http://git-wip-us.apache.org/repos/asf/spark-website/blob/d2bcf185/site/docs/2.1.0/structured-streaming-programming-guide.html ---------------------------------------------------------------------- diff --git a/site/docs/2.1.0/structured-streaming-programming-guide.html b/site/docs/2.1.0/structured-streaming-programming-guide.html index e54c101..3a1ac5f 100644 --- a/site/docs/2.1.0/structured-streaming-programming-guide.html +++ b/site/docs/2.1.0/structured-streaming-programming-guide.html @@ -127,45 +127,50 @@ <ul id="markdown-toc"> - <li><a href="#overview" id="markdown-toc-overview">Overview</a></li> - <li><a href="#quick-example" id="markdown-toc-quick-example">Quick Example</a></li> - <li><a href="#programming-model" id="markdown-toc-programming-model">Programming Model</a> <ul> - <li><a href="#basic-concepts" id="markdown-toc-basic-concepts">Basic Concepts</a></li> - <li><a href="#handling-event-time-and-late-data" id="markdown-toc-handling-event-time-and-late-data">Handling Event-time and Late Data</a></li> - <li><a href="#fault-tolerance-semantics" id="markdown-toc-fault-tolerance-semantics">Fault Tolerance Semantics</a></li> + <li><a href="#overview">Overview</a></li> + <li><a href="#quick-example">Quick Example</a></li> + <li><a href="#programming-model">Programming Model</a> <ul> + <li><a href="#basic-concepts">Basic Concepts</a></li> + <li><a href="#handling-event-time-and-late-data">Handling Event-time and Late Data</a></li> + <li><a href="#fault-tolerance-semantics">Fault Tolerance Semantics</a></li> </ul> </li> - <li><a href="#api-using-datasets-and-dataframes" id="markdown-toc-api-using-datasets-and-dataframes">API using Datasets and DataFrames</a> <ul> - <li><a href="#creating-streaming-dataframes-and-streaming-datasets" id="markdown-toc-creating-streaming-dataframes-and-streaming-datasets">Creating streaming DataFrames and streaming Datasets</a> <ul> - <li><a href="#data-sources" id="markdown-toc-data-sources">Data Sources</a></li> - <li><a href="#schema-inference-and-partition-of-streaming-dataframesdatasets" id="markdown-toc-schema-inference-and-partition-of-streaming-dataframesdatasets">Schema inference and partition of streaming DataFrames/Datasets</a></li> + <li><a href="#api-using-datasets-and-dataframes">API using Datasets and DataFrames</a> <ul> + <li><a href="#creating-streaming-dataframes-and-streaming-datasets">Creating streaming DataFrames and streaming Datasets</a> <ul> + <li><a href="#data-sources">Data Sources</a></li> + <li><a href="#schema-inference-and-partition-of-streaming-dataframesdatasets">Schema inference and partition of streaming DataFrames/Datasets</a></li> </ul> </li> - <li><a href="#operations-on-streaming-dataframesdatasets" id="markdown-toc-operations-on-streaming-dataframesdatasets">Operations on streaming DataFrames/Datasets</a> <ul> - <li><a href="#basic-operations---selection-projection-aggregation" id="markdown-toc-basic-operations---selection-projection-aggregation">Basic Operations - Selection, Projection, Aggregation</a></li> - <li><a href="#window-operations-on-event-time" id="markdown-toc-window-operations-on-event-time">Window Operations on Event Time</a></li> - <li><a href="#join-operations" id="markdown-toc-join-operations">Join Operations</a></li> - <li><a href="#unsupported-operations" id="markdown-toc-unsupported-operations">Unsupported Operations</a></li> + <li><a href="#operations-on-streaming-dataframesdatasets">Operations on streaming DataFrames/Datasets</a> <ul> + <li><a href="#basic-operations---selection-projection-aggregation">Basic Operations - Selection, Projection, Aggregation</a></li> + <li><a href="#window-operations-on-event-time">Window Operations on Event Time</a></li> + <li><a href="#handling-late-data-and-watermarking">Handling Late Data and Watermarking</a></li> + <li><a href="#join-operations">Join Operations</a></li> + <li><a href="#unsupported-operations">Unsupported Operations</a></li> </ul> </li> - <li><a href="#starting-streaming-queries" id="markdown-toc-starting-streaming-queries">Starting Streaming Queries</a> <ul> - <li><a href="#output-modes" id="markdown-toc-output-modes">Output Modes</a></li> - <li><a href="#output-sinks" id="markdown-toc-output-sinks">Output Sinks</a></li> - <li><a href="#using-foreach" id="markdown-toc-using-foreach">Using Foreach</a></li> + <li><a href="#starting-streaming-queries">Starting Streaming Queries</a> <ul> + <li><a href="#output-modes">Output Modes</a></li> + <li><a href="#output-sinks">Output Sinks</a></li> + <li><a href="#using-foreach">Using Foreach</a></li> </ul> </li> - <li><a href="#managing-streaming-queries" id="markdown-toc-managing-streaming-queries">Managing Streaming Queries</a></li> - <li><a href="#monitoring-streaming-queries" id="markdown-toc-monitoring-streaming-queries">Monitoring Streaming Queries</a></li> - <li><a href="#recovering-from-failures-with-checkpointing" id="markdown-toc-recovering-from-failures-with-checkpointing">Recovering from Failures with Checkpointing</a></li> + <li><a href="#managing-streaming-queries">Managing Streaming Queries</a></li> + <li><a href="#monitoring-streaming-queries">Monitoring Streaming Queries</a> <ul> + <li><a href="#interactive-apis">Interactive APIs</a></li> + <li><a href="#asynchronous-api">Asynchronous API</a></li> + </ul> + </li> + <li><a href="#recovering-from-failures-with-checkpointing">Recovering from Failures with Checkpointing</a></li> </ul> </li> - <li><a href="#where-to-go-from-here" id="markdown-toc-where-to-go-from-here">Where to go from here</a></li> + <li><a href="#where-to-go-from-here">Where to go from here</a></li> </ul> <h1 id="overview">Overview</h1> <p>Structured Streaming is a scalable and fault-tolerant stream processing engine built on the Spark SQL engine. You can express your streaming computation the same way you would express a batch computation on static data.The Spark SQL engine will take care of running it incrementally and continuously and updating the final result as streaming data continues to arrive. You can use the <a href="sql-programming-guide.html">Dataset/DataFrame API</a> in Scala, Java or Python to express streaming aggregations, event-time windows, stream-to-batch joins, etc. The computation is executed on the same optimized Spark SQL engine. Finally, the system ensures end-to-end exactly-once fault-tolerance guarantees through checkpointing and Write Ahead Logs. In short, <em>Structured Streaming provides fast, scalable, fault-tolerant, end-to-end exactly-once stream processing without the user having to reason about streaming.</em></p> -<p><strong>Spark 2.0 is the ALPHA RELEASE of Structured Streaming</strong> and the APIs are still experimental. In this guide, we are going to walk you through the programming model and the APIs. First, let’s start with a simple example - a streaming word count.</p> +<p><strong>Structured Streaming is still ALPHA in Spark 2.1</strong> and the APIs are still experimental. In this guide, we are going to walk you through the programming model and the APIs. First, let’s start with a simple example - a streaming word count. </p> <h1 id="quick-example">Quick Example</h1> <p>Letâs say you want to maintain a running word count of text data received from a data server listening on a TCP socket. Letâs see how you can express this using Structured Streaming. You can see the full code in @@ -175,7 +180,7 @@ And if you <a href="http://spark.apache.org/downloads.html";>download Spark</a>, <div class="codetabs"> <div data-lang="scala"> - <div class="highlight"><pre><code class="language-scala" data-lang="scala"><span class="k">import</span> <span class="nn">org.apache.spark.sql.functions._</span> + <figure class="highlight"><pre><code class="language-scala" data-lang="scala"><span></span><span class="k">import</span> <span class="nn">org.apache.spark.sql.functions._</span> <span class="k">import</span> <span class="nn">org.apache.spark.sql.SparkSession</span> <span class="k">val</span> <span class="n">spark</span> <span class="k">=</span> <span class="nc">SparkSession</span> @@ -183,12 +188,12 @@ And if you <a href="http://spark.apache.org/downloads.html";>download Spark</a>, <span class="o">.</span><span class="n">appName</span><span class="o">(</span><span class="s">"StructuredNetworkWordCount"</span><span class="o">)</span> <span class="o">.</span><span class="n">getOrCreate</span><span class="o">()</span> -<span class="k">import</span> <span class="nn">spark.implicits._</span></code></pre></div> +<span class="k">import</span> <span class="nn">spark.implicits._</span></code></pre></figure> </div> <div data-lang="java"> - <div class="highlight"><pre><code class="language-java" data-lang="java"><span class="kn">import</span> <span class="nn">org.apache.spark.api.java.function.FlatMapFunction</span><span class="o">;</span> + <figure class="highlight"><pre><code class="language-java" data-lang="java"><span></span><span class="kn">import</span> <span class="nn">org.apache.spark.api.java.function.FlatMapFunction</span><span class="o">;</span> <span class="kn">import</span> <span class="nn">org.apache.spark.sql.*</span><span class="o">;</span> <span class="kn">import</span> <span class="nn">org.apache.spark.sql.streaming.StreamingQuery</span><span class="o">;</span> @@ -198,19 +203,19 @@ And if you <a href="http://spark.apache.org/downloads.html";>download Spark</a>, <span class="n">SparkSession</span> <span class="n">spark</span> <span class="o">=</span> <span class="n">SparkSession</span> <span class="o">.</span><span class="na">builder</span><span class="o">()</span> <span class="o">.</span><span class="na">appName</span><span class="o">(</span><span class="s">"JavaStructuredNetworkWordCount"</span><span class="o">)</span> - <span class="o">.</span><span class="na">getOrCreate</span><span class="o">();</span></code></pre></div> + <span class="o">.</span><span class="na">getOrCreate</span><span class="o">();</span></code></pre></figure> </div> <div data-lang="python"> - <div class="highlight"><pre><code class="language-python" data-lang="python"><span class="kn">from</span> <span class="nn">pyspark.sql</span> <span class="kn">import</span> <span class="n">SparkSession</span> + <figure class="highlight"><pre><code class="language-python" data-lang="python"><span></span><span class="kn">from</span> <span class="nn">pyspark.sql</span> <span class="kn">import</span> <span class="n">SparkSession</span> <span class="kn">from</span> <span class="nn">pyspark.sql.functions</span> <span class="kn">import</span> <span class="n">explode</span> <span class="kn">from</span> <span class="nn">pyspark.sql.functions</span> <span class="kn">import</span> <span class="n">split</span> <span class="n">spark</span> <span class="o">=</span> <span class="n">SparkSession</span> \ <span class="o">.</span><span class="n">builder</span> \ - <span class="o">.</span><span class="n">appName</span><span class="p">(</span><span class="s">"StructuredNetworkWordCount"</span><span class="p">)</span> \ - <span class="o">.</span><span class="n">getOrCreate</span><span class="p">()</span></code></pre></div> + <span class="o">.</span><span class="n">appName</span><span class="p">(</span><span class="s2">"StructuredNetworkWordCount"</span><span class="p">)</span> \ + <span class="o">.</span><span class="n">getOrCreate</span><span class="p">()</span></code></pre></figure> </div> </div> @@ -220,7 +225,7 @@ And if you <a href="http://spark.apache.org/downloads.html";>download Spark</a>, <div class="codetabs"> <div data-lang="scala"> - <div class="highlight"><pre><code class="language-scala" data-lang="scala"><span class="c1">// Create DataFrame representing the stream of input lines from connection to localhost:9999</span> + <figure class="highlight"><pre><code class="language-scala" data-lang="scala"><span></span><span class="c1">// Create DataFrame representing the stream of input lines from connection to localhost:9999</span> <span class="k">val</span> <span class="n">lines</span> <span class="k">=</span> <span class="n">spark</span><span class="o">.</span><span class="n">readStream</span> <span class="o">.</span><span class="n">format</span><span class="o">(</span><span class="s">"socket"</span><span class="o">)</span> <span class="o">.</span><span class="n">option</span><span class="o">(</span><span class="s">"host"</span><span class="o">,</span> <span class="s">"localhost"</span><span class="o">)</span> @@ -231,14 +236,14 @@ And if you <a href="http://spark.apache.org/downloads.html";>download Spark</a>, <span class="k">val</span> <span class="n">words</span> <span class="k">=</span> <span class="n">lines</span><span class="o">.</span><span class="n">as</span><span class="o">[</span><span class="kt">String</span><span class="o">].</span><span class="n">flatMap</span><span class="o">(</span><span class="k">_</span><span class="o">.</span><span class="n">split</span><span class="o">(</span><span class="s">" "</span><span class="o">))</span> <span class="c1">// Generate running word count</span> -<span class="k">val</span> <span class="n">wordCounts</span> <span class="k">=</span> <span class="n">words</span><span class="o">.</span><span class="n">groupBy</span><span class="o">(</span><span class="s">"value"</span><span class="o">).</span><span class="n">count</span><span class="o">()</span></code></pre></div> +<span class="k">val</span> <span class="n">wordCounts</span> <span class="k">=</span> <span class="n">words</span><span class="o">.</span><span class="n">groupBy</span><span class="o">(</span><span class="s">"value"</span><span class="o">).</span><span class="n">count</span><span class="o">()</span></code></pre></figure> <p>This <code>lines</code> DataFrame represents an unbounded table containing the streaming text data. This table contains one column of strings named “value”, and each line in the streaming text data becomes a row in the table. Note, that this is not currently receiving any data as we are just setting up the transformation, and have not yet started it. Next, we have converted the DataFrame to a Dataset of String using <code>.as[String]</code>, so that we can apply the <code>flatMap</code> operation to split each line into multiple words. The resultant <code>words</code> Dataset contains all the words. Finally, we have defined the <code>wordCounts</code> DataFrame by grouping by the unique values in the Dataset and counting them. Note that this is a streaming DataFrame which represents the running word counts of the stream.</p> </div> <div data-lang="java"> - <div class="highlight"><pre><code class="language-java" data-lang="java"><span class="c1">// Create DataFrame representing the stream of input lines from connection to localhost:9999</span> + <figure class="highlight"><pre><code class="language-java" data-lang="java"><span></span><span class="c1">// Create DataFrame representing the stream of input lines from connection to localhost:9999</span> <span class="n">Dataset</span><span class="o"><</span><span class="n">Row</span><span class="o">></span> <span class="n">lines</span> <span class="o">=</span> <span class="n">spark</span> <span class="o">.</span><span class="na">readStream</span><span class="o">()</span> <span class="o">.</span><span class="na">format</span><span class="o">(</span><span class="s">"socket"</span><span class="o">)</span> @@ -258,30 +263,30 @@ And if you <a href="http://spark.apache.org/downloads.html";>download Spark</a>, <span class="o">},</span> <span class="n">Encoders</span><span class="o">.</span><span class="na">STRING</span><span class="o">());</span> <span class="c1">// Generate running word count</span> -<span class="n">Dataset</span><span class="o"><</span><span class="n">Row</span><span class="o">></span> <span class="n">wordCounts</span> <span class="o">=</span> <span class="n">words</span><span class="o">.</span><span class="na">groupBy</span><span class="o">(</span><span class="s">"value"</span><span class="o">).</span><span class="na">count</span><span class="o">();</span></code></pre></div> +<span class="n">Dataset</span><span class="o"><</span><span class="n">Row</span><span class="o">></span> <span class="n">wordCounts</span> <span class="o">=</span> <span class="n">words</span><span class="o">.</span><span class="na">groupBy</span><span class="o">(</span><span class="s">"value"</span><span class="o">).</span><span class="na">count</span><span class="o">();</span></code></pre></figure> <p>This <code>lines</code> DataFrame represents an unbounded table containing the streaming text data. This table contains one column of strings named “value”, and each line in the streaming text data becomes a row in the table. Note, that this is not currently receiving any data as we are just setting up the transformation, and have not yet started it. Next, we have converted the DataFrame to a Dataset of String using <code>.as(Encoders.STRING())</code>, so that we can apply the <code>flatMap</code> operation to split each line into multiple words. The resultant <code>words</code> Dataset contains all the words. Finally, we have defined the <code>wordCounts</code> DataFrame by grouping by the unique values in the Dataset and counting them. Note that this is a streaming DataFrame which represents the running word counts of the stream.</p> </div> <div data-lang="python"> - <div class="highlight"><pre><code class="language-python" data-lang="python"><span class="c"># Create DataFrame representing the stream of input lines from connection to localhost:9999</span> + <figure class="highlight"><pre><code class="language-python" data-lang="python"><span></span><span class="c1"># Create DataFrame representing the stream of input lines from connection to localhost:9999</span> <span class="n">lines</span> <span class="o">=</span> <span class="n">spark</span> \ <span class="o">.</span><span class="n">readStream</span> \ - <span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="s">"socket"</span><span class="p">)</span> \ - <span class="o">.</span><span class="n">option</span><span class="p">(</span><span class="s">"host"</span><span class="p">,</span> <span class="s">"localhost"</span><span class="p">)</span> \ - <span class="o">.</span><span class="n">option</span><span class="p">(</span><span class="s">"port"</span><span class="p">,</span> <span class="mi">9999</span><span class="p">)</span> \ + <span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="s2">"socket"</span><span class="p">)</span> \ + <span class="o">.</span><span class="n">option</span><span class="p">(</span><span class="s2">"host"</span><span class="p">,</span> <span class="s2">"localhost"</span><span class="p">)</span> \ + <span class="o">.</span><span class="n">option</span><span class="p">(</span><span class="s2">"port"</span><span class="p">,</span> <span class="mi">9999</span><span class="p">)</span> \ <span class="o">.</span><span class="n">load</span><span class="p">()</span> -<span class="c"># Split the lines into words</span> +<span class="c1"># Split the lines into words</span> <span class="n">words</span> <span class="o">=</span> <span class="n">lines</span><span class="o">.</span><span class="n">select</span><span class="p">(</span> <span class="n">explode</span><span class="p">(</span> - <span class="n">split</span><span class="p">(</span><span class="n">lines</span><span class="o">.</span><span class="n">value</span><span class="p">,</span> <span class="s">" "</span><span class="p">)</span> - <span class="p">)</span><span class="o">.</span><span class="n">alias</span><span class="p">(</span><span class="s">"word"</span><span class="p">)</span> + <span class="n">split</span><span class="p">(</span><span class="n">lines</span><span class="o">.</span><span class="n">value</span><span class="p">,</span> <span class="s2">" "</span><span class="p">)</span> + <span class="p">)</span><span class="o">.</span><span class="n">alias</span><span class="p">(</span><span class="s2">"word"</span><span class="p">)</span> <span class="p">)</span> -<span class="c"># Generate running word count</span> -<span class="n">wordCounts</span> <span class="o">=</span> <span class="n">words</span><span class="o">.</span><span class="n">groupBy</span><span class="p">(</span><span class="s">"word"</span><span class="p">)</span><span class="o">.</span><span class="n">count</span><span class="p">()</span></code></pre></div> +<span class="c1"># Generate running word count</span> +<span class="n">wordCounts</span> <span class="o">=</span> <span class="n">words</span><span class="o">.</span><span class="n">groupBy</span><span class="p">(</span><span class="s2">"word"</span><span class="p">)</span><span class="o">.</span><span class="n">count</span><span class="p">()</span></code></pre></figure> <p>This <code>lines</code> DataFrame represents an unbounded table containing the streaming text data. This table contains one column of strings named “value”, and each line in the streaming text data becomes a row in the table. Note, that this is not currently receiving any data as we are just setting up the transformation, and have not yet started it. Next, we have used two built-in SQL functions - split and explode, to split each line into multiple rows with a word each. In addition, we use the function <code>alias</code> to name the new column as “word”. Finally, we have defined the <code>wordCounts</code> DataFrame by grouping by the unique values in the Dataset and counting them. Note that this is a streaming DataFrame which represents the running word counts of the stream.</p> @@ -293,36 +298,36 @@ And if you <a href="http://spark.apache.org/downloads.html";>download Spark</a>, <div class="codetabs"> <div data-lang="scala"> - <div class="highlight"><pre><code class="language-scala" data-lang="scala"><span class="c1">// Start running the query that prints the running counts to the console</span> + <figure class="highlight"><pre><code class="language-scala" data-lang="scala"><span></span><span class="c1">// Start running the query that prints the running counts to the console</span> <span class="k">val</span> <span class="n">query</span> <span class="k">=</span> <span class="n">wordCounts</span><span class="o">.</span><span class="n">writeStream</span> <span class="o">.</span><span class="n">outputMode</span><span class="o">(</span><span class="s">"complete"</span><span class="o">)</span> <span class="o">.</span><span class="n">format</span><span class="o">(</span><span class="s">"console"</span><span class="o">)</span> <span class="o">.</span><span class="n">start</span><span class="o">()</span> -<span class="n">query</span><span class="o">.</span><span class="n">awaitTermination</span><span class="o">()</span></code></pre></div> +<span class="n">query</span><span class="o">.</span><span class="n">awaitTermination</span><span class="o">()</span></code></pre></figure> </div> <div data-lang="java"> - <div class="highlight"><pre><code class="language-java" data-lang="java"><span class="c1">// Start running the query that prints the running counts to the console</span> + <figure class="highlight"><pre><code class="language-java" data-lang="java"><span></span><span class="c1">// Start running the query that prints the running counts to the console</span> <span class="n">StreamingQuery</span> <span class="n">query</span> <span class="o">=</span> <span class="n">wordCounts</span><span class="o">.</span><span class="na">writeStream</span><span class="o">()</span> <span class="o">.</span><span class="na">outputMode</span><span class="o">(</span><span class="s">"complete"</span><span class="o">)</span> <span class="o">.</span><span class="na">format</span><span class="o">(</span><span class="s">"console"</span><span class="o">)</span> <span class="o">.</span><span class="na">start</span><span class="o">();</span> -<span class="n">query</span><span class="o">.</span><span class="na">awaitTermination</span><span class="o">();</span></code></pre></div> +<span class="n">query</span><span class="o">.</span><span class="na">awaitTermination</span><span class="o">();</span></code></pre></figure> </div> <div data-lang="python"> - <div class="highlight"><pre><code class="language-python" data-lang="python"><span class="c"># Start running the query that prints the running counts to the console</span> + <figure class="highlight"><pre><code class="language-python" data-lang="python"><span></span> <span class="c1"># Start running the query that prints the running counts to the console</span> <span class="n">query</span> <span class="o">=</span> <span class="n">wordCounts</span> \ <span class="o">.</span><span class="n">writeStream</span> \ - <span class="o">.</span><span class="n">outputMode</span><span class="p">(</span><span class="s">"complete"</span><span class="p">)</span> \ - <span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="s">"console"</span><span class="p">)</span> \ + <span class="o">.</span><span class="n">outputMode</span><span class="p">(</span><span class="s2">"complete"</span><span class="p">)</span> \ + <span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="s2">"console"</span><span class="p">)</span> \ <span class="o">.</span><span class="n">start</span><span class="p">()</span> -<span class="n">query</span><span class="o">.</span><span class="n">awaitTermination</span><span class="p">()</span></code></pre></div> +<span class="n">query</span><span class="o">.</span><span class="n">awaitTermination</span><span class="p">()</span></code></pre></figure> </div> </div> @@ -341,17 +346,17 @@ And if you <a href="http://spark.apache.org/downloads.html";>download Spark</a>, <div class="codetabs"> <div data-lang="scala"> - <div class="highlight"><pre><code class="language-bash" data-lang="bash"><span class="nv">$ </span>./bin/run-example org.apache.spark.examples.sql.streaming.StructuredNetworkWordCount localhost 9999</code></pre></div> + <figure class="highlight"><pre><code class="language-bash" data-lang="bash"><span></span>$ ./bin/run-example org.apache.spark.examples.sql.streaming.StructuredNetworkWordCount localhost <span class="m">9999</span></code></pre></figure> </div> <div data-lang="java"> - <div class="highlight"><pre><code class="language-bash" data-lang="bash"><span class="nv">$ </span>./bin/run-example org.apache.spark.examples.sql.streaming.JavaStructuredNetworkWordCount localhost 9999</code></pre></div> + <figure class="highlight"><pre><code class="language-bash" data-lang="bash"><span></span>$ ./bin/run-example org.apache.spark.examples.sql.streaming.JavaStructuredNetworkWordCount localhost <span class="m">9999</span></code></pre></figure> </div> <div data-lang="python"> - <div class="highlight"><pre><code class="language-bash" data-lang="bash"><span class="nv">$ </span>./bin/spark-submit examples/src/main/python/sql/streaming/structured_network_wordcount.py localhost 9999</code></pre></div> + <figure class="highlight"><pre><code class="language-bash" data-lang="bash"><span></span>$ ./bin/spark-submit examples/src/main/python/sql/streaming/structured_network_wordcount.py localhost <span class="m">9999</span></code></pre></figure> </div> </div> @@ -361,10 +366,10 @@ And if you <a href="http://spark.apache.org/downloads.html";>download Spark</a>, <table width="100%"> <td> -<div class="highlight"><pre><code class="language-bash" data-lang="bash"><span class="c"># TERMINAL 1:</span> -<span class="c"># Running Netcat</span> +<figure class="highlight"><pre><code class="language-bash" data-lang="bash"><span></span><span class="c1"># TERMINAL 1:</span> +<span class="c1"># Running Netcat</span> -<span class="nv">$ </span>nc -lk 9999 +$ nc -lk <span class="m">9999</span> apache spark apache hadoop @@ -386,7 +391,7 @@ apache hadoop -...</code></pre></div> +...</code></pre></figure> </td> <td width="2%"></td> @@ -395,90 +400,90 @@ apache hadoop <div data-lang="scala"> - <div class="highlight"><pre><code class="language-bash" data-lang="bash"><span class="c"># TERMINAL 2: RUNNING StructuredNetworkWordCount</span> + <figure class="highlight"><pre><code class="language-bash" data-lang="bash"><span></span><span class="c1"># TERMINAL 2: RUNNING StructuredNetworkWordCount</span> -<span class="nv">$ </span>./bin/run-example org.apache.spark.examples.sql.streaming.StructuredNetworkWordCount localhost 9999 +$ ./bin/run-example org.apache.spark.examples.sql.streaming.StructuredNetworkWordCount localhost <span class="m">9999</span> ------------------------------------------- -Batch: 0 +Batch: <span class="m">0</span> ------------------------------------------- +------+-----+ <span class="p">|</span> value<span class="p">|</span>count<span class="p">|</span> +------+-----+ -<span class="p">|</span>apache<span class="p">|</span> 1<span class="p">|</span> -<span class="p">|</span> spark<span class="p">|</span> 1<span class="p">|</span> +<span class="p">|</span>apache<span class="p">|</span> <span class="m">1</span><span class="p">|</span> +<span class="p">|</span> spark<span class="p">|</span> <span class="m">1</span><span class="p">|</span> +------+-----+ ------------------------------------------- -Batch: 1 +Batch: <span class="m">1</span> ------------------------------------------- +------+-----+ <span class="p">|</span> value<span class="p">|</span>count<span class="p">|</span> +------+-----+ -<span class="p">|</span>apache<span class="p">|</span> 2<span class="p">|</span> -<span class="p">|</span> spark<span class="p">|</span> 1<span class="p">|</span> -<span class="p">|</span>hadoop<span class="p">|</span> 1<span class="p">|</span> +<span class="p">|</span>apache<span class="p">|</span> <span class="m">2</span><span class="p">|</span> +<span class="p">|</span> spark<span class="p">|</span> <span class="m">1</span><span class="p">|</span> +<span class="p">|</span>hadoop<span class="p">|</span> <span class="m">1</span><span class="p">|</span> +------+-----+ -...</code></pre></div> +...</code></pre></figure> </div> <div data-lang="java"> - <div class="highlight"><pre><code class="language-bash" data-lang="bash"><span class="c"># TERMINAL 2: RUNNING JavaStructuredNetworkWordCount</span> + <figure class="highlight"><pre><code class="language-bash" data-lang="bash"><span></span><span class="c1"># TERMINAL 2: RUNNING JavaStructuredNetworkWordCount</span> -<span class="nv">$ </span>./bin/run-example org.apache.spark.examples.sql.streaming.JavaStructuredNetworkWordCount localhost 9999 +$ ./bin/run-example org.apache.spark.examples.sql.streaming.JavaStructuredNetworkWordCount localhost <span class="m">9999</span> ------------------------------------------- -Batch: 0 +Batch: <span class="m">0</span> ------------------------------------------- +------+-----+ <span class="p">|</span> value<span class="p">|</span>count<span class="p">|</span> +------+-----+ -<span class="p">|</span>apache<span class="p">|</span> 1<span class="p">|</span> -<span class="p">|</span> spark<span class="p">|</span> 1<span class="p">|</span> +<span class="p">|</span>apache<span class="p">|</span> <span class="m">1</span><span class="p">|</span> +<span class="p">|</span> spark<span class="p">|</span> <span class="m">1</span><span class="p">|</span> +------+-----+ ------------------------------------------- -Batch: 1 +Batch: <span class="m">1</span> ------------------------------------------- +------+-----+ <span class="p">|</span> value<span class="p">|</span>count<span class="p">|</span> +------+-----+ -<span class="p">|</span>apache<span class="p">|</span> 2<span class="p">|</span> -<span class="p">|</span> spark<span class="p">|</span> 1<span class="p">|</span> -<span class="p">|</span>hadoop<span class="p">|</span> 1<span class="p">|</span> +<span class="p">|</span>apache<span class="p">|</span> <span class="m">2</span><span class="p">|</span> +<span class="p">|</span> spark<span class="p">|</span> <span class="m">1</span><span class="p">|</span> +<span class="p">|</span>hadoop<span class="p">|</span> <span class="m">1</span><span class="p">|</span> +------+-----+ -...</code></pre></div> +...</code></pre></figure> </div> <div data-lang="python"> - <div class="highlight"><pre><code class="language-bash" data-lang="bash"><span class="c"># TERMINAL 2: RUNNING structured_network_wordcount.py</span> + <figure class="highlight"><pre><code class="language-bash" data-lang="bash"><span></span><span class="c1"># TERMINAL 2: RUNNING structured_network_wordcount.py</span> -<span class="nv">$ </span>./bin/spark-submit examples/src/main/python/sql/streaming/structured_network_wordcount.py localhost 9999 +$ ./bin/spark-submit examples/src/main/python/sql/streaming/structured_network_wordcount.py localhost <span class="m">9999</span> ------------------------------------------- -Batch: 0 +Batch: <span class="m">0</span> ------------------------------------------- +------+-----+ <span class="p">|</span> value<span class="p">|</span>count<span class="p">|</span> +------+-----+ -<span class="p">|</span>apache<span class="p">|</span> 1<span class="p">|</span> -<span class="p">|</span> spark<span class="p">|</span> 1<span class="p">|</span> +<span class="p">|</span>apache<span class="p">|</span> <span class="m">1</span><span class="p">|</span> +<span class="p">|</span> spark<span class="p">|</span> <span class="m">1</span><span class="p">|</span> +------+-----+ ------------------------------------------- -Batch: 1 +Batch: <span class="m">1</span> ------------------------------------------- +------+-----+ <span class="p">|</span> value<span class="p">|</span>count<span class="p">|</span> +------+-----+ -<span class="p">|</span>apache<span class="p">|</span> 2<span class="p">|</span> -<span class="p">|</span> spark<span class="p">|</span> 1<span class="p">|</span> -<span class="p">|</span>hadoop<span class="p">|</span> 1<span class="p">|</span> +<span class="p">|</span>apache<span class="p">|</span> <span class="m">2</span><span class="p">|</span> +<span class="p">|</span> spark<span class="p">|</span> <span class="m">1</span><span class="p">|</span> +<span class="p">|</span>hadoop<span class="p">|</span> <span class="m">1</span><span class="p">|</span> +------+-----+ -...</code></pre></div> +...</code></pre></figure> </div> </div> @@ -500,15 +505,15 @@ arriving on the stream is like a new row being appended to the Input Table.</p> <p><img src="img/structured-streaming-stream-as-a-table.png" alt="Stream as a Table" title="Stream as a Table" /></p> -<p>A query on the input will generate the “Result Table”. Every trigger interval (say, every 1 second), new rows get appended to the Input Table, which eventually updates the Result Table. Whenever the result table gets updated, we would want to write the changed result rows to an external sink.</p> +<p>A query on the input will generate the “Result Table”. Every trigger interval (say, every 1 second), new rows get appended to the Input Table, which eventually updates the Result Table. Whenever the result table gets updated, we would want to write the changed result rows to an external sink. </p> <p><img src="img/structured-streaming-model.png" alt="Model" /></p> -<p>The “Output” is defined as what gets written out to the external storage. The output can be defined in different modes</p> +<p>The “Output” is defined as what gets written out to the external storage. The output can be defined in different modes </p> <ul> <li> - <p><em>Complete Mode</em> - The entire updated Result Table will be written to the external storage. It is up to the storage connector to decide how to handle writing of the entire table.</p> + <p><em>Complete Mode</em> - The entire updated Result Table will be written to the external storage. It is up to the storage connector to decide how to handle writing of the entire table. </p> </li> <li> <p><em>Append Mode</em> - Only the new rows appended in the Result Table since the last trigger will be written to the external storage. This is applicable only on the queries where existing rows in the Result Table are not expected to change.</p> @@ -542,7 +547,14 @@ see how this model handles event-time based processing and late arriving data.</ <h2 id="handling-event-time-and-late-data">Handling Event-time and Late Data</h2> <p>Event-time is the time embedded in the data itself. For many applications, you may want to operate on this event-time. For example, if you want to get the number of events generated by IoT devices every minute, then you probably want to use the time when the data was generated (that is, event-time in the data), rather than the time Spark receives them. This event-time is very naturally expressed in this model – each event from the devices is a row in the table, and event-time is a column value in the row. This allows window-based aggregations (e.g. number of events every minute) to be just a special type of grouping and aggregation on the even-time column – each time window is a group and each row can belong to multiple windows/groups. Therefore, such event-time-window-based aggregation queries can be defined consistently on both a static dataset (e.g. from collected device events logs) as well as on a data stream, making the life of the user much easier.</p> -<p>Furthermore, this model naturally handles data that has arrived later than expected based on its event-time. Since Spark is updating the Result Table, it has full control over updating/cleaning up the aggregates when there is late data. While not yet implemented in Spark 2.0, event-time watermarking will be used to manage this data. These are explained later in more details in the <a href="#window-operations-on-event-time">Window Operations</a> section.</p> +<p>Furthermore, this model naturally handles data that has arrived later than +expected based on its event-time. Since Spark is updating the Result Table, +it has full control over updating old aggregates when there is late data, +as well as cleaning up old aggregates to limit the size of intermediate +state data. Since Spark 2.1, we have support for watermarking which +allows the user to specify the threshold of late data, and allows the engine +to accordingly clean up old state. These are explained later in more +details in the <a href="#window-operations-on-event-time">Window Operations</a> section.</p> <h2 id="fault-tolerance-semantics">Fault Tolerance Semantics</h2> <p>Delivering end-to-end exactly-once semantics was one of key goals behind the design of Structured Streaming. To achieve that, we have designed the Structured Streaming sources, the sinks and the execution engine to reliably track the exact progress of the processing so that it can handle any kind of failure by restarting and/or reprocessing. Every streaming source is assumed to have offsets (similar to Kafka offsets, or Kinesis sequence numbers) @@ -570,7 +582,7 @@ returned by <code>SparkSession.readStream()</code>. Similar to the read interfac <p><strong>Kafka source</strong> - Poll data from Kafka. It’s compatible with Kafka broker versions 0.10.0 or higher. See the <a href="structured-streaming-kafka-integration.html">Kafka Integration Guide</a> for more details.</p> </li> <li> - <p><strong>Socket source (for testing)</strong> - Reads UTF8 text data from a socket connection. The listening server socket is at the driver. Note that this should be used only for testing as this does not provide end-to-end fault-tolerance guarantees.</p> + <p><strong>Socket source (for testing)</strong> - Reads UTF8 text data from a socket connection. The listening server socket is at the driver. Note that this should be used only for testing as this does not provide end-to-end fault-tolerance guarantees. </p> </li> </ul> @@ -579,7 +591,7 @@ returned by <code>SparkSession.readStream()</code>. Similar to the read interfac <div class="codetabs"> <div data-lang="scala"> - <div class="highlight"><pre><code class="language-scala" data-lang="scala"><span class="k">val</span> <span class="n">spark</span><span class="k">:</span> <span class="kt">SparkSession</span> <span class="o">=</span> <span class="o">...</span> + <figure class="highlight"><pre><code class="language-scala" data-lang="scala"><span></span><span class="k">val</span> <span class="n">spark</span><span class="k">:</span> <span class="kt">SparkSession</span> <span class="o">=</span> <span class="o">...</span> <span class="c1">// Read text from socket </span> <span class="k">val</span> <span class="n">socketDF</span> <span class="k">=</span> <span class="n">spark</span> @@ -599,12 +611,12 @@ returned by <code>SparkSession.readStream()</code>. Similar to the read interfac <span class="o">.</span><span class="n">readStream</span> <span class="o">.</span><span class="n">option</span><span class="o">(</span><span class="s">"sep"</span><span class="o">,</span> <span class="s">";"</span><span class="o">)</span> <span class="o">.</span><span class="n">schema</span><span class="o">(</span><span class="n">userSchema</span><span class="o">)</span> <span class="c1">// Specify schema of the csv files</span> - <span class="o">.</span><span class="n">csv</span><span class="o">(</span><span class="s">"/path/to/directory"</span><span class="o">)</span> <span class="c1">// Equivalent to format("csv").load("/path/to/directory")</span></code></pre></div> + <span class="o">.</span><span class="n">csv</span><span class="o">(</span><span class="s">"/path/to/directory"</span><span class="o">)</span> <span class="c1">// Equivalent to format("csv").load("/path/to/directory")</span></code></pre></figure> </div> <div data-lang="java"> - <div class="highlight"><pre><code class="language-java" data-lang="java"><span class="n">SparkSession</span> <span class="n">spark</span> <span class="o">=</span> <span class="o">...</span> + <figure class="highlight"><pre><code class="language-java" data-lang="java"><span></span><span class="n">SparkSession</span> <span class="n">spark</span> <span class="o">=</span> <span class="o">...</span> <span class="c1">// Read text from socket </span> <span class="n">Dataset</span><span class="o">[</span><span class="n">Row</span><span class="o">]</span> <span class="n">socketDF</span> <span class="o">=</span> <span class="n">spark</span> @@ -619,37 +631,37 @@ returned by <code>SparkSession.readStream()</code>. Similar to the read interfac <span class="n">socketDF</span><span class="o">.</span><span class="na">printSchema</span><span class="o">();</span> <span class="c1">// Read all the csv files written atomically in a directory</span> -<span class="n">StructType</span> <span class="n">userSchema</span> <span class="o">=</span> <span class="k">new</span> <span class="nf">StructType</span><span class="o">().</span><span class="na">add</span><span class="o">(</span><span class="s">"name"</span><span class="o">,</span> <span class="s">"string"</span><span class="o">).</span><span class="na">add</span><span class="o">(</span><span class="s">"age"</span><span class="o">,</span> <span class="s">"integer"</span><span class="o">);</span> +<span class="n">StructType</span> <span class="n">userSchema</span> <span class="o">=</span> <span class="k">new</span> <span class="n">StructType</span><span class="o">().</span><span class="na">add</span><span class="o">(</span><span class="s">"name"</span><span class="o">,</span> <span class="s">"string"</span><span class="o">).</span><span class="na">add</span><span class="o">(</span><span class="s">"age"</span><span class="o">,</span> <span class="s">"integer"</span><span class="o">);</span> <span class="n">Dataset</span><span class="o">[</span><span class="n">Row</span><span class="o">]</span> <span class="n">csvDF</span> <span class="o">=</span> <span class="n">spark</span> <span class="o">.</span><span class="na">readStream</span><span class="o">()</span> <span class="o">.</span><span class="na">option</span><span class="o">(</span><span class="s">"sep"</span><span class="o">,</span> <span class="s">";"</span><span class="o">)</span> <span class="o">.</span><span class="na">schema</span><span class="o">(</span><span class="n">userSchema</span><span class="o">)</span> <span class="c1">// Specify schema of the csv files</span> - <span class="o">.</span><span class="na">csv</span><span class="o">(</span><span class="s">"/path/to/directory"</span><span class="o">);</span> <span class="c1">// Equivalent to format("csv").load("/path/to/directory")</span></code></pre></div> + <span class="o">.</span><span class="na">csv</span><span class="o">(</span><span class="s">"/path/to/directory"</span><span class="o">);</span> <span class="c1">// Equivalent to format("csv").load("/path/to/directory")</span></code></pre></figure> </div> <div data-lang="python"> - <div class="highlight"><pre><code class="language-python" data-lang="python"><span class="n">spark</span> <span class="o">=</span> <span class="n">SparkSession</span><span class="o">.</span> <span class="o">...</span> + <figure class="highlight"><pre><code class="language-python" data-lang="python"><span></span><span class="n">spark</span> <span class="o">=</span> <span class="n">SparkSession</span><span class="o">.</span> <span class="o">...</span> -<span class="c"># Read text from socket </span> +<span class="c1"># Read text from socket </span> <span class="n">socketDF</span> <span class="o">=</span> <span class="n">spark</span> \ <span class="o">.</span><span class="n">readStream</span><span class="p">()</span> \ - <span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="s">"socket"</span><span class="p">)</span> \ - <span class="o">.</span><span class="n">option</span><span class="p">(</span><span class="s">"host"</span><span class="p">,</span> <span class="s">"localhost"</span><span class="p">)</span> \ - <span class="o">.</span><span class="n">option</span><span class="p">(</span><span class="s">"port"</span><span class="p">,</span> <span class="mi">9999</span><span class="p">)</span> \ + <span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="s2">"socket"</span><span class="p">)</span> \ + <span class="o">.</span><span class="n">option</span><span class="p">(</span><span class="s2">"host"</span><span class="p">,</span> <span class="s2">"localhost"</span><span class="p">)</span> \ + <span class="o">.</span><span class="n">option</span><span class="p">(</span><span class="s2">"port"</span><span class="p">,</span> <span class="mi">9999</span><span class="p">)</span> \ <span class="o">.</span><span class="n">load</span><span class="p">()</span> -<span class="n">socketDF</span><span class="o">.</span><span class="n">isStreaming</span><span class="p">()</span> <span class="c"># Returns True for DataFrames that have streaming sources</span> +<span class="n">socketDF</span><span class="o">.</span><span class="n">isStreaming</span><span class="p">()</span> <span class="c1"># Returns True for DataFrames that have streaming sources</span> <span class="n">socketDF</span><span class="o">.</span><span class="n">printSchema</span><span class="p">()</span> -<span class="c"># Read all the csv files written atomically in a directory</span> -<span class="n">userSchema</span> <span class="o">=</span> <span class="n">StructType</span><span class="p">()</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="s">"name"</span><span class="p">,</span> <span class="s">"string"</span><span class="p">)</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="s">"age"</span><span class="p">,</span> <span class="s">"integer"</span><span class="p">)</span> +<span class="c1"># Read all the csv files written atomically in a directory</span> +<span class="n">userSchema</span> <span class="o">=</span> <span class="n">StructType</span><span class="p">()</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="s2">"name"</span><span class="p">,</span> <span class="s2">"string"</span><span class="p">)</span><span class="o">.</span><span class="n">add</span><span class="p">(</span><span class="s2">"age"</span><span class="p">,</span> <span class="s2">"integer"</span><span class="p">)</span> <span class="n">csvDF</span> <span class="o">=</span> <span class="n">spark</span> \ <span class="o">.</span><span class="n">readStream</span><span class="p">()</span> \ - <span class="o">.</span><span class="n">option</span><span class="p">(</span><span class="s">"sep"</span><span class="p">,</span> <span class="s">";"</span><span class="p">)</span> \ + <span class="o">.</span><span class="n">option</span><span class="p">(</span><span class="s2">"sep"</span><span class="p">,</span> <span class="s2">";"</span><span class="p">)</span> \ <span class="o">.</span><span class="n">schema</span><span class="p">(</span><span class="n">userSchema</span><span class="p">)</span> \ - <span class="o">.</span><span class="n">csv</span><span class="p">(</span><span class="s">"/path/to/directory"</span><span class="p">)</span> <span class="c"># Equivalent to format("csv").load("/path/to/directory")</span></code></pre></div> + <span class="o">.</span><span class="n">csv</span><span class="p">(</span><span class="s2">"/path/to/directory"</span><span class="p">)</span> <span class="c1"># Equivalent to format("csv").load("/path/to/directory")</span></code></pre></figure> </div> </div> @@ -671,7 +683,7 @@ returned by <code>SparkSession.readStream()</code>. Similar to the read interfac <div class="codetabs"> <div data-lang="scala"> - <div class="highlight"><pre><code class="language-scala" data-lang="scala"><span class="k">case</span> <span class="k">class</span> <span class="nc">DeviceData</span><span class="o">(</span><span class="n">device</span><span class="k">:</span> <span class="kt">String</span><span class="o">,</span> <span class="n">type</span><span class="k">:</span> <span class="kt">String</span><span class="o">,</span> <span class="n">signal</span><span class="k">:</span> <span class="kt">Double</span><span class="o">,</span> <span class="n">time</span><span class="k">:</span> <span class="kt">DateTime</span><span class="o">)</span> + <figure class="highlight"><pre><code class="language-scala" data-lang="scala"><span></span><span class="k">case</span> <span class="k">class</span> <span class="nc">DeviceData</span><span class="o">(</span><span class="n">device</span><span class="k">:</span> <span class="kt">String</span><span class="o">,</span> <span class="n">type</span><span class="k">:</span> <span class="kt">String</span><span class="o">,</span> <span class="n">signal</span><span class="k">:</span> <span class="kt">Double</span><span class="o">,</span> <span class="n">time</span><span class="k">:</span> <span class="kt">DateTime</span><span class="o">)</span> <span class="k">val</span> <span class="n">df</span><span class="k">:</span> <span class="kt">DataFrame</span> <span class="o">=</span> <span class="o">...</span> <span class="c1">// streaming DataFrame with IOT device data with schema { device: string, type: string, signal: double, time: string }</span> <span class="k">val</span> <span class="n">ds</span><span class="k">:</span> <span class="kt">Dataset</span><span class="o">[</span><span class="kt">DeviceData</span><span class="o">]</span> <span class="k">=</span> <span class="n">df</span><span class="o">.</span><span class="n">as</span><span class="o">[</span><span class="kt">DeviceData</span><span class="o">]</span> <span class="c1">// streaming Dataset with IOT device data</span> @@ -685,12 +697,12 @@ returned by <code>SparkSession.readStream()</code>. Similar to the read interfac <span class="c1">// Running average signal for each device type</span> <span class="k">import</span> <span class="nn">org.apache.spark.sql.expressions.scalalang.typed._</span> -<span class="n">ds</span><span class="o">.</span><span class="n">groupByKey</span><span class="o">(</span><span class="k">_</span><span class="o">.</span><span class="n">type</span><span class="o">).</span><span class="n">agg</span><span class="o">(</span><span class="n">typed</span><span class="o">.</span><span class="n">avg</span><span class="o">(</span><span class="k">_</span><span class="o">.</span><span class="n">signal</span><span class="o">))</span> <span class="c1">// using typed API</span></code></pre></div> +<span class="n">ds</span><span class="o">.</span><span class="n">groupByKey</span><span class="o">(</span><span class="k">_</span><span class="o">.</span><span class="n">type</span><span class="o">).</span><span class="n">agg</span><span class="o">(</span><span class="n">typed</span><span class="o">.</span><span class="n">avg</span><span class="o">(</span><span class="k">_</span><span class="o">.</span><span class="n">signal</span><span class="o">))</span> <span class="c1">// using typed API</span></code></pre></figure> </div> <div data-lang="java"> - <div class="highlight"><pre><code class="language-java" data-lang="java"><span class="kn">import</span> <span class="nn">org.apache.spark.api.java.function.*</span><span class="o">;</span> + <figure class="highlight"><pre><code class="language-java" data-lang="java"><span></span><span class="kn">import</span> <span class="nn">org.apache.spark.api.java.function.*</span><span class="o">;</span> <span class="kn">import</span> <span class="nn">org.apache.spark.sql.*</span><span class="o">;</span> <span class="kn">import</span> <span class="nn">org.apache.spark.sql.expressions.javalang.typed</span><span class="o">;</span> <span class="kn">import</span> <span class="nn">org.apache.spark.sql.catalyst.encoders.ExpressionEncoder</span><span class="o">;</span> @@ -735,24 +747,24 @@ returned by <code>SparkSession.readStream()</code>. Similar to the read interfac <span class="kd">public</span> <span class="n">Double</span> <span class="nf">call</span><span class="o">(</span><span class="n">DeviceData</span> <span class="n">value</span><span class="o">)</span> <span class="kd">throws</span> <span class="n">Exception</span> <span class="o">{</span> <span class="k">return</span> <span class="n">value</span><span class="o">.</span><span class="na">getSignal</span><span class="o">();</span> <span class="o">}</span> -<span class="o">}));</span></code></pre></div> +<span class="o">}));</span></code></pre></figure> </div> <div data-lang="python"> - <div class="highlight"><pre><code class="language-python" data-lang="python"><span class="n">df</span> <span class="o">=</span> <span class="o">...</span> <span class="c"># streaming DataFrame with IOT device data with schema { device: string, type: string, signal: double, time: DateType }</span> + <figure class="highlight"><pre><code class="language-python" data-lang="python"><span></span><span class="n">df</span> <span class="o">=</span> <span class="o">...</span> <span class="c1"># streaming DataFrame with IOT device data with schema { device: string, type: string, signal: double, time: DateType }</span> -<span class="c"># Select the devices which have signal more than 10</span> -<span class="n">df</span><span class="o">.</span><span class="n">select</span><span class="p">(</span><span class="s">"device"</span><span class="p">)</span><span class="o">.</span><span class="n">where</span><span class="p">(</span><span class="s">"signal > 10"</span><span class="p">)</span> +<span class="c1"># Select the devices which have signal more than 10</span> +<span class="n">df</span><span class="o">.</span><span class="n">select</span><span class="p">(</span><span class="s2">"device"</span><span class="p">)</span><span class="o">.</span><span class="n">where</span><span class="p">(</span><span class="s2">"signal > 10"</span><span class="p">)</span> -<span class="c"># Running count of the number of updates for each device type</span> -<span class="n">df</span><span class="o">.</span><span class="n">groupBy</span><span class="p">(</span><span class="s">"type"</span><span class="p">)</span><span class="o">.</span><span class="n">count</span><span class="p">()</span></code></pre></div> +<span class="c1"># Running count of the number of updates for each device type</span> +<span class="n">df</span><span class="o">.</span><span class="n">groupBy</span><span class="p">(</span><span class="s2">"type"</span><span class="p">)</span><span class="o">.</span><span class="n">count</span><span class="p">()</span></code></pre></figure> </div> </div> <h3 id="window-operations-on-event-time">Window Operations on Event Time</h3> -<p>Aggregations over a sliding event-time window are straightforward with Structured Streaming. The key idea to understand about window-based aggregations are very similar to grouped aggregations. In a grouped aggregation, aggregate values (e.g. counts) are maintained for each unique value in the user-specified grouping column. In case of window-based aggregations, aggregate values are maintained for each window the event-time of a row falls into. Let’s understand this with an illustration.</p> +<p>Aggregations over a sliding event-time window are straightforward with Structured Streaming. The key idea to understand about window-based aggregations are very similar to grouped aggregations. In a grouped aggregation, aggregate values (e.g. counts) are maintained for each unique value in the user-specified grouping column. In case of window-based aggregations, aggregate values are maintained for each window the event-time of a row falls into. Let’s understand this with an illustration. </p> <p>Imagine our <a href="#quick-example">quick example</a> is modified and the stream now contains lines along with the time when the line was generated. Instead of running word counts, we want to count words within 10 minute windows, updating every 5 minutes. That is, word counts in words received between 10 minute windows 12:00 - 12:10, 12:05 - 12:15, 12:10 - 12:20, etc. Note that 12:00 - 12:10 means data that arrived after 12:00 but before 12:10. Now, consider a word that was received at 12:07. This word should increment the counts corresponding to two windows 12:00 - 12:10 and 12:05 - 12:15. So the counts will be indexed by both, the grouping key (i.e. the word) and the window (can be calculated from the event-time).</p> @@ -766,7 +778,7 @@ returned by <code>SparkSession.readStream()</code>. Similar to the read interfac <div class="codetabs"> <div data-lang="scala"> - <div class="highlight"><pre><code class="language-scala" data-lang="scala"><span class="k">import</span> <span class="nn">spark.implicits._</span> + <figure class="highlight"><pre><code class="language-scala" data-lang="scala"><span></span><span class="k">import</span> <span class="nn">spark.implicits._</span> <span class="k">val</span> <span class="n">words</span> <span class="k">=</span> <span class="o">...</span> <span class="c1">// streaming DataFrame of schema { timestamp: Timestamp, word: String }</span> @@ -774,66 +786,178 @@ returned by <code>SparkSession.readStream()</code>. Similar to the read interfac <span class="k">val</span> <span class="n">windowedCounts</span> <span class="k">=</span> <span class="n">words</span><span class="o">.</span><span class="n">groupBy</span><span class="o">(</span> <span class="n">window</span><span class="o">(</span><span class="n">$</span><span class="s">"timestamp"</span><span class="o">,</span> <span class="s">"10 minutes"</span><span class="o">,</span> <span class="s">"5 minutes"</span><span class="o">),</span> <span class="n">$</span><span class="s">"word"</span> -<span class="o">).</span><span class="n">count</span><span class="o">()</span></code></pre></div> +<span class="o">).</span><span class="n">count</span><span class="o">()</span></code></pre></figure> </div> <div data-lang="java"> - <div class="highlight"><pre><code class="language-java" data-lang="java"><span class="n">Dataset</span><span class="o"><</span><span class="n">Row</span><span class="o">></span> <span class="n">words</span> <span class="o">=</span> <span class="o">...</span> <span class="c1">// streaming DataFrame of schema { timestamp: Timestamp, word: String }</span> + <figure class="highlight"><pre><code class="language-java" data-lang="java"><span></span><span class="n">Dataset</span><span class="o"><</span><span class="n">Row</span><span class="o">></span> <span class="n">words</span> <span class="o">=</span> <span class="o">...</span> <span class="c1">// streaming DataFrame of schema { timestamp: Timestamp, word: String }</span> <span class="c1">// Group the data by window and word and compute the count of each group</span> <span class="n">Dataset</span><span class="o"><</span><span class="n">Row</span><span class="o">></span> <span class="n">windowedCounts</span> <span class="o">=</span> <span class="n">words</span><span class="o">.</span><span class="na">groupBy</span><span class="o">(</span> <span class="n">functions</span><span class="o">.</span><span class="na">window</span><span class="o">(</span><span class="n">words</span><span class="o">.</span><span class="na">col</span><span class="o">(</span><span class="s">"timestamp"</span><span class="o">),</span> <span class="s">"10 minutes"</span><span class="o">,</span> <span class="s">"5 minutes"</span><span class="o">),</span> <span class="n">words</span><span class="o">.</span><span class="na">col</span><span class="o">(</span><span class="s">"word"</span><span class="o">)</span> -<span class="o">).</span><span class="na">count</span><span class="o">();</span></code></pre></div> +<span class="o">).</span><span class="na">count</span><span class="o">();</span></code></pre></figure> </div> <div data-lang="python"> - <div class="highlight"><pre><code class="language-python" data-lang="python"><span class="n">words</span> <span class="o">=</span> <span class="o">...</span> <span class="c"># streaming DataFrame of schema { timestamp: Timestamp, word: String }</span> + <figure class="highlight"><pre><code class="language-python" data-lang="python"><span></span><span class="n">words</span> <span class="o">=</span> <span class="o">...</span> <span class="c1"># streaming DataFrame of schema { timestamp: Timestamp, word: String }</span> -<span class="c"># Group the data by window and word and compute the count of each group</span> +<span class="c1"># Group the data by window and word and compute the count of each group</span> <span class="n">windowedCounts</span> <span class="o">=</span> <span class="n">words</span><span class="o">.</span><span class="n">groupBy</span><span class="p">(</span> - <span class="n">window</span><span class="p">(</span><span class="n">words</span><span class="o">.</span><span class="n">timestamp</span><span class="p">,</span> <span class="s">"10 minutes"</span><span class="p">,</span> <span class="s">"5 minutes"</span><span class="p">),</span> + <span class="n">window</span><span class="p">(</span><span class="n">words</span><span class="o">.</span><span class="n">timestamp</span><span class="p">,</span> <span class="s2">"10 minutes"</span><span class="p">,</span> <span class="s2">"5 minutes"</span><span class="p">),</span> <span class="n">words</span><span class="o">.</span><span class="n">word</span> -<span class="p">)</span><span class="o">.</span><span class="n">count</span><span class="p">()</span></code></pre></div> +<span class="p">)</span><span class="o">.</span><span class="n">count</span><span class="p">()</span></code></pre></figure> </div> </div> +<h3 id="handling-late-data-and-watermarking">Handling Late Data and Watermarking</h3> <p>Now consider what happens if one of the events arrives late to the application. -For example, a word that was generated at 12:04 but it was received at 12:11. -Since this windowing is based on the time in the data, the time 12:04 should be considered for windowing. This occurs naturally in our window-based grouping â the late data is automatically placed in the proper windows and the correct aggregates are updated as illustrated below.</p> +For example, say, a word generated at 12:04 (i.e. event time) could be received received by +the application at 12:11. The application should use the time 12:04 instead of 12:11 +to update the older counts for the window <code>12:00 - 12:10</code>. This occurs +naturally in our window-based grouping â Structured Streaming can maintain the intermediate state +for partial aggregates for a long period of time such that late data can update aggregates of +old windows correctly, as illustrated below.</p> <p><img src="img/structured-streaming-late-data.png" alt="Handling Late Data" /></p> +<p>However, to run this query for days, its necessary for the system to bound the amount of +intermediate in-memory state it accumulates. This means the system needs to know when an old +aggregate can be dropped from the in-memory state because the application is not going to receive +late data for that aggregate any more. To enable this, in Spark 2.1, we have introduced +<strong>watermarking</strong>, which let’s the engine automatically track the current event time in the data and +and attempt to clean up old state accordingly. You can define the watermark of a query by +specifying the event time column and the threshold on how late the data is expected be in terms of +event time. For a specific window starting at time <code>T</code>, the engine will maintain state and allow late +data to be update the state until <code>(max event time seen by the engine - late threshold > T)</code>. +In other words, late data within the threshold will be aggregated, +but data later than the threshold will be dropped. Let’s understand this with an example. We can +easily define watermarking on the previous example using <code>withWatermark()</code> as shown below.</p> + +<div class="codetabs"> +<div data-lang="scala"> + + <figure class="highlight"><pre><code class="language-scala" data-lang="scala"><span></span><span class="k">import</span> <span class="nn">spark.implicits._</span> + +<span class="k">val</span> <span class="n">words</span> <span class="k">=</span> <span class="o">...</span> <span class="c1">// streaming DataFrame of schema { timestamp: Timestamp, word: String }</span> + +<span class="c1">// Group the data by window and word and compute the count of each group</span> +<span class="k">val</span> <span class="n">windowedCounts</span> <span class="k">=</span> <span class="n">words</span> + <span class="o">.</span><span class="n">withWatermark</span><span class="o">(</span><span class="s">"timestamp"</span><span class="o">,</span> <span class="s">"10 minutes"</span><span class="o">)</span> + <span class="o">.</span><span class="n">groupBy</span><span class="o">(</span> + <span class="n">window</span><span class="o">(</span><span class="n">$</span><span class="s">"timestamp"</span><span class="o">,</span> <span class="s">"10 minutes"</span><span class="o">,</span> <span class="s">"5 minutes"</span><span class="o">),</span> + <span class="n">$</span><span class="s">"word"</span><span class="o">)</span> + <span class="o">.</span><span class="n">count</span><span class="o">()</span></code></pre></figure> + + </div> +<div data-lang="java"> + + <figure class="highlight"><pre><code class="language-java" data-lang="java"><span></span><span class="n">Dataset</span><span class="o"><</span><span class="n">Row</span><span class="o">></span> <span class="n">words</span> <span class="o">=</span> <span class="o">...</span> <span class="c1">// streaming DataFrame of schema { timestamp: Timestamp, word: String }</span> + +<span class="c1">// Group the data by window and word and compute the count of each group</span> +<span class="n">Dataset</span><span class="o"><</span><span class="n">Row</span><span class="o">></span> <span class="n">windowedCounts</span> <span class="o">=</span> <span class="n">words</span> + <span class="o">.</span><span class="na">withWatermark</span><span class="o">(</span><span class="s">"timestamp"</span><span class="o">,</span> <span class="s">"10 minutes"</span><span class="o">)</span> + <span class="o">.</span><span class="na">groupBy</span><span class="o">(</span> + <span class="n">functions</span><span class="o">.</span><span class="na">window</span><span class="o">(</span><span class="n">words</span><span class="o">.</span><span class="na">col</span><span class="o">(</span><span class="s">"timestamp"</span><span class="o">),</span> <span class="s">"10 minutes"</span><span class="o">,</span> <span class="s">"5 minutes"</span><span class="o">),</span> + <span class="n">words</span><span class="o">.</span><span class="na">col</span><span class="o">(</span><span class="s">"word"</span><span class="o">))</span> + <span class="o">.</span><span class="na">count</span><span class="o">();</span></code></pre></figure> + + </div> +<div data-lang="python"> + + <figure class="highlight"><pre><code class="language-python" data-lang="python"><span></span><span class="n">words</span> <span class="o">=</span> <span class="o">...</span> <span class="c1"># streaming DataFrame of schema { timestamp: Timestamp, word: String }</span> + +<span class="c1"># Group the data by window and word and compute the count of each group</span> +<span class="n">windowedCounts</span> <span class="o">=</span> <span class="n">words</span> + <span class="o">.</span><span class="n">withWatermark</span><span class="p">(</span><span class="s2">"timestamp"</span><span class="p">,</span> <span class="s2">"10 minutes"</span><span class="p">)</span> + <span class="o">.</span><span class="n">groupBy</span><span class="p">(</span> + <span class="n">window</span><span class="p">(</span><span class="n">words</span><span class="o">.</span><span class="n">timestamp</span><span class="p">,</span> <span class="s2">"10 minutes"</span><span class="p">,</span> <span class="s2">"5 minutes"</span><span class="p">),</span> + <span class="n">words</span><span class="o">.</span><span class="n">word</span><span class="p">)</span> + <span class="o">.</span><span class="n">count</span><span class="p">()</span></code></pre></figure> + + </div> +</div> + +<p>In this example, we are defining the watermark of the query on the value of the column “timestamp”, +and also defining “10 minutes” as the threshold of how late is the data allowed to be. If this query +is run in Append output mode (discussed later in <a href="#output-modes">Output Modes</a> section), +the engine will track the current event time from the column “timestamp” and wait for additional +“10 minutes” in event time before finalizing the windowed counts and adding them to the Result Table. +Here is an illustration. </p> + +<p><img src="img/structured-streaming-watermark.png" alt="Watermarking in Append Mode" /></p> + +<p>As shown in the illustration, the maximum event time tracked by the engine is the +<em>blue dashed line</em>, and the watermark set as <code>(max event time - '10 mins')</code> +at the beginning of every trigger is the red line For example, when the engine observes the data +<code>(12:14, dog)</code>, it sets the watermark for the next trigger as <code>12:04</code>. +For the window <code>12:00 - 12:10</code>, the partial counts are maintained as internal state while the system +is waiting for late data. After the system finds data (i.e. <code>(12:21, owl)</code>) such that the +watermark exceeds 12:10, the partial count is finalized and appended to the table. This count will +not change any further as all “too-late” data older than 12:10 will be ignored. </p> + +<p>Note that in Append output mode, the system has to wait for “late threshold” time +before it can output the aggregation of a window. This may not be ideal if data can be very late, +(say 1 day) and you like to have partial counts without waiting for a day. In future, we will add +Update output mode which would allows every update to aggregates to be written to sink every trigger. </p> + +<p><strong>Conditions for watermarking to clean aggregation state</strong> +It is important to note that the following conditions must be satisfied for the watermarking to +clean the state in aggregation queries <em>(as of Spark 2.1, subject to change in the future)</em>.</p> + +<ul> + <li> + <p><strong>Output mode must be Append.</strong> Complete mode requires all aggregate data to be preserved, and hence +cannot use watermarking to drop intermediate state. See the <a href="#output-modes">Output Modes</a> section +for detailed explanation of the semantics of each output mode.</p> + </li> + <li> + <p>The aggregation must have either the event-time column, or a <code>window</code> on the event-time column. </p> + </li> + <li> + <p><code>withWatermark</code> must be called on the +same column as the timestamp column used in the aggregate. For example, +<code>df.withWatermark("time", "1 min").groupBy("time2").count()</code> is invalid +in Append output mode, as watermark is defined on a different column +as the aggregation column.</p> + </li> + <li> + <p><code>withWatermark</code> must be called before the aggregation for the watermark details to be used. +For example, <code>df.groupBy("time").count().withWatermark("time", "1 min")</code> is invalid in Append +output mode.</p> + </li> +</ul> + <h3 id="join-operations">Join Operations</h3> <p>Streaming DataFrames can be joined with static DataFrames to create new streaming DataFrames. Here are a few examples.</p> <div class="codetabs"> <div data-lang="scala"> - <div class="highlight"><pre><code class="language-scala" data-lang="scala"><span class="k">val</span> <span class="n">staticDf</span> <span class="k">=</span> <span class="n">spark</span><span class="o">.</span><span class="n">read</span><span class="o">.</span> <span class="o">...</span> + <figure class="highlight"><pre><code class="language-scala" data-lang="scala"><span></span><span class="k">val</span> <span class="n">staticDf</span> <span class="k">=</span> <span class="n">spark</span><span class="o">.</span><span class="n">read</span><span class="o">.</span> <span class="o">...</span> <span class="k">val</span> <span class="n">streamingDf</span> <span class="k">=</span> <span class="n">spark</span><span class="o">.</span><span class="n">readStream</span><span class="o">.</span> <span class="o">...</span> <span class="n">streamingDf</span><span class="o">.</span><span class="n">join</span><span class="o">(</span><span class="n">staticDf</span><span class="o">,</span> <span class="s">"type"</span><span class="o">)</span> <span class="c1">// inner equi-join with a static DF</span> -<span class="n">streamingDf</span><span class="o">.</span><span class="n">join</span><span class="o">(</span><span class="n">staticDf</span><span class="o">,</span> <span class="s">"type"</span><span class="o">,</span> <span class="s">"right_join"</span><span class="o">)</span> <span class="c1">// right outer join with a static DF</span></code></pre></div> +<span class="n">streamingDf</span><span class="o">.</span><span class="n">join</span><span class="o">(</span><span class="n">staticDf</span><span class="o">,</span> <span class="s">"type"</span><span class="o">,</span> <span class="s">"right_join"</span><span class="o">)</span> <span class="c1">// right outer join with a static DF </span></code></pre></figure> </div> <div data-lang="java"> - <div class="highlight"><pre><code class="language-java" data-lang="java"><span class="n">Dataset</span><span class="o"><</span><span class="n">Row</span><span class="o">></span> <span class="n">staticDf</span> <span class="o">=</span> <span class="n">spark</span><span class="o">.</span><span class="na">read</span><span class="o">.</span> <span class="o">...;</span> + <figure class="highlight"><pre><code class="language-java" data-lang="java"><span></span><span class="n">Dataset</span><span class="o"><</span><span class="n">Row</span><span class="o">></span> <span class="n">staticDf</span> <span class="o">=</span> <span class="n">spark</span><span class="o">.</span><span class="na">read</span><span class="o">.</span> <span class="o">...;</span> <span class="n">Dataset</span><span class="o"><</span><span class="n">Row</span><span class="o">></span> <span class="n">streamingDf</span> <span class="o">=</span> <span class="n">spark</span><span class="o">.</span><span class="na">readStream</span><span class="o">.</span> <span class="o">...;</span> <span class="n">streamingDf</span><span class="o">.</span><span class="na">join</span><span class="o">(</span><span class="n">staticDf</span><span class="o">,</span> <span class="s">"type"</span><span class="o">);</span> <span class="c1">// inner equi-join with a static DF</span> -<span class="n">streamingDf</span><span class="o">.</span><span class="na">join</span><span class="o">(</span><span class="n">staticDf</span><span class="o">,</span> <span class="s">"type"</span><span class="o">,</span> <span class="s">"right_join"</span><span class="o">);</span> <span class="c1">// right outer join with a static DF</span></code></pre></div> +<span class="n">streamingDf</span><span class="o">.</span><span class="na">join</span><span class="o">(</span><span class="n">staticDf</span><span class="o">,</span> <span class="s">"type"</span><span class="o">,</span> <span class="s">"right_join"</span><span class="o">);</span> <span class="c1">// right outer join with a static DF</span></code></pre></figure> </div> <div data-lang="python"> - <div class="highlight"><pre><code class="language-python" data-lang="python"><span class="n">staticDf</span> <span class="o">=</span> <span class="n">spark</span><span class="o">.</span><span class="n">read</span><span class="o">.</span> <span class="o">...</span> + <figure class="highlight"><pre><code class="language-python" data-lang="python"><span></span><span class="n">staticDf</span> <span class="o">=</span> <span class="n">spark</span><span class="o">.</span><span class="n">read</span><span class="o">.</span> <span class="o">...</span> <span class="n">streamingDf</span> <span class="o">=</span> <span class="n">spark</span><span class="o">.</span><span class="n">readStream</span><span class="o">.</span> <span class="o">...</span> -<span class="n">streamingDf</span><span class="o">.</span><span class="n">join</span><span class="p">(</span><span class="n">staticDf</span><span class="p">,</span> <span class="s">"type"</span><span class="p">)</span> <span class="c"># inner equi-join with a static DF</span> -<span class="n">streamingDf</span><span class="o">.</span><span class="n">join</span><span class="p">(</span><span class="n">staticDf</span><span class="p">,</span> <span class="s">"type"</span><span class="p">,</span> <span class="s">"right_join"</span><span class="p">)</span> <span class="c"># right outer join with a static DF</span></code></pre></div> +<span class="n">streamingDf</span><span class="o">.</span><span class="n">join</span><span class="p">(</span><span class="n">staticDf</span><span class="p">,</span> <span class="s2">"type"</span><span class="p">)</span> <span class="c1"># inner equi-join with a static DF</span> +<span class="n">streamingDf</span><span class="o">.</span><span class="n">join</span><span class="p">(</span><span class="n">staticDf</span><span class="p">,</span> <span class="s2">"type"</span><span class="p">,</span> <span class="s2">"right_join"</span><span class="p">)</span> <span class="c1"># right outer join with a static DF</span></code></pre></figure> </div> </div> @@ -878,7 +1002,7 @@ Since this windowing is based on the time in the data, the time 12:04 should be <ul> <li> - <p><code>count()</code> - Cannot return a single count from a streaming Dataset. Instead, use <code>ds.groupBy.count()</code> which returns a streaming Dataset containing a running count.</p> + <p><code>count()</code> - Cannot return a single count from a streaming Dataset. Instead, use <code>ds.groupBy.count()</code> which returns a streaming Dataset containing a running count. </p> </li> <li> <p><code>foreach()</code> - Instead use <code>ds.writeStream.foreach(...)</code> (see next section).</p> @@ -897,7 +1021,7 @@ returned through <code>Dataset.writeStream()</code>. You will have to specify on <ul> <li> - <p><em>Details of the output sink:</em> Data format, location, etc.</p> + <p><em>Details of the output sink:</em> Data format, location, etc. </p> </li> <li> <p><em>Output mode:</em> Specify what gets written to the output sink.</p> @@ -914,23 +1038,86 @@ returned through <code>Dataset.writeStream()</code>. You will have to specify on </ul> <h4 id="output-modes">Output Modes</h4> -<p>There are two types of output mode currently implemented.</p> +<p>There are a few types of output modes.</p> <ul> <li> - <p><strong>Append mode (default)</strong> - This is the default mode, where only the new rows added to the result table since the last trigger will be outputted to the sink. This is only applicable to queries that <em>do not have any aggregations</em> (e.g. queries with only <code>select</code>, <code>where</code>, <code>map</code>, <code>flatMap</code>, <code>filter</code>, <code>join</code>, etc.).</p> + <p><strong>Append mode (default)</strong> - This is the default mode, where only the +new rows added to the Result Table since the last trigger will be +outputted to the sink. This is supported for only those queries where +rows added to the Result Table is never going to change. Hence, this mode +guarantees that each row will be output only once (assuming +fault-tolerant sink). For example, queries with only <code>select</code>, +<code>where</code>, <code>map</code>, <code>flatMap</code>, <code>filter</code>, <code>join</code>, etc. will support Append mode.</p> + </li> + <li> + <p><strong>Complete mode</strong> - The whole Result Table will be outputted to the sink after every trigger. + This is supported for aggregation queries.</p> </li> <li> - <p><strong>Complete mode</strong> - The whole result table will be outputted to the sink.This is only applicable to queries that <em>have aggregations</em>.</p> + <p><strong>Update mode</strong> - (<em>not available in Spark 2.1</em>) Only the rows in the Result Table that were +updated since the last trigger will be outputted to the sink. +More information to be added in future releases.</p> </li> </ul> +<p>Different types of streaming queries support different output modes. +Here is the compatibility matrix.</p> + +<table class="table"> + <tr> + <th>Query Type</th> + <th></th> + <th>Supported Output Modes</th> + <th>Notes</th> + </tr> + <tr> + <td colspan="2" valign="middle"><br />Queries without aggregation</td> + <td>Append</td> + <td> + Complete mode note supported as it is infeasible to keep all data in the Result Table. + </td> + </tr> + <tr> + <td rowspan="2">Queries with aggregation</td> + <td>Aggregation on event-time with watermark</td> + <td>Append, Complete</td> + <td> + Append mode uses watermark to drop old aggregation state. But the output of a + windowed aggregation is delayed the late threshold specified in `withWatermark()` as by + the modes semantics, rows can be added to the Result Table only once after they are + finalized (i.e. after watermark is crossed). See + <a href="#handling-late-data">Late Data</a> section for more details. + <br /><br /> + Complete mode does drop not old aggregation state since by definition this mode + preserves all data in the Result Table. + </td> + </tr> + <tr> + <td>Other aggregations</td> + <td>Complete</td> + <td> + Append mode is not supported as aggregates can update thus violating the semantics of + this mode. + <br /><br /> + Complete mode does drop not old aggregation state since by definition this mode + preserves all data in the Result Table. + </td> + </tr> + <tr> + <td></td> + <td></td> + <td></td> + <td></td> + </tr> +</table> + <h4 id="output-sinks">Output Sinks</h4> <p>There are a few types of built-in output sinks.</p> <ul> <li> - <p><strong>File sink</strong> - Stores the output to a directory. As of Spark 2.0, this only supports Parquet file format, and Append output mode.</p> + <p><strong>File sink</strong> - Stores the output to a directory. </p> </li> <li> <p><strong>Foreach sink</strong> - Runs arbitrary computation on the records in the output. See later in the section for more details.</p> @@ -954,7 +1141,7 @@ returned through <code>Dataset.writeStream()</code>. You will have to specify on <th>Notes</th> </tr> <tr> - <td><b>File Sink</b><br />(only parquet in Spark 2.0)</td> + <td><b>File Sink</b></td> <td>Append</td> <td><pre>writeStream<br /> .format("parquet")<br /> .start()</pre></td> <td>Yes</td> @@ -980,7 +1167,14 @@ returned through <code>Dataset.writeStream()</code>. You will have to specify on <td><pre>writeStream<br /> .format("memory")<br /> .queryName("table")<br /> .start()</pre></td> <td>No</td> <td>Saves the output data as a table, for interactive querying. Table name is the query name.</td> - </tr> + </tr> + <tr> + <td></td> + <td></td> + <td></td> + <td></td> + <td></td> + </tr> </table> <p>Finally, you have to call <code>start()</code> to actually start the execution of the query. This returns a StreamingQuery object which is a handle to the continuously running execution. You can use this object to manage the query, which we will discuss in the next subsection. For now, letâs understand all this with a few examples.</p> @@ -988,7 +1182,7 @@ returned through <code>Dataset.writeStream()</code>. You will have to specify on <div class="codetabs"> <div data-lang="scala"> - <div class="highlight"><pre><code class="language-scala" data-lang="scala"><span class="c1">// ========== DF with no ag
<TRUNCATED> --------------------------------------------------------------------- To unsubscribe, e-mail: commits-unsubscr...@spark.apache.org For additional commands, e-mail: commits-h...@spark.apache.org
