diff --git a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/Optimizer.scala b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/Optimizer.scala
index 91208479be03b..b63468ac24a68 100644
--- a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/Optimizer.scala
+++ b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/optimizer/Optimizer.scala
@@ -58,6 +58,7 @@ abstract class Optimizer(sessionCatalog: SessionCatalog)
         LimitPushDown,
         ColumnPruning,
         InferFiltersFromConstraints,
+        TransitPredicateInOuterJoin,
         // Operator combine
         CollapseRepartition,
         CollapseProject,
@@ -1071,6 +1072,66 @@ object PushPredicateThroughJoin extends Rule[LogicalPlan] with PredicateHelper {
   }
 }
 
+/**
+ * Infer and transit predicate from the preserved side to the null-supplying side
+ * of an outer join. The predicate is inferred from the preserved side based on the
+ * join condition and will be pushed over to the null-supplying side. For example,
+ * if the preserved side has constraints of the form 'a > 5' and the join condition
+ * is 'a = b', in which 'b' is an attribute from the null-supplying side, a [[Filter]]
+ * operator of 'b > 5' will be applied to the null-supplying side.
+ *
+ * Applying this rule will not change the constraints of the [[Join]] operator, so
+ * aside from its child being transformed, there is no side-effect to the [[Join]]
+ * operator itself.
+ */
+object TransitPredicateInOuterJoin extends Rule[LogicalPlan] with PredicateHelper {
+  def apply(plan: LogicalPlan): LogicalPlan =
+    if (!plan.conf.constraintPropagationEnabled) {
+      plan
+    } else plan transform {
+      case j@Join(left, right, joinType, joinCondition) =>
+        joinType match {
+          case RightOuter if joinCondition.isDefined =>
+            val rightConstraints = right.constraints.union(
+              splitConjunctivePredicates(joinCondition.get).toSet)
+            val inferredConstraints = ExpressionSet(
+              QueryPlanConstraints.inferAdditionalConstraints(rightConstraints))
+            val leftConditions = inferredConstraints
+              .filter(_.deterministic)
+              .filter(_.references.subsetOf(left.outputSet))
+            if (leftConditions.isEmpty) {
+              j
+            } else {
+              // push the predicate down to left side sub query.
+              val newLeft = leftConditions.
+                reduceLeftOption(And).map(Filter(_, left)).getOrElse(left)
+              val newRight = right
+
+              Join(newLeft, newRight, RightOuter, joinCondition)
+            }
+          case LeftOuter if joinCondition.isDefined =>
+            val leftConstraints = left.constraints.union(
+              splitConjunctivePredicates(joinCondition.get).toSet)
+            val inferredConstraints = ExpressionSet(
+              QueryPlanConstraints.inferAdditionalConstraints(leftConstraints))
+            val rightConditions = inferredConstraints
+              .filter(_.deterministic)
+              .filter(_.references.subsetOf(right.outputSet))
+            if (rightConditions.isEmpty) {
+              j
+            } else {
+              // push the predicate down to right side sub query.
+              val newRight = rightConditions.
+                reduceLeftOption(And).map(Filter(_, right)).getOrElse(right)
+              val newLeft = left
+
+              Join(newLeft, newRight, LeftOuter, joinCondition)
+            }
+          case _ => j
+        }
+    }
+}
+
 /**
  * Combines two adjacent [[Limit]] operators into one, merging the
  * expressions into one single expression.
diff --git a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/plans/logical/QueryPlanConstraints.scala b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/plans/logical/QueryPlanConstraints.scala
index 046848875548b..de49d749d86e1 100644
--- a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/plans/logical/QueryPlanConstraints.scala
+++ b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/plans/logical/QueryPlanConstraints.scala
@@ -29,7 +29,7 @@ trait QueryPlanConstraints { self: LogicalPlan =>
   lazy val allConstraints: ExpressionSet = {
     if (conf.constraintPropagationEnabled) {
       ExpressionSet(validConstraints
-        .union(inferAdditionalConstraints(validConstraints))
+        .union(QueryPlanConstraints.inferAdditionalConstraints(validConstraints))
         .union(constructIsNotNullConstraints(validConstraints)))
     } else {
       ExpressionSet(Set.empty)
@@ -96,13 +96,16 @@ trait QueryPlanConstraints { self: LogicalPlan =>
     case _: NullIntolerant => expr.children.flatMap(scanNullIntolerantAttribute)
     case _ => Seq.empty[Attribute]
   }
+}
+
+object QueryPlanConstraints {
 
   /**
    * Infers an additional set of constraints from a given set of equality constraints.
    * For e.g., if an operator has constraints of the form (`a = 5`, `a = b`), this returns an
    * additional constraint of the form `b = 5`.
    */
-  private def inferAdditionalConstraints(constraints: Set[Expression]): Set[Expression] = {
+  def inferAdditionalConstraints(constraints: Set[Expression]): Set[Expression] = {
     var inferredConstraints = Set.empty[Expression]
     constraints.foreach {
       case eq @ EqualTo(l: Attribute, r: Attribute) =>
diff --git a/sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/optimizer/InferFiltersFromConstraintsSuite.scala b/sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/optimizer/InferFiltersFromConstraintsSuite.scala
index f78c2356e35a5..c877fffdec57d 100644
--- a/sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/optimizer/InferFiltersFromConstraintsSuite.scala
+++ b/sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/optimizer/InferFiltersFromConstraintsSuite.scala
@@ -33,6 +33,7 @@ class InferFiltersFromConstraintsSuite extends PlanTest {
         PushPredicateThroughJoin,
         PushDownPredicate,
         InferFiltersFromConstraints,
+        TransitPredicateInOuterJoin,
         CombineFilters,
         SimplifyBinaryComparison,
         BooleanSimplification) :: Nil
@@ -204,4 +205,40 @@ class InferFiltersFromConstraintsSuite extends PlanTest {
     val optimized = Optimize.execute(originalQuery)
     comparePlans(optimized, correctAnswer)
   }
+
+  test("SPARK-21479: Outer join after-join filters push down to null-supplying side") {
+    val x = testRelation.subquery('x)
+    val y = testRelation.subquery('y)
+    val condition = Some("x.a".attr === "y.a".attr)
+    val originalQuery = x.join(y, LeftOuter, condition).where("x.a".attr === 2).analyze
+    val left = x.where(IsNotNull('a) && 'a === 2)
+    val right = y.where(IsNotNull('a) && 'a === 2)
+    val correctAnswer = left.join(right, LeftOuter, condition).analyze
+    val optimized = Optimize.execute(originalQuery)
+    comparePlans(optimized, correctAnswer)
+  }
+
+  test("SPARK-21479: Outer join pre-existing filters push down to null-supplying side") {
+    val x = testRelation.subquery('x)
+    val y = testRelation.subquery('y)
+    val condition = Some("x.a".attr === "y.a".attr)
+    val originalQuery = x.join(y.where("y.a".attr > 5), RightOuter, condition).analyze
+    val left = x.where(IsNotNull('a) && 'a > 5)
+    val right = y.where(IsNotNull('a) && 'a > 5)
+    val correctAnswer = left.join(right, RightOuter, condition).analyze
+    val optimized = Optimize.execute(originalQuery)
+    comparePlans(optimized, correctAnswer)
+  }
+
+  test("SPARK-21479: Outer join no filter push down to preserved side") {
+    val x = testRelation.subquery('x)
+    val y = testRelation.subquery('y)
+    val condition = Some("x.a".attr === "y.a".attr)
+    val originalQuery = x.join(y.where("y.a".attr === 1), LeftOuter, condition).analyze
+    val left = x
+    val right = y.where(IsNotNull('a) && 'a === 1)
+    val correctAnswer = left.join(right, LeftOuter, condition).analyze
+    val optimized = Optimize.execute(originalQuery)
+    comparePlans(optimized, correctAnswer)
+  }
 }