Pig Latin is the high-level data-flow language used by Apache Pig to express extract-transform-load and analysis pipelines over large datasets, compiling down to MapReduce (and later Tez/Spark) jobs on Hadoop.
Pig Latin
Pig Latin is the procedural data-flow language behind Apache Pig, created to make large-scale data processing on Hadoop expressible without writing raw MapReduce code. A Pig Latin script reads like a sequence of transformations — load, filter, group, join, and store — that the Pig engine compiles into MapReduce (and later Tez or Spark) jobs. It was one of the early attempts to give data engineers a readable, script-first path into the Hadoop ecosystem.
- Data-flow, not declarative - Unlike SQL, Pig Latin describes an ordered pipeline of operations, which maps naturally to ETL work.
- Relational operators -
LOAD,FILTER,GROUP,JOIN,FOREACH, andSTOREcover the common shapes of batch data processing. - User-defined functions - Custom logic can be plugged in via UDFs written in Java, Python, and other languages.
- Runs on Hadoop - Scripts execute across a cluster, abstracting away the distributed-execution details from the author.
In practice Pig Latin shows up in legacy big-data pipelines and batch analytics workloads that predate the SQL-on-Hadoop wave, and it is far less central today than newer engines. For an API-focused catalog it is worth tracking mostly as historical context: the data transformations Pig once handled in overnight batch jobs are increasingly exposed through APIs and streaming interfaces, which is where the discoverability and governance work now lives.
