官术网_书友最值得收藏!

Predicate push-down on smart data sources

Smart data sources are those that support data processing directly in their own engine--where the data resides--by preventing unnecessary data to be sent to Apache Spark.

On example is a relational SQL database with a smart data source. Consider a table with three columns: column1, column2, and column3, where the third column contains a timestamp. In addition, consider an ApacheSparkSQL query using this JDBC data source but only accessing a subset of columns and rows based using projection and selection. The following SQL query is an example of such a task:

select column2,column3 from tab where column3>1418812500

Running on a smart data source, data locality is made use of by letting the SQL database do the filtering of rows based on timestamp and removal of column1.

Let's have a look at a practical example on how this is implemented in the Apache Spark MongoDB connector. First, we'll take a look at the class definition:

private[spark] case class MongoRelation(mongoRDD: MongoRDD[BsonDocument], _schema: Option[StructType])(@transient val sqlContext: SQLContext)
extends BaseRelation
with PrunedFilteredScan
with InsertableRelation
with LoggingTrait {

As you can see, the MongoRelation class extends BaseRelation. This is all that is needed to create a new plugin to the DataSource API in order to support an additional data source. However, this class also implemented the PrunedFilteredScan trait adding the buildScan method in order to support filtering on the data source itself. So let's take a look at the implementation of this method:

override def buildScan(requiredColumns: Array[String], filters: Array[Filter]): RDD[Row] = {
// Fields that explicitly aren't nullable must also be added to the filters
val pipelineFilters = schema
.fields
.filter(!_.nullable)
.map(_.name)
.map(IsNotNull)
++ filters

if (requiredColumns.nonEmpty || pipelineFilters.nonEmpty) {
logInfo(s"requiredColumns: ${requiredColumns.mkString(", ")},
filters: ${pipelineFilters.mkString(", ")}")
}
mongoRDD.appendPipeline(createPipeline(requiredColumns, pipelineFilters))
.map(doc => documentToRow(doc, schema, requiredColumns))
}

It is not necessary to understand the complete code snippet, but you can see that two parameters are passed to the buildScan method: requiredColumns and filters. This means that the code can use this information to remove columns and rows directly using the MongoDB API.

主站蜘蛛池模板: 宣城市| 章丘市| 临清市| 娄烦县| 买车| 会宁县| 斗六市| 富川| 右玉县| 固安县| 郯城县| 怀仁县| 青神县| 漳州市| 长治市| 绵阳市| 兴安盟| 海丰县| 扎赉特旗| 民勤县| 突泉县| 响水县| 凤冈县| 张家口市| 光泽县| 耿马| 安义县| 岑巩县| 巴里| 尼木县| 嵊泗县| 潼南县| 云南省| 吉林省| 如皋市| 吕梁市| 沧州市| 黑山县| 芜湖市| 翼城县| 扶余县|