In this chapter, I gave an overview of what the MapReduce pattern looked like in a general sense and demonstrated how MapReduce works with some example code. From there, we reviewed the MapReduce pattern as applied to serverless architectures. We stepped through the details of implementing this pattern by parsing 1.5 GB of email data and counting the unique occurrences of From
and To
email addresses. I showed that a serverless system could be built using this pattern to perform our task in less than a minute, on average.
We covered some of the limitations of this pattern when implemented on a serverless platform. Finally, we discussed alternative solutions for general data analysis problems using serverless platforms such as AWS Athena and managed systems such as EMR, as well as ways to use a centralized data store such as Redis in a serverless MapReduce system.