In essence, it's an attempt at an open source management and orchestration layer for virtualized telecoms networks (i.e. 5G). In the telecoms world, since everything moves in 10 year standards cycles, they're only just moving off black box dedicated hardware towards white box.
Until now, management and orchestration has been vendor proprietary in the telecoms world, and this is an attempt to standardise an open and interoperable way of doing it, more like how we do in IT.
Don't read this to be a criticism of telecoms as being outdated or backward - they just have very different tolerances of breakage (which IT tolerates as inevitable or necessary or "move fast and break things"), so this is a big step. In many countries they have a 7 or 8x 9's uptime requirement on their main telecoms core network, and similar expectations even in access networks. NFV is network function virtualization, which is the move towards each component of a 5G network being software, rather than dedicated hardware, running in a hypervisor instead of on dedicated bespoke silicon.
In essence, it's an attempt at an open source management and orchestration layer for virtualized telecoms networks (i.e. 5G). In the telecoms world, since everything moves in 10 year standards cycles, they're only just moving off black box dedicated hardware towards white box.
Until now, management and orchestration has been vendor proprietary in the telecoms world, and this is an attempt to standardise an open and interoperable way of doing it, more like how we do in IT.
Don't read this to be a criticism of telecoms as being outdated or backward - they just have very different tolerances of breakage (which IT tolerates as inevitable or necessary or "move fast and break things"), so this is a big step. In many countries they have a 7 or 8x 9's uptime requirement on their main telecoms core network, and similar expectations even in access networks. NFV is network function virtualization, which is the move towards each component of a 5G network being software, rather than dedicated hardware, running in a hypervisor instead of on dedicated bespoke silicon.