ffeee submitted a new resource:
Looking into that OpenWrt forum discussion, the issue with slow nft set lookups seems largely tied to how big datasets are processed on constrained devices. When a set contains a very large number of elements, performance can degrade significantly, especially if the hardware has limited CPU and memory resources. This is somewhat similar to how
video transcription online services handle large inputs - performance depends not just on size, but on how efficiently the data is structured and accessed. In the case of nftables, optimizing the composition of sets, reducing redundancy, or restructuring entries can help improve lookup times. It’s less about the lookup operation itself and more about how the data is prepared and maintained, which ultimately determines how responsive the system will be under load.
