Work out nfs mount throughput instantly with clear inputs, formula shown and shareable results.
NFS throughput is limited by the same bandwidth-delay logic as TCP: outstanding requests multiplied by rsize divided by round-trip time. Small rsize values or too few outstanding requests leave a fast network idle waiting for round trips, which is why raising rsize and wsize to 1 MB and increasing the RPC slot count matter so much on higher-latency links.
NFS throughput
latency limit = rsize x outstanding requests x 8 / RTT; effective = min(0.9 x network rate, latency limit)
No. TCP handles retransmission and congestion properly, and UDP support is deprecated. NFSv4 over TCP is the correct choice.
Usually too few outstanding RPCs or a small rsize. Also check for synchronous writes, which force a round trip per operation and destroy throughput.