I have understood that real networks have routing changes, leading to out-of-order delivered IP packets. Real networks also have frequent packet loss.
However, what I have not understood is why could a perfectly operating network in addition to out-of-order and lost packets deliver packets as duplicates. What could cause delivery of such duplicate packets?
Could spanning tree protocol (STP) before converging on a spanning tree cause duplicate packets?
Note that I don't currently have a network that would frequently duplicate packets, so this is more of a theoretical question than a practical one.
Best Answer
If a network suffers frequent packet loss it is either badly designed or its load has outgrown its capacity. Packet loss does appear in almost any network but should usually be below 1 ppm.
(M/R)STP can cause duplicate packets when it's not configured correctly. Fast start should not be used on inter-switch links as it effectively causes loops for short periods.
Additionally, various link aggregation misconfigurations can cause duplicates, especially with "static", non-protocol LAGs.
Also, bloated router buffers can cause duplicates when the sender starts resending TCP segments that are still buffered on the router. Unbalanced links with different latencies and load-balancing can very easily cause this.