Case 4:23-cv-01147-ALM Document 101-2 Filed 03/11/25 Page 1 of 6 PageID #:
`6581
`
`Exhibit A
`
`
`
`

`

`Case 4:23-cv-01147-ALM Document 101-2 Filed 03/11/25 Page 2 of 6 PageID #:
`6582
`
`

`

`Case 4:23-cv-01147-ALM Document 101-2 Filed 03/11/25 Page 3 of 6 PageID #:
`6583
`
`
`
`8776.) “Spark uses the Hadoop core library to talk to HDFS and other Hadoop-supported storage
`
`systems.” (Id. at 8774.)
`
`1079. As discussed above, R2 accuses open-source Apache Spark functionality that is
`
`implemented by the Databricks Data Intelligence Platform as purportedly infringing the asserted
`
`claims of the ’610 patent. I disagree that open-source Apache Spark meets the requirements of the
`
`asserted claims for at least the reasons discussed in Databricks’ response to Interrogatory No. 7,
`
`and as I will further explain in my rebuttal non-infringement report. But, even if one were to
`
`assume that the accused open-source Apache Spark functionality meets the requirements of the
`
`asserted claims, Yahoo granted a license to the ’610 patent—as well as every other Yahoo patent—
`
`that purportedly covers open-source Apache Spark by virtue of (1) Yahoo’s contributions to open-
`
`source Apache Spark under the Apache 2.0 license, and (2) Yahoo’s contributions to open-source
`
`projects that were combined with, based on and derived from open-source Apache Spark, including
`
`Pig on Spark, CaffeOnSpark, and streaming-benchmarks. As explained below, Yahoo made these
`
`contributions during the time it owned the ’610 patent—i.e., between the October 5, 2006 filing
`
`date of the ’610 patent and the April 18, 2016 assignment to Excalibur, and thus, granted all users
`
`of open-source Apache Spark a license to the ’610 patent. (R2_DB_000466 at 0468 (patent
`
`assignment from named inventors to Yahoo dated October 4, 2006); R2_DB_000471 (patent
`
`assignment from Yahoo to Excalibur IP, LLC dated April 18, 2016).)
`
`a.
`
`The Accused Open-Source Apache Spark Functionality is in the
`Versions of Apache Spark to which Yahoo Contributed
`
`1080. As an initial matter, R2 alleges Databricks first infringed the ’610 patent on June
`
`15, 2015, when open-source Apache Spark first introduced SortMergeJoin functionality. This
`
`functionality was contributed to Apache Spark while Yahoo owned the ’610 patent and was first
`
`committed on April 15, 2015. (JWeissman_00000294-0325; JWeissman_00003139-3140.) Thus,
`
`
`
`462
`
`

`

`Case 4:23-cv-01147-ALM Document 101-2 Filed 03/11/25 Page 4 of 6 PageID #:
`6584
`
`
`
`under R2’s own allegations that that the accused functionality as of June 2015 purportedly meets
`
`the asserted claims of the ’610 patent (which I disagree with), Yahoo granted all users of Apache
`
`Spark a license to the ’610 patent under the Apache 2.0 license.
`
`1081. Below I further analyze the versions of Apache Spark that implemented the files
`
`that R2 accuses in its infringement contentions. I reserve the right to supplement my opinions in
`
`view of R2’s forthcoming opening report.
`
`(1)
`
`Apache Spark 1.0
`
`1082. The GitHub repository for Apache Spark Branch 1.0 includes the Apache 2.0
`
`license and is part of the Apache-licensed “Work.” (Databricks_R2_00116013-6031.) Thomas
`
`Graves was a Yahoo employee that was managing Yahoo’s internal open-source Apache Spark
`
`team, and he committed to Apache Spark Branch 1.0 between October 30, 2013 and June 12,
`
`2014.18 (Databricks_R2_00116032-6039.) Additionally, Mridul Muralidharan was also a Yahoo
`
`employee on Thomas Graves’ internal Yahoo team working on open-source Apache Spark and
`
`committed code to Apache Spark Branch 1.0 between April 7, 2013 and March 19, 2014.19
`
`(JWeissman_00000103-0109.)
`
`1083. Beginning with Apache Spark 1.0, released May 30, 2014, Apache Spark “included
`
`shuffle functionality . . . functionality to merge datasets . . . [and] the ability to join datasets that
`
`had different schema.” (Graves Dep. at 92:2-22.) Thomas Graves personally “contribute[d] to the
`
`[Spark] shuffle functionality . . . running on YARN.” (Id. at 224:12-22.) Additionally, Thomas
`
`Graves and Mridul Muralidharan contributed to at least the following additional functionalities,
`
`including,
`
`for
`
`example, RDDs, SparkContext,
`
`and Schedulers.
`
`
`
`(See,
`
`e.g.,
`
`
`18 Thomas Graves testified that his contributions in GitHub are under the name “tgravescs.”
`(Graves Dep. at 93:14-18.)
`19 Thomas Graves testified that Mridul Muralidharan’s contributions in GitHub are under the
`name “mridulm.” (Id. at 109:21-110:3.)
`
`
`
`463
`
`

`

`Case 4:23-cv-01147-ALM Document 101-2 Filed 03/11/25 Page 5 of 6 PageID #:
`6585
`
`
`
`2016, when Yahoo owned the ’610 patent and was contributing to Apache Spark.
`
`(Databricks_R2_00117016; Databricks_R2_00117802.)
`
`1124. I understand
`
`that R2
`
`alleges org/apache/spark/sql/catalyst/expressions/
`
`namedExpressions.scala allegedly meets the claimed mapping functionality. (See generally R2
`
`Infringement Contentions, Ex. A.) Such code was available in Apache Spark-1.6.1 as
`
`sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/
`
`expressions/namedExpressions.scala
`
`prior to April 18, 2016, when Yahoo owned the ’610 patent and was contributing to Apache Spark.
`
`(Databricks_R2_00117127; Databricks_R2_00117831.)
`
`1125. I
`
`understand
`
`that
`
`R2
`
`alleges
`
`org/apache/spark/sql/execution/joins/SortMergeJoinExec.scala allegedly meets
`
`the claimed
`
`mapping functionality. (See generally R2 Infringement Contentions, Ex. A.) Such code was
`
`available in Apache Spark-1.6.1 as sql/core/src/main/scala/org/apache/spark/sql/execution/joins/
`
`SortMergeJoin.scala prior to April 18, 2016, when Yahoo owned the ’610 patent and was
`
`contributing
`
`to Apache Spark.
`
`
`
`(Databricks_R2_00117119; Databricks_R2_00117834.)
`
`SortMergeJoin.scala was renamed to SortMergeJoinExec.scala in a patch to add the “Exec” suffix
`
`to all physical operators on April 22, 2016. (JWeissman_00000151-0262.)
`
`1126. I
`
`understand
`
`that R2
`
`alleges
`
`org/apache/spark/sql/execution/joins/
`
`SortMergeJoinEvaluatorFactory.scala allegedly meets the claimed mapping functionality. (See
`
`generally R2 Infringement Contentions, Ex. A.) SortMergeJoin.scala has the same or substantially
`
`similar functionality as the functionality R2 accuses within SortMergeJoinEvaluatorFactory.scala.
`
`Such
`
`code was
`
`available
`
`in Apache Spark-1.6.1
`
`as
`
`sql/core/src/main/scala/org/
`
`apache/spark/sql/execution/ joins/SortMergeJoin.scala prior to April 18, 2016, when Yahoo
`
`owned the ’610 patent and was contributing to Apache Spark. (JWeissman_00000284-0290;
`
`
`
`476
`
`

`

`Case 4:23-cv-01147-ALM Document 101-2 Filed 03/11/25 Page 6 of 6 PageID #:
`6586
`
`
`
`JWeissman_00000291-0293.) I further understand that in response to a question regarding
`
`whether
`
`R2
`
`alleges
`
`that
`
`Databricks
`
`infringes
`
`in
`
`the
`
`absence
`
`of
`
`SortMergeJoinEvaluatorFactory.scala, R2 alleged that Databricks first infringed when it
`
`“introduced the SortMergeJoin functionality as part of its offerings.” (JWeissman_00003139-
`
`3140.) SortMergeJoin.scala was available in Apache Spark-1.6.1 as of April 15, 2015.
`
`(JWeissman_00000294-0325.)
`
`(c)
`
`Reducing
`
`1127. I understand that R2 alleges org/apache/spark/rdd/ZippedPartitionsRDD.scala,
`
`org/apache/spark/sql/execution/joins/SortMergeJoinExec.scala,
`
`org/apache/spark/sql/execution/joins/BroadcastHashJoinExec.scala,
`
`org/apache/spark/sql/DataFrameJoinSuite.scala,
`
`org/apache/spark/sql/execution/datasources/DataSourceStrategy.scala,
`
`and org/apache/spark/sql/execution/joins/SortMergeJoinEvaluatorFactory.scala allegedly meet
`
`the claimed reducing functionality. (See generally R2 Infringement Contentions, Ex. A.) As
`
`explained above, such code was available in Apache Spark-1.6.1 prior to April 18, 2016, when
`
`Yahoo owned the ’610 patent and was contributing to Apache Spark.
`
`1128. I understand
`
`that R2 alleges org/apache/spark/rdd/MapPartitionsRDD.scala
`
`allegedly meets the claimed reducing functionality. (See generally R2 Infringement Contentions,
`
`Ex. A.)
`
` Such code was available in Apache Spark-1.6.1 as core/src/main/scala/org/
`
`apache/spark/rdd/MapPartitionsRDD.scala prior to April 18, 2016, when Yahoo owned the ’610
`
`patent
`
`and was
`
`contributing
`
`to Apache
`
`Spark.
`
`
`
`(Databricks_R2_00117134;
`
`Databricks_R2_00117842.)
`
`
`
`477
`
`