Which of the following is true regarding RDDs?

RDDs, or Resilient Distributed Datasets, are a core abstraction in Apache Spark that embodies several important characteristics of the framework. The statement asserting that RDDs are distributed across multiple nodes is true and highlights one of the primary features of RDDs. Spark's ability to process large data sets efficiently stems from its design to distribute data across a cluster of machines. This distribution allows concurrent processing of data, leading to improved performance and utilization of resources.

By being distributed, RDDs benefit from parallel computing capabilities, which is critical for processing large volumes of data quickly. This characteristic also facilitates load balancing and resource optimization, as calculations can be executed on different nodes simultaneously. The distributed nature of RDDs allows for high scalability when dealing with big data workloads.

In contrast, RDDs are immutable, meaning they cannot be modified after creation, and they rely on lazy evaluation, thus cannot be computed immediately. Furthermore, RDDs are designed with fault tolerance in mind, as they can recover from node failures through lineage information. These aspects are crucial in ensuring that computations remain reliable and that the data processing pipeline is resilient to failures, further emphasizing the significance of their distributed nature.