diff options
Diffstat (limited to 'docs/streaming-programming-guide.md')
| -rw-r--r-- | docs/streaming-programming-guide.md | 45 |
1 files changed, 28 insertions, 17 deletions
diff --git a/docs/streaming-programming-guide.md b/docs/streaming-programming-guide.md index 8cd1b0cd66..c7df172024 100644 --- a/docs/streaming-programming-guide.md +++ b/docs/streaming-programming-guide.md @@ -13,6 +13,14 @@ A Spark Streaming application is very similar to a Spark application; it consist This guide shows some how to start programming with DStreams. +# Linking with Spark Streaming + +Add the following SBT or Maven dependency to your project to use Spark Streaming: + + groupId = org.apache.spark + artifactId = spark-streaming_{{site.SCALA_VERSION}} + version = {{site.SPARK_VERSION}} + # Initializing Spark Streaming The first thing a Spark Streaming program must do is create a `StreamingContext` object, which tells Spark how to access a cluster. A `StreamingContext` can be created by using @@ -34,7 +42,7 @@ ssc.textFileStream(directory) // Creates a stream by monitoring and process ssc.socketStream(hostname, port) // Creates a stream that uses a TCP socket to read data from hostname:port {% endhighlight %} -We also provide a input streams for Kafka, Flume, Akka actor, etc. For a complete list of input streams, take a look at the [StreamingContext API documentation](api/streaming/index.html#spark.streaming.StreamingContext). +We also provide a input streams for Kafka, Flume, Akka actor, etc. For a complete list of input streams, take a look at the [StreamingContext API documentation](api/streaming/index.html#org.apache.spark.streaming.StreamingContext). @@ -156,7 +164,7 @@ Spark Streaming features windowed computations, which allow you to apply transfo </table> -A complete list of DStream operations is available in the API documentation of [DStream](api/streaming/index.html#spark.streaming.DStream) and [PairDStreamFunctions](api/streaming/index.html#spark.streaming.PairDStreamFunctions). +A complete list of DStream operations is available in the API documentation of [DStream](api/streaming/index.html#org.apache.spark.streaming.DStream) and [PairDStreamFunctions](api/streaming/index.html#org.apache.spark.streaming.PairDStreamFunctions). ## Output Operations When an output operator is called, it triggers the computation of a stream. Currently the following output operators are defined: @@ -203,11 +211,11 @@ ssc.stop() {% endhighlight %} # Example -A simple example to start off is the [NetworkWordCount](https://github.com/mesos/spark/tree/master/examples/src/main/scala/spark/streaming/examples/NetworkWordCount.scala). This example counts the words received from a network server every second. Given below is the relevant sections of the source code. You can find the full source code in `<Spark repo>/streaming/src/main/scala/spark/streaming/examples/NetworkWordCount.scala` . +A simple example to start off is the [NetworkWordCount](https://github.com/apache/incubator-spark/tree/master/examples/src/main/scala/spark/streaming/examples/NetworkWordCount.scala). This example counts the words received from a network server every second. Given below is the relevant sections of the source code. You can find the full source code in `<Spark repo>/streaming/src/main/scala/spark/streaming/examples/NetworkWordCount.scala` . {% highlight scala %} -import spark.streaming.{Seconds, StreamingContext} -import spark.streaming.StreamingContext._ +import org.apache.spark.streaming.{Seconds, StreamingContext} +import StreamingContext._ ... // Create the context and set up a network input stream to receive from a host:port @@ -234,7 +242,7 @@ $ nc -lk 9999 Then, in a different terminal, you can start NetworkWordCount by using {% highlight bash %} -$ ./run spark.streaming.examples.NetworkWordCount local[2] localhost 9999 +$ ./run-example org.apache.spark.streaming.examples.NetworkWordCount local[2] localhost 9999 {% endhighlight %} This will make NetworkWordCount connect to the netcat server. Any lines typed in the terminal running the netcat server will be counted and printed on screen. @@ -272,7 +280,7 @@ Time: 1357008430000 ms </td> </table> -You can find more examples in `<Spark repo>/streaming/src/main/scala/spark/streaming/examples/`. They can be run in the similar manner using `./run spark.streaming.examples....` . Executing without any parameter would give the required parameter list. Further explanation to run them can be found in comments in the files. +You can find more examples in `<Spark repo>/streaming/src/main/scala/spark/streaming/examples/`. They can be run in the similar manner using `./run-example org.apache.spark.streaming.examples....` . Executing without any parameter would give the required parameter list. Further explanation to run them can be found in comments in the files. # DStream Persistence Similar to RDDs, DStreams also allow developers to persist the stream's data in memory. That is, using `persist()` method on a DStream would automatically persist every RDD of that DStream in memory. This is useful if the data in the DStream will be computed multiple times (e.g., multiple operations on the same data). For window-based operations like `reduceByWindow` and `reduceByKeyAndWindow` and state-based operations like `updateStateByKey`, this is implicitly true. Hence, DStreams generated by window-based operations are automatically persisted in memory, without the developer calling `persist()`. @@ -301,6 +309,9 @@ dstream.checkpoint(checkpointInterval) // checkpointInterval must be a multiple For DStreams that must be checkpointed (that is, DStreams created by `updateStateByKey` and `reduceByKeyAndWindow` with inverse function), the checkpoint interval of the DStream is by default set to a multiple of the DStream's sliding interval such that its at least 10 seconds. +## Custom Receivers +Spark comes with a built in support for most common usage scenarios where input stream source can be either a network socket stream to support for a few message queues. Apart from that it is also possible to supply your own custom receiver via a convenient API. Find more details at [Custom Receiver Guide](streaming-custom-receivers.html). + # Performance Tuning Getting the best performance of a Spark Streaming application on a cluster requires a bit of tuning. This section explains a number of the parameters and configurations that can tuned to improve the performance of you application. At a high level, you need to consider two things: <ol> @@ -312,7 +323,7 @@ Getting the best performance of a Spark Streaming application on a cluster requi There are a number of optimizations that can be done in Spark to minimize the processing time of each batch. These have been discussed in detail in [Tuning Guide](tuning.html). This section highlights some of the most important ones. ### Level of Parallelism -Cluster resources maybe under-utilized if the number of parallel tasks used in any stage of the computation is not high enough. For example, for distributed reduce operations like `reduceByKey` and `reduceByKeyAndWindow`, the default number of parallel tasks is 8. You can pass the level of parallelism as an argument (see the [`spark.PairDStreamFunctions`](api/streaming/index.html#spark.PairDStreamFunctions) documentation), or set the system property `spark.default.parallelism` to change the default. +Cluster resources maybe under-utilized if the number of parallel tasks used in any stage of the computation is not high enough. For example, for distributed reduce operations like `reduceByKey` and `reduceByKeyAndWindow`, the default number of parallel tasks is 8. You can pass the level of parallelism as an argument (see the [`PairDStreamFunctions`](api/streaming/index.html#org.apache.spark.PairDStreamFunctions) documentation), or set the system property `spark.default.parallelism` to change the default. ### Data Serialization The overhead of data serialization can be significant, especially when sub-second batch sizes are to be achieved. There are two aspects to it. @@ -342,7 +353,7 @@ This value is closely tied with any window operation that is being used. Any win ## Memory Tuning Tuning the memory usage and GC behavior of Spark applications have been discussed in great detail in the [Tuning Guide](tuning.html). It is recommended that you read that. In this section, we highlight a few customizations that are strongly recommended to minimize GC related pauses in Spark Streaming applications and achieving more consistent batch processing times. -* **Default persistence level of DStreams**: Unlike RDDs, the default persistence level of DStreams serializes the data in memory (that is, [StorageLevel.MEMORY_ONLY_SER](api/core/index.html#spark.storage.StorageLevel$) for DStream compared to [StorageLevel.MEMORY_ONLY](api/core/index.html#spark.storage.StorageLevel$) for RDDs). Even though keeping the data serialized incurs a higher serialization overheads, it significantly reduces GC pauses. +* **Default persistence level of DStreams**: Unlike RDDs, the default persistence level of DStreams serializes the data in memory (that is, [StorageLevel.MEMORY_ONLY_SER](api/core/index.html#org.apache.spark.storage.StorageLevel$) for DStream compared to [StorageLevel.MEMORY_ONLY](api/core/index.html#org.apache.spark.storage.StorageLevel$) for RDDs). Even though keeping the data serialized incurs a higher serialization overheads, it significantly reduces GC pauses. * **Concurrent garbage collector**: Using the concurrent mark-and-sweep GC further minimizes the variability of GC pauses. Even though concurrent GC is known to reduce the overall processing throughput of the system, its use is still recommended to achieve more consistent batch processing times. @@ -464,10 +475,10 @@ If the driver had crashed in the middle of the processing of time 3, then it wil # Java API -Similar to [Spark's Java API](java-programming-guide.html), we also provide a Java API for Spark Streaming which allows all its features to be accessible from a Java program. This is defined in [spark.streaming.api.java] (api/streaming/index.html#spark.streaming.api.java.package) package and includes [JavaStreamingContext](api/streaming/index.html#spark.streaming.api.java.JavaStreamingContext) and [JavaDStream](api/streaming/index.html#spark.streaming.api.java.JavaDStream) classes that provide the same methods as their Scala counterparts, but take Java functions (that is, Function, and Function2) and return Java data and collection types. Some of the key points to note are: +Similar to [Spark's Java API](java-programming-guide.html), we also provide a Java API for Spark Streaming which allows all its features to be accessible from a Java program. This is defined in [org.apache.spark.streaming.api.java] (api/streaming/index.html#org.apache.spark.streaming.api.java.package) package and includes [JavaStreamingContext](api/streaming/index.html#org.apache.spark.streaming.api.java.JavaStreamingContext) and [JavaDStream](api/streaming/index.html#org.apache.spark.streaming.api.java.JavaDStream) classes that provide the same methods as their Scala counterparts, but take Java functions (that is, Function, and Function2) and return Java data and collection types. Some of the key points to note are: -1. Functions for transformations must be implemented as subclasses of [Function](api/core/index.html#spark.api.java.function.Function) and [Function2](api/core/index.html#spark.api.java.function.Function2) -1. Unlike the Scala API, the Java API handles DStreams for key-value pairs using a separate [JavaPairDStream](api/streaming/index.html#spark.streaming.api.java.JavaPairDStream) class(similar to [JavaRDD and JavaPairRDD](java-programming-guide.html#rdd-classes). DStream functions like `map` and `filter` are implemented separately by JavaDStreams and JavaPairDStream to return DStreams of appropriate types. +1. Functions for transformations must be implemented as subclasses of [Function](api/core/index.html#org.apache.spark.api.java.function.Function) and [Function2](api/core/index.html#org.apache.spark.api.java.function.Function2) +1. Unlike the Scala API, the Java API handles DStreams for key-value pairs using a separate [JavaPairDStream](api/streaming/index.html#org.apache.spark.streaming.api.java.JavaPairDStream) class(similar to [JavaRDD and JavaPairRDD](java-programming-guide.html#rdd-classes). DStream functions like `map` and `filter` are implemented separately by JavaDStreams and JavaPairDStream to return DStreams of appropriate types. Spark's [Java Programming Guide](java-programming-guide.html) gives more ideas about using the Java API. To extends the ideas presented for the RDDs to DStreams, we present parts of the Java version of the same NetworkWordCount example presented above. The full source code is given at `<spark repo>/examples/src/main/java/spark/streaming/examples/JavaNetworkWordCount.java` @@ -479,7 +490,7 @@ JavaDStream<String> lines = ssc.socketTextStream(ip, port); {% endhighlight %} -Then the `lines` are split into words by using the `flatMap` function and [FlatMapFunction](api/core/index.html#spark.api.java.function.FlatMapFunction). +Then the `lines` are split into words by using the `flatMap` function and [FlatMapFunction](api/core/index.html#org.apache.spark.api.java.function.FlatMapFunction). {% highlight java %} JavaDStream<String> words = lines.flatMap( @@ -491,7 +502,7 @@ JavaDStream<String> words = lines.flatMap( }); {% endhighlight %} -The `words` is then mapped to a [JavaPairDStream](api/streaming/index.html#spark.streaming.api.java.JavaPairDStream) of `(word, 1)` pairs using `map` and [PairFunction](api/core/index.html#spark.api.java.function.PairFunction). This is reduced by using `reduceByKey` and [Function2](api/core/index.html#spark.api.java.function.Function2). +The `words` is then mapped to a [JavaPairDStream](api/streaming/index.html#org.apache.spark.streaming.api.java.JavaPairDStream) of `(word, 1)` pairs using `map` and [PairFunction](api/core/index.html#org.apache.spark.api.java.function.PairFunction). This is reduced by using `reduceByKey` and [Function2](api/core/index.html#org.apache.spark.api.java.function.Function2). {% highlight java %} JavaPairDStream<String, Integer> wordCounts = words.map( @@ -510,8 +521,8 @@ JavaPairDStream<String, Integer> wordCounts = words.map( {% endhighlight %} - # Where to Go from Here -* API docs - [Scala](api/streaming/index.html#spark.streaming.package) and [Java](api/streaming/index.html#spark.streaming.api.java.package) -* More examples - [Scala](https://github.com/mesos/spark/tree/master/examples/src/main/scala/spark/streaming/examples) and [Java](https://github.com/mesos/spark/tree/master/examples/src/main/java/spark/streaming/examples) + +* API docs - [Scala](api/streaming/index.html#org.apache.spark.streaming.package) and [Java](api/streaming/index.html#org.apache.spark.streaming.api.java.package) +* More examples - [Scala](https://github.com/apache/incubator-spark/tree/master/examples/src/main/scala/spark/streaming/examples) and [Java](https://github.com/apache/incubator-spark/tree/master/examples/src/main/java/spark/streaming/examples) * [Paper describing Spark Streaming](http://www.eecs.berkeley.edu/Pubs/TechRpts/2012/EECS-2012-259.pdf) |