VMware NSX
Tunnels are formed between transport nodes as needed and it is expected to see a different number of tunnels from one transport node to the next.
Once an ESX host transport node has at least one VM running on it that uses an NSX-backed segment, at least one tunnel will be formed. The segment is dependent on an NSX logical router and that logical router may be configured to use one or more edge nodes. You should see at least one tunnel between the host where the VM is running and the edge node backing the segment/logical router. If the logical router is configured to use more than one edge node (active-active or active-standby), tunnels will be formed between the host and both edge nodes.
When there are other ESX host transport nodes with VMs running on them that use the same segment as a VM running on the original host in question, tunnels will be formed between the original host and the other hosts with VMs on the same segments.
It can be fairly difficult in a large environment to get an accurate accounting for all of the tunnels on a host and why they were formed. The main thing to be concerned with is that all tunnels are up.