Target session is not limiting number... max sessions is. With dedupe you wish to avoid multiplexing to get better dedupe ration so I guess this is why recommended values are such as they are (for all de-dupe appliances). As with VTLs, nothing stops you to create bunch of DD devices thus this issue stops to be the the issue (given that dedupe ration is global anyway and not per device).
My question arises from this doubt:Networker by design on the same storage node with several devices (with several target session) uses the minum target session value for all the devices. So if i have some constraints for RMAN i've to create a dedicated storagenode using for example DataDomain Devices with DDboost.
this monday i've added to a storagenode with 10physical deevices (target session 18) one DDdevice DDboost (target session 4). All devices (physical too) on that storage node ,during the night, has worked with target session 4 with a big number of mounting requests
Target session 18 on physical device? Why? Did you ever try restore from such multiplexed backup?
Ok, back to your thing... that sounds as some sort of bug. I know similar rule applies to clients and parallelism (that sucks), but it should not to devices per storage node. The best thing would be to open a ticket with support to investigate this and try to replicate it.