What is data locality and why is it critical in distributed processing frameworks?

Data locality means processing data where it physically resides to reduce network transfer overhead.

In distributed systems like Spark and Hadoop, moving computation closer to data significantly reduces network I/O, which is one of the most expensive operations. Data locality levels include process-local, node-local, rack-local, and off-rack. Scheduling engines try to execute tasks where data already exists to maximize performance and minimize shuffle operations.