon, and Kim; and c. claims 7 and 14 under 35 U.S.C. § 103 as obvious over the combined teachings of Li, Waldspurger, Bivens, Simon, and Thorat. 2. Thus, on this record, claims 1-4, 6-11, 13-18, and 20 are not patentable. CONCLUSION We affirm the rejections of claims 1-4, 6-11, 13-18, and 20. DECISION SUMMARY In summary: Claim(s) Rejected 35 U.S.C. § Reference(s)/Basis Affirmed Reversed 1, 3, 4, 6, 8, 10, 11, 13, 15, 17, 18, 20 103 Li, Waldspurger, Bivens, Simon 1, 3, 4, 6, 8, 10, 11, 13, 15, 17, 18, 20 2, 9, 16 103 Li, Waldspurger, Bivens, Simon, Kim 2, 9, 16 7, 14 103 Li, Waldspurger, Bivens, Simon, Thorat 7, 14 Overall Outcome 1-4, 6-11, 13-18, 20 Appeal 2020-006790 Application 15/213,816 22 TIME PERIOD FOR RESPONSE No time period for taking any subsequent action in connection with this appeal may be extended under 37 C.F.R. § 1.136(a). See 37 C.F.R. § 1.136(a)(1)(iv) (2013). AFFIRMED