Tag Archives: functional programming

Scala On Spark – Streak

This is yet another programming example in my Scala-on-Spark blog series. Again, while it starts with the same minuscule weather data used in previous examples of the blog series, it can be viewed as an independent programming exercise.

In this example, we’re going to create a table that shows the streaks of consecutive months with non-zero precipitation.

Result should be similar to the following:

We’ll explore using Spark’s window functions in this example. As a side note, some of the previous examples in the blog series could be resolved using window functions as well. By means of aggregating over partitioned sliding windows of data, Spark’s window functions readily perform certain kinds of complex aggregations which would otherwise require repetitive nested groupings. They are similar to how PostgreSQL’s window functions work.

Now, let’s load up the same old minuscule weather data.

First, create a DataFrame of precipitation by weather station and month and filter it to consist of only months with positive precipitation..

Next, using window function, we capture sequences of row numbers ordered by month over partitions by weather station. For each row, we then use an UDF to calculate the base date by dating back from the corresponding month of the row in accordance with the row number. As shown in the following table, these base dates help trace chunks of contiguous months back to their common base dates.

Finally, we apply another row-number window function, but this time, over partitions by weather station as well as base date. This partitioning allows contiguous common base dates to generate new row numbers as the wanted streaks.

Using the same logic flow, we can also generate similar streak reports for temperature high/low (e.g. streak of temperature high above 75F). I’ll leave that as exercise for the readers.

Scala On Spark – Sum Over Periods

This is another programming example in my Scala-on-Spark blog series. While it uses the same minuscule weather data created in the first example of the blog series, it can be viewed as an independent programming exercise.

In this example, we want a table of total precipitation over custom past periods by weather stations. The specific periods in this example are the previous month, previous 3 months, and all previous months. We have data from July through December, and let’s say it’s now January hence the previous month is December.

The result should be like this:

User-defined functions (UDF) will be used in this example. Spark’s UDF supplements its API by allowing the vast library of Scala (or any of the other supported languages) functions to be used. That said, a method from Spark’s API should be picked over an UDF of same functionality as the former would likely perform more optimally.

First, let’s load up the said weather data.

We first create a DataFrame of precipitation by weather station and month, each with the number of months that lag the current month.

Next, we combine the list of months-lagged with monthly precipitation by means of a UDF to create a map column. To do that, we use Scala’s zip method within the UDF to create a list of tuples from the two input lists and convert the resulting list into a map.

Note that the map content might look different depending on when it is generated, as the months-lagged is relative to the current month when the application is run.

Using another UDF to sum precipitation counting backward from the previous months based on the number of months lagged, we create the result DataFrame.

Again, note that the months-lagged is relative to the current month when the application is executed, hence the months-lagged parameters for the aggMapValues UDF should be adjusted accordingly.

We can use similar approach to come up with a table for temperature high/low over the custom periods. Below are the steps for creating the result table for temperature high.

I’ll leave creating the temperature low result table as a programming exercise for the readers. Note that rather than calculating temperature high and low separately, one could aggregate both of them together in some of the steps with little code change. For those who are up for a slightly more challenging exercise, both temperature high and low data can in fact be transformed together in every step of the way.

Scala On Spark – Cumulative Pivot Sum

In a couple of recent R&D projects, I was using Apache Spark rather extensively to address some data processing needs on Hadoop clusters. Although there is an abundance of big data processing platforms these days, it didn’t take long for me to settle on Spark. One of the main reasons is that the programming language for the R&D is Scala, which is what Spark itself is written in. In particular, Spark’s inherent support for functional programming and compositional transformations on immutable data enables high performance at scale as well as readability. Other main reasons are very much in line with some of the key factors attributing to Spark’s rising popularity.

I’m starting a mini blog series on Scala-on-Spark (SoS) with each blog post demonstrating with some Scala programming example on Apache Spark. In the blog series, I’m going to illustrate how the functionality-rich SoS is able to resolve some non-trivial data processing problems with seemingly little effort. If nothing else, they are good brain-teasing programming exercise in Scala on Spark.

As the source data for the example, let’s consider a minuscule set of weather data stored in a DataFrame, which consists of the following columns:

  • Weather Station ID
  • Start Date of a half-month period
  • Temperature High (in Fahrenheit) over the period
  • Temperature Low (in Fahrenheit) over the period
  • Total Precipitation (in inches) over the period

Note that with a properly configured Spark cluster, the methods illustrated in the following example can be readily adapted to handle much more granular data at scale – e.g. down to sub-hourly weather data from tens of thousands of weather stations. It’s also worth mentioning that there can be other ways to solve the problems presented in the examples.

For illustration purpose, the following code snippets are executed on a Spark Shell. First thing is to generate a DataFrame with the said columns of sample data, which will be used as source data for this example and a couple following ones.

In this first example, the goal is to generate a table of cumulative precipitation by weather stations in month-by-month columns. By ‘cumulative sum’, it means the monthly precipitation will be cumulated from one month over to the next one (i.e. rolling sum). In other words, if July’s precipitation is 2 inches and August’s is 1 inch, the figure for August will be 3 inches. The result should look like the following table:

First, we transform the original DataFrame to include an additional year-month column, followed by using Spark’s groupBy, pivot and agg methods to generate the pivot table.

Next, we assemble a list of the year-month columns and traverse the list using method foldLeft, which is one of the most versatile Scala functions for custom iterative transformations. In this particular case, the data to be transformed by foldLeft is a tuple of (DataFrame, Double). Normally, transforming the DataFrame alone should suffice, but in this case we need an additional value to address to rolling cumulation requirement.

The tuple’s first DataFrame-type element, with monthlyPrecipDF as its initial value, will be transformed using the binary operator function specified as foldLeft’s second argument (i.e. (acc, c) => …). As for the tuple’s second Double-type element, with the first year-month as its initial value it’s for carrying the current month value over to the next iteration. The end result is a (DataFrame, Double) tuple successively transformed month-by-month.

Similar pivot aggregations can be applied to temperature high’s/low’s as well, with method sum replaced with method max/min.

Finally, we compute cumulative temperature high/low like cumulative precipitation, by replacing method sum with iterative max/min using Spark’s when-otherwise method.

Text Mining With Akka Streams

Reactive Systems, whose core characteristics are declared in the Reactive Manifesto, have started to emerge in recent years as message-driven systems that emphasize scalability, responsiveness and resilience. It’s pretty clear from the requirements that a system can’t be simply made Reactive. Rather, it should be built from the architectural level to be Reactive.

Akka’s actor systems, which rely on asynchronous message-passing among lightweight loosely-coupled actors, serve a great run-time platform for building Reactive Systems on the JVM (Java Virtual Machine). I posted a few blogs along with sample code about Akka actors in the past. This time I’m going to talk about something different but closely related.

Reactive Streams

While bearing a similar name, Reactive Streams is a separate initiative that mandates its implementations to be capable of processing stream data asynchronously and at the same time automatically regulating the stream flows in a non-blocking fashion.

Akka Streams, built on top of Akka actor systems, is an implementation of Reactive Streams. Equipped with the back-pressure functionality, it eliminates the need of manually buffering stream flows or custom-building stream buffering mechanism to avoid buffer overflow problems.

Extracting n-grams from text

In text mining, n-grams are useful data in the area of NLP (natural language processing). In this blog post, I’ll illustrate extracting n-grams from a stream of text messages using Akka Streams with Scala as the programming language.

First thing first, let’s create an object with methods for generating random text content:

Source code: TextMessage.scala

Some minimal effort has been made to generate random clauses of likely pronounceable fake words along with punctuations. To make it a little more flexible, lengths of individual words and clauses would be supplied as parameters.

Next, create another object with text processing methods responsible for extracting n-grams from input text, with n being an input parameter. Using Scala’s sliding(size, step) iterator method with size n and step default to 1, a new iterator of sliding window view is generated to produce the wanted n-grams.

Source code: TextProcessor.scala

Now that the text processing tools are in place, we can focus on building the main streaming application in which Akka Streams plays the key role.

First, make sure we have the necessary library dependencies included in build.sbt:

Source code: build.sbt

As Akka Streams is relatively new development work, more recent Akka versions (2.4.9 or higher) should be used.

Let’s start with a simple stream for this text mining application:

Source code: NgramStream_v01.scala

As shown in the source code, constructing a simple stream like this is just defining and chaining together the text-generating source, the text-processing flow and the text-display sink as follows:

Graph DSL

Akka Streams provides a Graph DSL (domain-specific language) that helps build the topology of stream flows using predefined fan-in/fan-out functions.

What Graph DSL does is somewhat similar to how Apache Storm‘s TopologyBuilder pieces together its spouts (i.e. stream sources), bolts (i.e. stream processors) and stream grouping/partitioning functions, as illustrated in a previous blog about HBase streaming.


Now, let’s branch off the stream using Graph DSL to illustrate how the integral back-pressure feature is at play.

Source code: NgramStream_v02.scala

Streaming to a file should be significantly slower than streaming to the console. To make the difference more noticeable, a delay is deliberately added to streaming each line of text in the file sink.

Running the application and you will notice that the console display is slowed down. It’s the result of the upstream data flow being regulated to accommodate the relatively slow file I/O outlet even though the other console outlet is able to consume relatively faster – all that being conducted in a non-blocking fashion.

Graph DSL create() methods

To build a streaming topology using Graph DSL, you’ll need to use one of the create() methods defined within trait GraphApply, which is extended by object GraphDSL. Here are the signatures of the create() methods:

Note that the sbt-boilerplate template language is needed to interpret the create() method being used in the application that takes multiple stream components as input parameters.

Materialized values

In Akka Streams, materializing a constructed stream is the step of actually running the stream with the necessary resources. To run the stream, the implicitly passed factory method ActorMaterializer() is required to allocate the resources for stream execution. That includes starting up the underlying Akka actors to process the stream.

Every processing stage of the stream can produce a materialized value. By default, using the via(flow) and to(sink) functions, the materialized value of the left-most stage will be preserved. As in the following example, for graph1, the materialized value of the source is preserved:

To allow one to selectively capture the materialized values of the specific stream components, Akka Streams provides functions viaMat(flow) and toMat(sink) along with a combiner function, Keep. As shown in the above example, for graph2, the materialized value of the flow is preserved, whereas for graph3, materialized values for both the flow and sink are preserved.

Back to our fileSink function as listed below, toMat(fileIOSink)(Keep.right) instructs Akka Streams to keep the materialized value of the fileIOSink as a Future value of type IOResult:

Using Graph DSL, as seen earlier in the signature of the create() method, one can select what materialized value is to be preserved by specifying the associated stream components accordingly as the curried parameters:

In our case, we want the materialized value of fileSink, thus the curried parameters should look like this:

Defining the stream graph

Akka Streams provides a number of functions for fan-out (e.g. Broadcast, Balance) and fan-in (e.g. Merge, Concat). In our example, we want a simple topology with a single text source and the same n-gram generator flow branching off to two sinks in parallel:

Adding a message counter

Let’s further expand our n-gram extraction application to include displaying a count. A simple count-flow is created to map each message string into numeric 1, and a count-sink to sum up all these 1′s streamed to the sink. Adding them as the third flow and sink to the existing stream topology yields something similar to the following:

Source code: NgramStream_v03.scala

Full source code of the application is at GitHub.

Final thoughts

Having used Apache Storm, I see it a rather different beast compared with Akka Streams. A full comparison between the two would obviously be an extensive exercise by itself, but it suffices to say that both are great platforms for streaming applications.

Perhaps one of the biggest differences between the two is that Storm provides granular message delivery options (at most / at least / exactly once, guaranteed message delivery) whereas Akka Streams by design questions the premise of reliable messaging on distributed systems. For instance, if guaranteed message delivery is a requirement, Akka Streams would probably not be the best choice.

Back-pressure has recently been added to Storm’s v.1.0.x built-in feature list, so there is indeed some flavor of reactiveness in it. Aside from message delivery options, choosing between the two technologies might be a decision basing more on other factors such as engineering staff’s expertise, concurrency model preference, etc.

Outside of the turf of typical streaming systems, Akka Streams also plays a key role as the underlying platform for an emerging service stack. Viewed as the next-generation of Spray.io, Akka HTTP is built on top of Akka Streams. Designed for building HTTP-based integration layers, Akka HTTP provides versatile streaming-oriented HTTP routing and request/response transformation mechanism. Under the hood, Akka Streams’ back-pressure functionality regulates data streaming between the server and the remote client, consequently conserving memory utilization on the server.