We've learned in the previous PySpark blog posts about the serialization overhead between the Python application and JVM. An intrinsic actor of this overhead are Python serializers that will be the topic of this article and hopefully, will provide a more complete overview of the Python <=> JVM serialization.
Modern data space is an exciting place with a lot of innovation these last years. The single drawback of that movement are all the new buzz words and the time required to understand and classify them into something we could use in the organization or not. Recently I see more and more "data contracts" in social media. It's also a new term and I'd like to see if and how it revolutionizes the data space.
There are days like that. You inherit a code and it doesn't really work as expected. While digging into issues you find usual weird warnings but also several new things. For me one of these things was the "Generated method too long to be JIT compiled..." info message.
It's time to start the 4th part of the Table file formats series. This time the topic will be Change Data Capture, so how to stream all changes made on the table. As for the 3rd part, I'm going to start with Delta Lake.
Message bus is a common architectural design in the Enterprise Design Patterns. But it's also present at a lower level to enable the event-driven behavior. Apache Spark is not an exception. It uses a publish/subscribe approach in various places.
PySpark has been getting interesting improvements making it more Python and user-friendly in each release. However, it's not the single Python-based framework for distributed data processing and people talk more and more often about the alternatives like Dask or Ray. Since both are completely new for me, I'm going to use this blog post to shed some light on them, and why not plan a deeper exploration next year?
Let's suppose you store the partitioned data under the /data/mydir location. What will be the difference if you read this directory with Apache Spark as /data/mydir/ and /data/mydir/* ? You should find the answer to the question just below.
And it shouldn't be, right? After all, it's "just" about using a Unit Test framework and defining the test cases. Well, that's "just" a theory!
The topic of this blog post is one of my first big surprises while I was learning the debugging of PySpark jobs. Usually I'm running the code locally in debug mode and the defined breakpoints help me understand what happens. That time, it was different!
After Delta Lake and Apache Iceberg it's time to see the reading part of Apache Hudi. Despite an apparent similarity with the aforementioned table formats, Apache Hudi has an interesting reading specificity related to the different table types.
The Scala API of Apache Spark SQL has various ways of transforming the data, from the native and User-Defined Function column-based functions, to more custom and row-level map functions. PySpark doesn't have this mapping feature but does have the User-Defined Functions with an optimized version called vectorized UDF!
Last week you could read about data reading in Delta Lake. Today it's time to cover this part in Apache Iceberg!
Observability is a hot topic nowadays, not only for the data but also the software industry. Apache Spark innovates in this field a lot, including new metrics for Structured Streaming and an important update added in the 3.0.0 release that I missed at the time, which are the observable metrics.
In the previous blog post about Delta Lake you discovered the logic for the writing part. Meantime Delta Lake 2 was released and it's for this brand new version that I'm going to share with you some findings related to the data reading.
Pushdowns in Apache Spark are great to delegate some operations to the data sources. It's a great way to reduce the data volume to be processed in the job. However, there is one important gotcha. Watch out the definition of your predicate because from time to time, even though the pushdown predicate is supported by the data source, the predicate can still be executed by the Apache Spark job!
You want to learn data engineering but have no idea where and how to start in this wide domain? Check if Become a Data Engineer can help you 💪
Bartosz Konieczny