Hi Mahmood,
I think FOSS often trades licensing/purchase costs for operational costs.
Sure, the software is "free" but because such software is rarely an exact fit for your use case off the shelf, you have to spend time (money) to:
customize for your operation
integrate into other operational systems - billing, provisioning, support, etc.
maintaining that expertise over time - either in-house or contracted out, will cost you $
Fred and Alex have pointed out some additional things on this.
But I think arguments can still be made for FOSS in these areas:
You can control the software, customize it, and are not reliant on a 3rd party software vendor to keep the software current.
With traditional softswitch vendors like MetaSwitch, Broadsoft, Squire all putting these products on the back-burner and not
investing in any more development of them, that could be a big organizational benefit.
The FOSS options are also, at this point in time, more likely to get future feature development. E.g. Asterisk, FreeSwitch,
Kamailio all support WebRTC and rich media - and the legacy alternatives do not.
They all help you build highly fault tolerant systems with any number of architectures to suit your needs.
They can all be made to scale from thousands, to millions of subscribers, and the legacy alternatives do not.
I would say, IF there is an off-the-shelf, commercially supported system that suits your needs now and into the future,
and has a cost model that meets your business needs, and supports your scale, you are almost always better off going with that.
But if the cost model is detached from your needs (nickel and dimed to death on 'feature licenses'), or, they don't
support features you need, or, you don't think they are going to be further developed or supported into the future,
or they won't scale the way you need, then, open source gives you the ability to build your own without having to
reinvent all the wheels.