pyspark.RDD.persist¶

RDD.persist(storageLevel: pyspark.storagelevel.StorageLevel = StorageLevel(False, True, False, False, 1)) → pyspark.rdd.RDD[T][source]¶

Set this RDD’s storage level to persist its values across operations after the first time it is computed. This can only be used to assign a new storage level if the RDD does not have a storage level set yet. If no storage level is specified defaults to (MEMORY_ONLY).

New in version 0.9.1.

Parameters

storageLevelStorageLevel, default MEMORY_ONLY: the target storage level

Returns

RDD: The same RDD with storage level set to storageLevel.

See also

RDD.cache()
RDD.unpersist()
RDD.getStorageLevel()

Examples

>>> rdd = sc.parallelize(["b", "a", "c"])
>>> rdd.persist().is_cached
True
>>> str(rdd.getStorageLevel())
'Memory Serialized 1x Replicated'
>>> _ = rdd.unpersist()
>>> rdd.is_cached
False

>>> from pyspark import StorageLevel
>>> rdd2 = sc.range(5)
>>> _ = rdd2.persist(StorageLevel.MEMORY_AND_DISK)
>>> rdd2.is_cached
True
>>> str(rdd2.getStorageLevel())
'Disk Memory Serialized 1x Replicated'

Can not override existing storage level

>>> _ = rdd2.persist(StorageLevel.MEMORY_ONLY_2)
Traceback (most recent call last):
    ...
py4j.protocol.Py4JJavaError: ...

Assign another storage level after unpersist

>>> _ = rdd2.unpersist()
>>> rdd2.is_cached
False
>>> _ = rdd2.persist(StorageLevel.MEMORY_ONLY_2)
>>> str(rdd2.getStorageLevel())
'Memory Serialized 2x Replicated'
>>> rdd2.is_cached
True
>>> _ = rdd2.unpersist()

pyspark.RDD.partitionBy pyspark.RDD.pipe