Category: Hadoop

Tomaz Kastrun did it: 25 posts in 25 days on Spark. Part 23 looks at Delta Live Tables:

Delta Live Tables is a framework for building reliable, maintainable, and testable data processing pipelines. User defines the transformations to be performed on the datasources and data, and the framework manages all the data engineering tasks: task orchestrations, cluster management, monitoring, data quality, and event error handling.

Delta Live Tables framework helps and manages how data is being transformed with help of target schema and can is a slight different experience with Databricks Tasks (with Apache Spark tasks in the background).

Part 24 takes us through a bit of visualization:

You can use any of the popular Python packages to do the visualisation; Plotly, Dash, Seaborn, Matplotlib, Bokeh, Leather, Glam, to name the couple and many others. Once the data is persisted in dataframe, you can use any of the packages. With the use of PySpark, plugin the Matplotlib. Here is an example

And part 25 wraps things up with links to additional resources:

To wrap up this year’s Advent of Spark 2021 – series of blogposts on Spark – it is essential to look at the list of additional learning resources for you to continue with this journey. Let’s divide this list not by type of the resource (book, on-line documentation, on-line courses, articles, Youtube channels, Discord channels, and others) but rather divide them by language flavour. Scala/Spark, R, and Python.

Great job on Tomaz’s part for gutting it out.

Tomaz Kastrun continues a series on working with Apache Spark. Part 10 looks at the DataFrame construct:

We have looked in datasets and seen that a dataset is distributed collection of data. A dataset can be constructer from JVM object and later manipulated with transformation operations (e.g.: filter(), Map(),…). API for these datasets are also available in Scala and in Java. But in both cases of Python and R, you can also access the columns or rows from datasets.

On the other hand, dataframe is organised dataset with named columns. It offers much better optimizations and computations and still resembles a typical table (as we know it from database world). Dataframes can be constructed from arrays or from matrices from variety of files, SQL tables, and datasets (RDDs). Dataframe API is available in all flavours: Java, Scala, R and Python and hence it’s popularity.

Part 11 looks at external R and Python packages and DataFrame support:

When you install Spark, the extension of not only languages but also other packages, systems is huge. For example with R, not only that you can harvest the capabilities of distributed and parallel computations, you can also extend the use of R language.

Part 12 gets into Spark SQL:

Spark SQL is a one of the Spark modules for structured data processing and analysing. Spark provides Spark SQL and also API for execution of SQL queries. Spark SQL can read data from Hive instance, but also from datasets and dataframe. The communication between Spark SQL and execution engine will always result in a dataset or datafrane.

These formats are interchangeable. So interacting with SQL against result from a different API is possible, respectively. Plugging in the Java JDBD or standard ODBC drivers will also give your SQL interface access to different sources. This unification means that developers can easily switch back and forth between different APIs based on which provides the most natural way to express a given transformation.

With API unification, user can access Spark SQL using Scala spark-shell, using Python pyspark or using R sparkR shell.

DataFrames are so popular that they’ve become the de facto standard for working with data in Spark, and .NET languages only work with DataFrames, not with the raw RDDs.

Tomaz Kastrun continues a series on Spark. Part 7 ties in R and gives us sample plotting in R and Python:

Let’s look into the local use of Spark. For R language, sparklyr package is availble and for Python pyspark is availble.

Part 8 gets us into the key data structure behind Spark’s success, the Resilient Distributed Dataset:

Spark is created around the concept of resilient distributed datasets (RDD). RDD is a fault-tolerant collection of files that can be used in parallel. RDDs can be created in two ways:
– parallelising an existing data collection in driver program
– referencing a datasets in external storage (HDFS, blob storage, shared filesystem, Hadoop InputFormat,…)

In a simple way, Spark RDD has two opeartions:
– transformations – creating a new RDD dataset on top of already existing one with the last transformation
– actions – to the action, and return a value to the driver program after running a computation on the dataset.

Part 9 looks a bit more at transformations and actions:

Two types of operations are available with RDD; transformations and actions. Transformations are lazy operations, meaning that they prepare the new RDD with every new operation but now show or return anything. We can say, that transformations are lazy because of updating existing RDD, these operations create another RDD. Actions on the other hand trigger the computations on RDD and show (return) the result of transformations.

Most modern work in Spark won’t directly use RDDs, though everything is built on top of them and it’s good to understand the foundation even if you don’t need to write all of those map(), fold(), and reduceByKey() operations yourself.