- Passing functions to Spark (be careful the reference to the containing object which need to be serializable)
class SearchFunctions(val query: String) {
def isMatch(s: String): Boolean = {
s.contains(query)
}
def getMatchesFunctionReference(rdd: RDD[String]): RDD[String] = {
// Problem: “isMatch” means “this.isMatch”, so we pass all of “this”
rdd.map(isMatch)
}
def getMatchesFieldReference(rdd: RDD[String]): RDD[String] = {
// Problem: “query” means “this.query”, so we pass all of “this”
rdd.map(x => x.split(query))
}
def getMatchesNoReference(rdd: RDD[String]): RDD[String] = {
// Safe: extract just the field we need into a local variable
val query_ = this.query
rdd.map(x => x.split(query_))
}
}
Note that passing in local serializable
variables or functions that are members of a top-level object is always safe
- Basic RDD transformations
- map, flatMap
- set operations, union, distinct, intersection, subtract, cartesian. pay attention to operations needing shuffle (multiple RDDs with same type)
- reduce
- collect
- count
- fold -> currying functions. provide "zero value" as first parameter which then applied as the first parameter of the function.
- aggregate, initial value, function 1 to accumulate value from each node, function 2 to merge all accumulated values.
- foreach , run function on distributed nodes
- take, returned result not in order
- RDD implicitly converted to real scala classes, like RDD[Double] to DoubleRDDFunctions
- Persist (eviction cache of partition computing result by LRU)
MEMORY_ONLY
MEMORY_ONLY_SER
MEMORY_AND_DISK
MEMORY_AND_DISK_SER
DISK_ONLY