[SPARK-4548] []SPARK-4517] improve performance of python broadcast

Re-implement the Python broadcast using file:

1) serialize the python object using cPickle, write into disks.
2) Create a wrapper in JVM (for the dumped file), it read data from during serialization
3) Using TorrentBroadcast or HttpBroadcast to transfer the data (compressed) into executors
4) During deserialization, writing the data into disk.
5) Passing the path into Python worker, read data from disk and unpickle it into python object, until the first access.

It fixes the performance regression introduced in #2659, has similar performance as 1.1, but support object larger than 2G, also improve the memory efficiency (only one compressed copy in driver and executor).

Testing with a 500M broadcast and 4 tasks (excluding the benefit from reused worker in 1.2):

         name |   1.1   | 1.2 with this patch |  improvement
---------|--------|---------|--------
      python-broadcast-w-bytes  |	25.20  |	9.33   |	170.13% |
        python-broadcast-w-set	  |     4.13	   |    4.50  |	-8.35%  |

Testing with 100 tasks (16 CPUs):

         name |   1.1   | 1.2 with this patch |  improvement
---------|--------|---------|--------
     python-broadcast-w-bytes	| 38.16	| 8.40	 | 353.98%
        python-broadcast-w-set	| 23.29	| 9.59 |	142.80%

Author: Davies Liu <davies@databricks.com>

Closes #3417 from davies/pybroadcast and squashes the following commits:

50a58e0 [Davies Liu] address comments
b98de1d [Davies Liu] disable gc while unpickle
e5ee6b9 [Davies Liu] support large string
09303b8 [Davies Liu] read all data into memory
dde02dd [Davies Liu] improve performance of python broadcast

(cherry picked from commit 6cf507685efd01df77d663145ae08e48c7f92948)
Signed-off-by: Josh Rosen <joshrosen@databricks.com>

8 files changed, 135 insertions, 258 deletions

diff --git a/core/src/main/scala/org/apache/spark/api/python/PythonRDD.scala b/core/src/main/scala/org/apache/spark/api/python/PythonRDD.scala
index b80c771d58..e0bc00e1eb 100644
--- a/core/src/main/scala/org/apache/spark/api/python/PythonRDD.scala
+++ b/core/src/main/scala/org/apache/spark/api/python/PythonRDD.scala
@@ -19,7 +19,7 @@ package org.apache.spark.api.python
 
 import java.io._
 import java.net._
-import java.util.{List => JList, ArrayList => JArrayList, Map => JMap, Collections}
+import java.util.{List => JList, ArrayList => JArrayList, Map => JMap, UUID, Collections}
 
 import org.apache.spark.input.PortableDataStream
 
@@ -47,7 +47,7 @@ private[spark] class PythonRDD(
     pythonIncludes: JList[String],
     preservePartitoning: Boolean,
     pythonExec: String,
-    broadcastVars: JList[Broadcast[Array[Array[Byte]]]],
+    broadcastVars: JList[Broadcast[PythonBroadcast]],
     accumulator: Accumulator[JList[Array[Byte]]])
   extends RDD[Array[Byte]](parent) {
 
@@ -230,8 +230,7 @@ private[spark] class PythonRDD(
           if (!oldBids.contains(broadcast.id)) {
             // send new broadcast
             dataOut.writeLong(broadcast.id)
-            dataOut.writeLong(broadcast.value.map(_.length.toLong).sum)
-            broadcast.value.foreach(dataOut.write)
+            PythonRDD.writeUTF(broadcast.value.path, dataOut)
             oldBids.add(broadcast.id)
           }
         }
@@ -368,24 +367,8 @@ private[spark] object PythonRDD extends Logging {
     }
   }
 
-  def readBroadcastFromFile(
-      sc: JavaSparkContext,
-      filename: String): Broadcast[Array[Array[Byte]]] = {
-    val size = new File(filename).length()
-    val file = new DataInputStream(new FileInputStream(filename))
-    val blockSize = 1 << 20
-    val n = ((size + blockSize - 1) / blockSize).toInt
-    val obj = new Array[Array[Byte]](n)
-    try {
-      for (i <- 0 until n) {
-        val length = if (i < (n - 1)) blockSize else (size % blockSize).toInt
-        obj(i) = new Array[Byte](length)
-        file.readFully(obj(i))
-      }
-    } finally {
-      file.close()
-    }
-    sc.broadcast(obj)
+  def readBroadcastFromFile(sc: JavaSparkContext, path: String): Broadcast[PythonBroadcast] = {
+    sc.broadcast(new PythonBroadcast(path))
   }
 
   def writeIteratorToStream[T](iter: Iterator[T], dataOut: DataOutputStream) {
@@ -824,3 +807,49 @@ private class PythonAccumulatorParam(@transient serverHost: String, serverPort:
     }
   }
 }
+
+/**
+ * An Wrapper for Python Broadcast, which is written into disk by Python. It also will
+ * write the data into disk after deserialization, then Python can read it from disks.
+ */
+private[spark] class PythonBroadcast(@transient var path: String) extends Serializable {
+
+  /**
+   * Read data from disks, then copy it to `out`
+   */
+  private def writeObject(out: ObjectOutputStream): Unit = Utils.tryOrIOException {
+    val in = new FileInputStream(new File(path))
+    try {
+      Utils.copyStream(in, out)
+    } finally {
+      in.close()
+    }
+  }
+
+  /**
+   * Write data into disk, using randomly generated name.
+   */
+  private def readObject(in: ObjectInputStream): Unit = Utils.tryOrIOException {
+    val dir = new File(Utils.getLocalDir(SparkEnv.get.conf))
+    val file = File.createTempFile("broadcast", "", dir)
+    path = file.getAbsolutePath
+    val out = new FileOutputStream(file)
+    try {
+      Utils.copyStream(in, out)
+    } finally {
+      out.close()
+    }
+  }
+
+  /**
+   * Delete the file once the object is GCed.
+   */
+  override def finalize() {
+    if (!path.isEmpty) {
+      val file = new File(path)
+      if (file.exists()) {
+        file.delete()
+      }
+    }
+  }
+}
diff --git a/python/pyspark/broadcast.py b/python/pyspark/broadcast.py
index 01cac3c72c..6b8a8b256a 100644
--- a/python/pyspark/broadcast.py
+++ b/python/pyspark/broadcast.py
@@ -15,21 +15,10 @@
 # limitations under the License.
 #
 
-"""
->>> from pyspark.context import SparkContext
->>> sc = SparkContext('local', 'test')
->>> b = sc.broadcast([1, 2, 3, 4, 5])
->>> b.value
-[1, 2, 3, 4, 5]
->>> sc.parallelize([0, 0]).flatMap(lambda x: b.value).collect()
-[1, 2, 3, 4, 5, 1, 2, 3, 4, 5]
->>> b.unpersist()
-
->>> large_broadcast = sc.broadcast(list(range(10000)))
-"""
 import os
-
-from pyspark.serializers import LargeObjectSerializer
+import cPickle
+import gc
+from tempfile import NamedTemporaryFile
 
 
 __all__ = ['Broadcast']
@@ -49,44 +38,88 @@ def _from_id(bid):
 class Broadcast(object):
 
     """
-    A broadcast variable created with
-    L{SparkContext.broadcast()<pyspark.context.SparkContext.broadcast>}.
+    A broadcast variable created with L{SparkContext.broadcast()}.
     Access its value through C{.value}.
+
+    Examples:
+
+    >>> from pyspark.context import SparkContext
+    >>> sc = SparkContext('local', 'test')
+    >>> b = sc.broadcast([1, 2, 3, 4, 5])
+    >>> b.value
+    [1, 2, 3, 4, 5]
+    >>> sc.parallelize([0, 0]).flatMap(lambda x: b.value).collect()
+    [1, 2, 3, 4, 5, 1, 2, 3, 4, 5]
+    >>> b.unpersist()
+
+    >>> large_broadcast = sc.broadcast(range(10000))
     """
 
-    def __init__(self, bid, value, java_broadcast=None,
-                 pickle_registry=None, path=None):
+    def __init__(self, sc=None, value=None, pickle_registry=None, path=None):
         """
-        Should not be called directly by users -- use
-        L{SparkContext.broadcast()<pyspark.context.SparkContext.broadcast>}
+        Should not be called directly by users -- use L{SparkContext.broadcast()}
         instead.
         """
-        self.bid = bid
-        if path is None:
-            self._value = value
-        self._jbroadcast = java_broadcast
-        self._pickle_registry = pickle_registry
-        self.path = path
+        if sc is not None:
+            f = NamedTemporaryFile(delete=False, dir=sc._temp_dir)
+            self._path = self.dump(value, f)
+            self._jbroadcast = sc._jvm.PythonRDD.readBroadcastFromFile(sc._jsc, self._path)
+            self._pickle_registry = pickle_registry
+        else:
+            self._jbroadcast = None
+            self._path = path
+
+    def dump(self, value, f):
+        if isinstance(value, basestring):
+            if isinstance(value, unicode):
+                f.write('U')
+                value = value.encode('utf8')
+            else:
+                f.write('S')
+            f.write(value)
+        else:
+            f.write('P')
+            cPickle.dump(value, f, 2)
+        f.close()
+        return f.name
+
+    def load(self, path):
+        with open(path, 'rb', 1 << 20) as f:
+            flag = f.read(1)
+            data = f.read()
+            if flag == 'P':
+                # cPickle.loads() may create lots of objects, disable GC
+                # temporary for better performance
+                gc.disable()
+                try:
+                    return cPickle.loads(data)
+                finally:
+                    gc.enable()
+            else:
+                return data.decode('utf8') if flag == 'U' else data
 
     @property
     def value(self):
         """ Return the broadcasted value
         """
-        if not hasattr(self, "_value") and self.path is not None:
-            ser = LargeObjectSerializer()
-            self._value = ser.load_stream(open(self.path)).next()
+        if not hasattr(self, "_value") and self._path is not None:
+            self._value = self.load(self._path)
         return self._value
 
     def unpersist(self, blocking=False):
         """
         Delete cached copies of this broadcast on the executors.
         """
+        if self._jbroadcast is None:
+            raise Exception("Broadcast can only be unpersisted in driver")
         self._jbroadcast.unpersist(blocking)
-        os.unlink(self.path)
+        os.unlink(self._path)
 
     def __reduce__(self):
+        if self._jbroadcast is None:
+            raise Exception("Broadcast can only be serialized in driver")
         self._pickle_registry.add(self)
-        return (_from_id, (self.bid, ))
+        return _from_id, (self._jbroadcast.id(),)
 
 
 if __name__ == "__main__":
diff --git a/python/pyspark/context.py b/python/pyspark/context.py
index ec67ec8d0f..ed7351d60c 100644
--- a/python/pyspark/context.py
+++ b/python/pyspark/context.py
@@ -29,7 +29,7 @@ from pyspark.conf import SparkConf
 from pyspark.files import SparkFiles
 from pyspark.java_gateway import launch_gateway
 from pyspark.serializers import PickleSerializer, BatchedSerializer, UTF8Deserializer, \
-    PairDeserializer, AutoBatchedSerializer, NoOpSerializer, LargeObjectSerializer
+    PairDeserializer, AutoBatchedSerializer, NoOpSerializer
 from pyspark.storagelevel import StorageLevel
 from pyspark.rdd import RDD
 from pyspark.traceback_utils import CallSite, first_spark_call
@@ -624,15 +624,7 @@ class SparkContext(object):
         object for reading it in distributed functions. The variable will
         be sent to each cluster only once.
         """
-        ser = LargeObjectSerializer()
-
-        # pass large object by py4j is very slow and need much memory
-        tempFile = NamedTemporaryFile(delete=False, dir=self._temp_dir)
-        ser.dump_stream([value], tempFile)
-        tempFile.close()
-        jbroadcast = self._jvm.PythonRDD.readBroadcastFromFile(self._jsc, tempFile.name)
-        return Broadcast(jbroadcast.id(), None, jbroadcast,
-                         self._pickled_broadcast_vars, tempFile.name)
+        return Broadcast(self, value, self._pickled_broadcast_vars)
 
     def accumulator(self, value, accum_param=None):
         """
diff --git a/python/pyspark/serializers.py b/python/pyspark/serializers.py
index 760a509f0e..33aa55f7f1 100644
--- a/python/pyspark/serializers.py
+++ b/python/pyspark/serializers.py
@@ -448,184 +448,20 @@ class AutoSerializer(FramedSerializer):
             raise ValueError("invalid sevialization type: %s" % _type)
 
 
-class SizeLimitedStream(object):
-    """
-    Read at most `limit` bytes from underlying stream
-
-    >>> from StringIO import StringIO
-    >>> io = StringIO()
-    >>> io.write("Hello world")
-    >>> io.seek(0)
-    >>> lio = SizeLimitedStream(io, 5)
-    >>> lio.read()
-    'Hello'
-    """
-    def __init__(self, stream, limit):
-        self.stream = stream
-        self.limit = limit
-
-    def read(self, n=0):
-        if n > self.limit or n == 0:
-            n = self.limit
-        buf = self.stream.read(n)
-        self.limit -= len(buf)
-        return buf
-
-
-class CompressedStream(object):
-    """
-    Compress the data using zlib
-
-    >>> from StringIO import StringIO
-    >>> io = StringIO()
-    >>> wio = CompressedStream(io, 'w')
-    >>> wio.write("Hello world")
-    >>> wio.flush()
-    >>> io.seek(0)
-    >>> rio = CompressedStream(io, 'r')
-    >>> rio.read()
-    'Hello world'
-    >>> rio.read()
-    ''
-    """
-    MAX_BATCH = 1 << 20  # 1MB
-
-    def __init__(self, stream, mode='w', level=1):
-        self.stream = stream
-        self.mode = mode
-        if mode == 'w':
-            self.compresser = zlib.compressobj(level)
-        elif mode == 'r':
-            self.decompresser = zlib.decompressobj()
-            self.buf = ''
-        else:
-            raise ValueError("can only support mode 'w' or 'r' ")
-
-    def write(self, buf):
-        assert self.mode == 'w', "It's not opened for write"
-        if len(buf) > self.MAX_BATCH:
-            # zlib can not compress string larger than 2G
-            batches = len(buf) / self.MAX_BATCH + 1  # last one may be empty
-            for i in xrange(batches):
-                self.write(buf[i * self.MAX_BATCH:(i + 1) * self.MAX_BATCH])
-        else:
-            compressed = self.compresser.compress(buf)
-            self.stream.write(compressed)
-
-    def flush(self, mode=zlib.Z_FULL_FLUSH):
-        if self.mode == 'w':
-            d = self.compresser.flush(mode)
-            self.stream.write(d)
-            self.stream.flush()
-
-    def close(self):
-        if self.mode == 'w':
-            self.flush(zlib.Z_FINISH)
-            self.stream.close()
-
-    def read(self, size=0):
-        assert self.mode == 'r', "It's not opened for read"
-        if not size:
-            data = self.stream.read()
-            result = self.decompresser.decompress(data)
-            last = self.decompresser.flush()
-            return self.buf + result + last
-
-        # fast path for small read()
-        if size <= len(self.buf):
-            result = self.buf[:size]
-            self.buf = self.buf[size:]
-            return result
-
-        result = [self.buf]
-        size -= len(self.buf)
-        self.buf = ''
-        while size:
-            need = min(size, self.MAX_BATCH)
-            input = self.stream.read(need)
-            if input:
-                buf = self.decompresser.decompress(input)
-            else:
-                buf = self.decompresser.flush()
-
-            if len(buf) >= size:
-                self.buf = buf[size:]
-                result.append(buf[:size])
-                return ''.join(result)
-
-            size -= len(buf)
-            result.append(buf)
-            if not input:
-                return ''.join(result)
-
-    def readline(self):
-        """
-        This is needed for pickle, but not used in protocol 2
-        """
-        line = []
-        b = self.read(1)
-        while b and b != '\n':
-            line.append(b)
-            b = self.read(1)
-        line.append(b)
-        return ''.join(line)
-
-
-class LargeObjectSerializer(Serializer):
-    """
-    Serialize large object which could be larger than 2G
-
-    It uses cPickle to serialize the objects
-    """
-    def dump_stream(self, iterator, stream):
-        stream = CompressedStream(stream, 'w')
-        for value in iterator:
-            if isinstance(value, basestring):
-                if isinstance(value, unicode):
-                    stream.write('U')
-                    value = value.encode("utf-8")
-                else:
-                    stream.write('S')
-                write_long(len(value), stream)
-                stream.write(value)
-            else:
-                stream.write('P')
-                cPickle.dump(value, stream, 2)
-        stream.flush()
-
-    def load_stream(self, stream):
-        stream = CompressedStream(stream, 'r')
-        while True:
-            type = stream.read(1)
-            if not type:
-                return
-            if type in ('S', 'U'):
-                length = read_long(stream)
-                value = stream.read(length)
-                if type == 'U':
-                    value = value.decode('utf-8')
-                yield value
-            elif type == 'P':
-                yield cPickle.load(stream)
-            else:
-                raise ValueError("unknown type: %s" % type)
-
-
-class CompressedSerializer(Serializer):
+class CompressedSerializer(FramedSerializer):
     """
     Compress the serialized data
     """
     def __init__(self, serializer):
+        FramedSerializer.__init__(self)
+        assert isinstance(serializer, FramedSerializer), "serializer must be a FramedSerializer"
         self.serializer = serializer
 
-    def load_stream(self, stream):
-        stream = CompressedStream(stream, "r")
-        return self.serializer.load_stream(stream)
+    def dumps(self, obj):
+        return zlib.compress(self.serializer.dumps(obj), 1)
 
-    def dump_stream(self, iterator, stream):
-        stream = CompressedStream(stream, "w")
-        self.serializer.dump_stream(iterator, stream)
-        stream.flush()
+    def loads(self, obj):
+        return self.serializer.loads(zlib.decompress(obj))
 
 
 class UTF8Deserializer(Serializer):
diff --git a/python/pyspark/tests.py b/python/pyspark/tests.py
index 29bcd38908..32645778c2 100644
--- a/python/pyspark/tests.py
+++ b/python/pyspark/tests.py
@@ -48,7 +48,7 @@ from pyspark.conf import SparkConf
 from pyspark.context import SparkContext
 from pyspark.files import SparkFiles
 from pyspark.serializers import read_int, BatchedSerializer, MarshalSerializer, PickleSerializer, \
-    CloudPickleSerializer, SizeLimitedStream, CompressedSerializer, LargeObjectSerializer
+    CloudPickleSerializer, CompressedSerializer
 from pyspark.shuffle import Aggregator, InMemoryMerger, ExternalMerger, ExternalSorter
 from pyspark.sql import SQLContext, IntegerType, Row, ArrayType, StructType, StructField, \
     UserDefinedType, DoubleType
@@ -237,26 +237,16 @@ class SerializationTestCase(unittest.TestCase):
         self.assertTrue("exit" in foo.func_code.co_names)
         ser.dumps(foo)
 
-    def _test_serializer(self, ser):
+    def test_compressed_serializer(self):
+        ser = CompressedSerializer(PickleSerializer())
         from StringIO import StringIO
         io = StringIO()
         ser.dump_stream(["abc", u"123", range(5)], io)
         io.seek(0)
         self.assertEqual(["abc", u"123", range(5)], list(ser.load_stream(io)))
-        size = io.tell()
         ser.dump_stream(range(1000), io)
         io.seek(0)
-        first = SizeLimitedStream(io, size)
-        self.assertEqual(["abc", u"123", range(5)], list(ser.load_stream(first)))
-        self.assertEqual(range(1000), list(ser.load_stream(io)))
-
-    def test_compressed_serializer(self):
-        ser = CompressedSerializer(PickleSerializer())
-        self._test_serializer(ser)
-
-    def test_large_object_serializer(self):
-        ser = LargeObjectSerializer()
-        self._test_serializer(ser)
+        self.assertEqual(["abc", u"123", range(5)] + range(1000), list(ser.load_stream(io)))
 
 
 class PySparkTestCase(unittest.TestCase):
diff --git a/python/pyspark/worker.py b/python/pyspark/worker.py
index e1552a0b0b..7e5343c973 100644
--- a/python/pyspark/worker.py
+++ b/python/pyspark/worker.py
@@ -30,8 +30,7 @@ from pyspark.accumulators import _accumulatorRegistry
 from pyspark.broadcast import Broadcast, _broadcastRegistry
 from pyspark.files import SparkFiles
 from pyspark.serializers import write_with_length, write_int, read_long, \
-    write_long, read_int, SpecialLengths, UTF8Deserializer, PickleSerializer, \
-    SizeLimitedStream, LargeObjectSerializer
+    write_long, read_int, SpecialLengths, UTF8Deserializer, PickleSerializer
 from pyspark import shuffle
 
 pickleSer = PickleSerializer()
@@ -78,14 +77,11 @@ def main(infile, outfile):
 
         # fetch names and values of broadcast variables
         num_broadcast_variables = read_int(infile)
-        bser = LargeObjectSerializer()
         for _ in range(num_broadcast_variables):
             bid = read_long(infile)
             if bid >= 0:
-                size = read_long(infile)
-                s = SizeLimitedStream(infile, size)
-                value = list((bser.load_stream(s)))[0]  # read out all the bytes
-                _broadcastRegistry[bid] = Broadcast(bid, value)
+                path = utf8_deserializer.loads(infile)
+                _broadcastRegistry[bid] = Broadcast(path=path)
             else:
                 bid = - bid - 1
                 _broadcastRegistry.pop(bid)
diff --git a/sql/core/src/main/scala/org/apache/spark/sql/UdfRegistration.scala b/sql/core/src/main/scala/org/apache/spark/sql/UdfRegistration.scala
index ddcb5db6c3..00d6b43a57 100644
--- a/sql/core/src/main/scala/org/apache/spark/sql/UdfRegistration.scala
+++ b/sql/core/src/main/scala/org/apache/spark/sql/UdfRegistration.scala
@@ -20,6 +20,7 @@ package org.apache.spark.sql
 import java.util.{List => JList, Map => JMap}
 
 import org.apache.spark.Accumulator
+import org.apache.spark.api.python.PythonBroadcast
 import org.apache.spark.broadcast.Broadcast
 import org.apache.spark.sql.catalyst.ScalaReflection
 import org.apache.spark.sql.catalyst.expressions.{Expression, ScalaUdf}
@@ -39,7 +40,7 @@ private[sql] trait UDFRegistration {
       envVars: JMap[String, String],
       pythonIncludes: JList[String],
       pythonExec: String,
-      broadcastVars: JList[Broadcast[Array[Array[Byte]]]],
+      broadcastVars: JList[Broadcast[PythonBroadcast]],
       accumulator: Accumulator[JList[Array[Byte]]],
       stringDataType: String): Unit = {
     log.debug(
diff --git a/sql/core/src/main/scala/org/apache/spark/sql/execution/pythonUdfs.scala b/sql/core/src/main/scala/org/apache/spark/sql/execution/pythonUdfs.scala
index f98cae3f17..2b4a88d5e8 100644
--- a/sql/core/src/main/scala/org/apache/spark/sql/execution/pythonUdfs.scala
+++ b/sql/core/src/main/scala/org/apache/spark/sql/execution/pythonUdfs.scala
@@ -26,7 +26,7 @@ import scala.collection.JavaConverters._
 
 import net.razorvine.pickle.{Pickler, Unpickler}
 import org.apache.spark.annotation.DeveloperApi
-import org.apache.spark.api.python.PythonRDD
+import org.apache.spark.api.python.{PythonBroadcast, PythonRDD}
 import org.apache.spark.broadcast.Broadcast
 import org.apache.spark.sql.catalyst.expressions.Row
 import org.apache.spark.sql.catalyst.expressions._
@@ -45,7 +45,7 @@ private[spark] case class PythonUDF(
     envVars: JMap[String, String],
     pythonIncludes: JList[String],
     pythonExec: String,
-    broadcastVars: JList[Broadcast[Array[Array[Byte]]]],
+    broadcastVars: JList[Broadcast[PythonBroadcast]],
     accumulator: Accumulator[JList[Array[Byte]]],
     dataType: DataType,
     children: Seq[Expression]) extends Expression with SparkLogging {