Requirements are the key to the development of every software product. Resource Description Framework (RDF) systems, propelled by the rise of Linked Data, have necessitated the development of nonrelational models for web-based representation of diverse and incomplete data. As a new and emerging storage format for Linked Data, it requires specialised data management systems for storage and querying purposes. This has led to several RDF engines. It is seen that RDF engine development lacks the proper requirement engineering models. RDF engines also have requirements of storage, reproducibility, scalability, etc., also known as non-functional requirements (NFRs). In the case of the development of traditional software, these NFRs are well understood and established. However, in RDF engines, much of the existing knowledge regarding NFRs is not much applicable. The ever-increasing RDF-based data has caused a shift in the application of NFRs. The idea and interpretations of NFRs in an RDF-based data context (e.g. storage, scalability, reproducibility, etc.) must be rethought. By reviewing case studies from traditional software development, the paper identifies the challenges in adapting these NFRs to RDF engines and proposes research directions to address them. Putting emphasis on the need for specialised data management systems, the paper outlines trade-offs among different NFRs in the RDF domain.