You are viewing a plain text version of this content. The canonical link for it is here.
Posted to github@beam.apache.org by GitBox <gi...@apache.org> on 2020/10/19 22:04:27 UTC
[GitHub] [beam] davidcavazos commented on a change in pull request #12963: [BEAM-10983] Add getting started from Spark page
davidcavazos commented on a change in pull request #12963:
URL: https://github.com/apache/beam/pull/12963#discussion_r508069958
##########
File path: website/www/site/content/en/get-started/from-spark.md
##########
@@ -0,0 +1,245 @@
+---
+title: "Getting started from Apache Spark"
+---
+<!--
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+-->
+
+# Getting started from Apache Spark
+
+{{< localstorage language language-py >}}
+
+If you already know [_Apache Spark_](http://spark.apache.org/),
+learning _Apache Beam_ is easy.
+Beam and Spark are mostly equivalent, so you already know the basic concepts.
Review comment:
Thanks, done
##########
File path: website/www/site/content/en/get-started/from-spark.md
##########
@@ -0,0 +1,245 @@
+---
+title: "Getting started from Apache Spark"
+---
+<!--
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+-->
+
+# Getting started from Apache Spark
+
+{{< localstorage language language-py >}}
+
+If you already know [_Apache Spark_](http://spark.apache.org/),
+learning _Apache Beam_ is easy.
+Beam and Spark are mostly equivalent, so you already know the basic concepts.
+
+A collection of elements in Spark is called a _Resilient Distributed Dataset_ (RDD),
+while in Beam it's called a _Parallel Collection_ (PCollection).
+A PCollection in Beam does _not_ have any ordering guarantees.
+
+Likewise, a transform in Beam is called a _Parallel Transform_ (PTransform).
+
+Here are some examples of common operations and their equivalent between PySpark and Beam.
+
+## Overview
+
+Here's a simple example of a PySpark pipeline that takes the numbers from one to four,
+multiplies them by two, adds all the values together, and prints the result.
+
+{{< highlight py >}}
+import pyspark
+
+with pyspark.SparkContext() as sc:
+ result = (
+ sc.parallelize([1, 2, 3, 4])
+ .map(lambda x: x * 2)
+ .reduce(lambda x, y: x + y)
+ )
+ print(result)
+{{< /highlight >}}
+
+In Beam you _pipe_ your data through the pipeline using the
Review comment:
I added a note just below the example.
##########
File path: website/www/site/content/en/get-started/from-spark.md
##########
@@ -0,0 +1,245 @@
+---
+title: "Getting started from Apache Spark"
+---
+<!--
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+-->
+
+# Getting started from Apache Spark
+
+{{< localstorage language language-py >}}
+
+If you already know [_Apache Spark_](http://spark.apache.org/),
+learning _Apache Beam_ is easy.
+Beam and Spark are mostly equivalent, so you already know the basic concepts.
+
+A collection of elements in Spark is called a _Resilient Distributed Dataset_ (RDD),
+while in Beam it's called a _Parallel Collection_ (PCollection).
+A PCollection in Beam does _not_ have any ordering guarantees.
+
+Likewise, a transform in Beam is called a _Parallel Transform_ (PTransform).
+
+Here are some examples of common operations and their equivalent between PySpark and Beam.
+
+## Overview
+
+Here's a simple example of a PySpark pipeline that takes the numbers from one to four,
+multiplies them by two, adds all the values together, and prints the result.
+
+{{< highlight py >}}
+import pyspark
+
+with pyspark.SparkContext() as sc:
+ result = (
+ sc.parallelize([1, 2, 3, 4])
+ .map(lambda x: x * 2)
+ .reduce(lambda x, y: x + y)
+ )
+ print(result)
+{{< /highlight >}}
+
+In Beam you _pipe_ your data through the pipeline using the
+_pipe operator_ `|` like `data | beam.Map(...)` instead of chaining
+methods like `data.map(...)`, but they're doing the same thing.
+
+Here's how an equivalent pipeline looks like in Beam.
+
+{{< highlight py >}}
+import apache_beam as beam
+
+with beam.Pipeline() as pipeline:
+ result = (
+ pipeline
+ | beam.Create([1, 2, 3, 4])
+ | beam.Map(lambda x: x * 2)
+ | beam.CombineGlobally(sum)
+ | beam.Map(print)
+ )
+{{< /highlight >}}
+
+> ℹ️ Note that we called `print` inside a `Map` transform.
+> That's because we can only access the elements of a PCollection
+> from within a PTransform.
+
+A label can optionally be added to a transform using the
+_right shift operator_ `>>` like `data | 'My description' >> beam.Map(...)`.
+This serves both as comments and makes your pipeline easier to debug.
+
+This is how the pipeline looks after adding labels.
+
+{{< highlight py >}}
+import apache_beam as beam
+
+with beam.Pipeline() as pipeline:
+ result = (
+ pipeline
+ | 'Create numbers' >> beam.Create([1, 2, 3, 4])
+ | 'Multiply by two' >> beam.Map(lambda x: x * 2)
+ | 'Sum everything' >> beam.CombineGlobally(sum)
+ | beam.Map(print)
+ )
+{{< /highlight >}}
+
+## Setup
+
+Here's a comparison on how to get started both in PySpark and Beam.
+
+{{< table >}}
+<table>
+<tr>
+ <th></th>
+ <th>PySpark</th>
+ <th>Beam</th>
+</tr>
+<tr>
+ <td><b>Install</b></td>
+ <td><code>$ pip install pyspark</code></td>
+ <td><code>$ pip install apache-beam</code></td>
+</tr>
+<tr>
+ <td><b>Imports</b></td>
+ <td><code>import pyspark</code></td>
+ <td><code>import apache_beam as beam</code></td>
+</tr>
+<tr>
+ <td><b>Creating a<br>local pipeline</b></td>
+ <td>
+ <code>with pyspark.SparkContext() as sc:</code><br>
+ <code> # Your pipeline code here.</code>
+ </td>
+ <td>
+ <code>with beam.Pipeline() as pipeline:</code><br>
+ <code> # Your pipeline code here.</code>
+ </td>
+</tr>
+<tr>
+ <td><b>Creating values</b></td>
+ <td><code>values = sc.parallelize([1, 2, 3, 4])</code></td>
+ <td><code>values = pipeline | beam.Create([1, 2, 3, 4])</code></td>
+</tr>
+<tr>
+ <td><b>Creating<br>key-value pairs</b></td>
+ <td>
+ <code>pairs = sc.parallelize([</code><br>
+ <code> ('key1', 'value1'),</code><br>
+ <code> ('key2', 'value2'),</code><br>
+ <code> ('key3', 'value3'),</code><br>
+ <code>])</code>
+ </td>
+ <td>
+ <code>pairs = pipeline | beam.Create([</code><br>
+ <code> ('key1', 'value1'),</code><br>
+ <code> ('key2', 'value2'),</code><br>
+ <code> ('key3', 'value3'),</code><br>
+ <code>])</code>
+ </td>
+</tr>
+<tr>
+ <td><b>Running a<br>local pipeline</b></td>
+ <td><code>$ spark-submit spark_pipeline.py</code></td>
+ <td><code>$ python beam_pipeline.py</code></td>
+</tr>
+</table>
+{{< /table >}}
+
+## Transforms
+
+Here are the equivalents of some common transforms in both PySpark and Beam.
+
+{{< table >}}
Review comment:
Sure, the table gets a little uglier, and definitely over 80 characters long, but I can do that.
PS: I tried using reference links, but the website doesn't like them to be inside a table (even HTML instead of Markdown), so they have to be inlined if we want them here.
##########
File path: website/www/site/content/en/get-started/from-spark.md
##########
@@ -0,0 +1,245 @@
+---
+title: "Getting started from Apache Spark"
+---
+<!--
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+-->
+
+# Getting started from Apache Spark
+
+{{< localstorage language language-py >}}
+
+If you already know [_Apache Spark_](http://spark.apache.org/),
+learning _Apache Beam_ is easy.
+Beam and Spark are mostly equivalent, so you already know the basic concepts.
+
+A collection of elements in Spark is called a _Resilient Distributed Dataset_ (RDD),
+while in Beam it's called a _Parallel Collection_ (PCollection).
+A PCollection in Beam does _not_ have any ordering guarantees.
+
+Likewise, a transform in Beam is called a _Parallel Transform_ (PTransform).
+
+Here are some examples of common operations and their equivalent between PySpark and Beam.
+
+## Overview
+
+Here's a simple example of a PySpark pipeline that takes the numbers from one to four,
+multiplies them by two, adds all the values together, and prints the result.
+
+{{< highlight py >}}
+import pyspark
+
+with pyspark.SparkContext() as sc:
+ result = (
+ sc.parallelize([1, 2, 3, 4])
+ .map(lambda x: x * 2)
+ .reduce(lambda x, y: x + y)
+ )
+ print(result)
+{{< /highlight >}}
+
+In Beam you _pipe_ your data through the pipeline using the
+_pipe operator_ `|` like `data | beam.Map(...)` instead of chaining
+methods like `data.map(...)`, but they're doing the same thing.
+
+Here's how an equivalent pipeline looks like in Beam.
+
+{{< highlight py >}}
+import apache_beam as beam
+
+with beam.Pipeline() as pipeline:
+ result = (
+ pipeline
+ | beam.Create([1, 2, 3, 4])
+ | beam.Map(lambda x: x * 2)
+ | beam.CombineGlobally(sum)
+ | beam.Map(print)
+ )
+{{< /highlight >}}
+
+> ℹ️ Note that we called `print` inside a `Map` transform.
+> That's because we can only access the elements of a PCollection
+> from within a PTransform.
+
+A label can optionally be added to a transform using the
+_right shift operator_ `>>` like `data | 'My description' >> beam.Map(...)`.
+This serves both as comments and makes your pipeline easier to debug.
+
+This is how the pipeline looks after adding labels.
+
+{{< highlight py >}}
+import apache_beam as beam
+
+with beam.Pipeline() as pipeline:
+ result = (
+ pipeline
+ | 'Create numbers' >> beam.Create([1, 2, 3, 4])
+ | 'Multiply by two' >> beam.Map(lambda x: x * 2)
+ | 'Sum everything' >> beam.CombineGlobally(sum)
+ | beam.Map(print)
+ )
+{{< /highlight >}}
+
+## Setup
+
+Here's a comparison on how to get started both in PySpark and Beam.
+
+{{< table >}}
+<table>
+<tr>
+ <th></th>
+ <th>PySpark</th>
+ <th>Beam</th>
+</tr>
+<tr>
+ <td><b>Install</b></td>
+ <td><code>$ pip install pyspark</code></td>
+ <td><code>$ pip install apache-beam</code></td>
+</tr>
+<tr>
+ <td><b>Imports</b></td>
+ <td><code>import pyspark</code></td>
+ <td><code>import apache_beam as beam</code></td>
+</tr>
+<tr>
+ <td><b>Creating a<br>local pipeline</b></td>
+ <td>
+ <code>with pyspark.SparkContext() as sc:</code><br>
+ <code> # Your pipeline code here.</code>
+ </td>
+ <td>
+ <code>with beam.Pipeline() as pipeline:</code><br>
+ <code> # Your pipeline code here.</code>
+ </td>
+</tr>
+<tr>
+ <td><b>Creating values</b></td>
+ <td><code>values = sc.parallelize([1, 2, 3, 4])</code></td>
+ <td><code>values = pipeline | beam.Create([1, 2, 3, 4])</code></td>
+</tr>
+<tr>
+ <td><b>Creating<br>key-value pairs</b></td>
+ <td>
+ <code>pairs = sc.parallelize([</code><br>
+ <code> ('key1', 'value1'),</code><br>
+ <code> ('key2', 'value2'),</code><br>
+ <code> ('key3', 'value3'),</code><br>
+ <code>])</code>
+ </td>
+ <td>
+ <code>pairs = pipeline | beam.Create([</code><br>
+ <code> ('key1', 'value1'),</code><br>
Review comment:
I don't have an @apache.org email so I can't open the discussion.
I thought tuples were supported better than lists. I've been using tuples for all the samples and I've never had an issue. I also think type-wise, tuples are "allowed" to have elements of any/different types, while lists should only be of one type. On a type checker like `mypy` or any other typed language, it would be a compile error to have a list of two different types, unless you use `Any` which should be discouraged.
I think I would prefer to keep these as tuples, @aaltay what do you think?
##########
File path: website/www/site/content/en/get-started/from-spark.md
##########
@@ -0,0 +1,245 @@
+---
+title: "Getting started from Apache Spark"
+---
+<!--
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+-->
+
+# Getting started from Apache Spark
+
+{{< localstorage language language-py >}}
+
+If you already know [_Apache Spark_](http://spark.apache.org/),
+learning _Apache Beam_ is easy.
+Beam and Spark are mostly equivalent, so you already know the basic concepts.
+
+A collection of elements in Spark is called a _Resilient Distributed Dataset_ (RDD),
+while in Beam it's called a _Parallel Collection_ (PCollection).
+A PCollection in Beam does _not_ have any ordering guarantees.
+
+Likewise, a transform in Beam is called a _Parallel Transform_ (PTransform).
+
+Here are some examples of common operations and their equivalent between PySpark and Beam.
+
+## Overview
+
+Here's a simple example of a PySpark pipeline that takes the numbers from one to four,
+multiplies them by two, adds all the values together, and prints the result.
+
+{{< highlight py >}}
+import pyspark
+
+with pyspark.SparkContext() as sc:
+ result = (
+ sc.parallelize([1, 2, 3, 4])
+ .map(lambda x: x * 2)
+ .reduce(lambda x, y: x + y)
+ )
+ print(result)
+{{< /highlight >}}
+
+In Beam you _pipe_ your data through the pipeline using the
+_pipe operator_ `|` like `data | beam.Map(...)` instead of chaining
+methods like `data.map(...)`, but they're doing the same thing.
+
+Here's how an equivalent pipeline looks like in Beam.
+
+{{< highlight py >}}
+import apache_beam as beam
+
+with beam.Pipeline() as pipeline:
+ result = (
+ pipeline
+ | beam.Create([1, 2, 3, 4])
+ | beam.Map(lambda x: x * 2)
+ | beam.CombineGlobally(sum)
+ | beam.Map(print)
+ )
+{{< /highlight >}}
+
+> ℹ️ Note that we called `print` inside a `Map` transform.
+> That's because we can only access the elements of a PCollection
+> from within a PTransform.
+
+A label can optionally be added to a transform using the
+_right shift operator_ `>>` like `data | 'My description' >> beam.Map(...)`.
+This serves both as comments and makes your pipeline easier to debug.
+
+This is how the pipeline looks after adding labels.
+
+{{< highlight py >}}
+import apache_beam as beam
+
+with beam.Pipeline() as pipeline:
+ result = (
+ pipeline
+ | 'Create numbers' >> beam.Create([1, 2, 3, 4])
+ | 'Multiply by two' >> beam.Map(lambda x: x * 2)
+ | 'Sum everything' >> beam.CombineGlobally(sum)
+ | beam.Map(print)
Review comment:
Added
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org