Apache Spark's default serialization relies on Java with the default readObject(...) and writeObject(...) methods for all Serializable classes. This is a very fine default behavior as long as you don't rely on it too much...