The momentum around Network Functions Virtualization or NFV is building. While still in its early stages, it promises to be a disruptive influence to the way service providers build networks and deliver services to their customers. It’s also going to shake up how traditional network equipment manufacturers do business in the future. IHS Infonetics projects the NFV market to grow to over $11B in 2019, and that $4 out of every $5 spent will be on software, not hardware.
Functionality and services will reside in Virtual Network Functions (VNFs) in this new telco communications cloud environment. But will costs come down when infrastructure functionality is taken off purpose-built hardware and moved to a common resource pool of virtualized compute, storage and networking? Will more hardware be required to support the processing of real-time applications in addition to the virtualization environment? Let’s face it, today’s COTS server is in the position of being a “jack of all trades”, but does that also mean that it is a “master of none”? Granted, overhead introduced by virtualization is coming down and getting close to native performance levels, but for real-time, rich media data plane processing, I/O and CPU speeds need to be fast. As we move towards 5G where envisioned application latencies are in the sub-millisecond range, processing speed will be paramount.
Moving applications to NFV is not as simple as taking existing software and dropping it on top of a virtual machine. This approach does not optimize application efficiency. Nor does it take full advantage of the benefits of a cloud-based approach, and service providers, that are already seeing margin erosion, run the risk of an unprofitable move towards an NFV environment. In a recent white paper, Jim Metzler outlined a few guiding principles service providers should consider when it comes to virtualizing applications. I’ve highlighted a few of them below:
- VNF automation, scalability and programmability are not “nice to have” concepts when moving to NFV, but should be “must have” goals to maximize profitability. VNFs should provide sufficient intelligence to assist in the process of automating the rapid scaling out and scaling up of capacity in response to traffic loads in an NFV environment.
- Modularity of software is critical to optimize VNF application performance and scalability and realize the full potential of a virtualized environment. Modular VNF components should be able to scale independently of each other so that scaling up or out does not require a new instance of the overall application. This approach saves time and resources when it comes to adding more capacity, and it also lends itself better to moving application functionality around geographically to reduce service delay.
- Virtualization technology is quickly evolving, and VNFs should be architected for flexibility to allow operators to accommodate advancements at the virtualization layer such as moving from virtual machines to container technology. Containers are lightweight compared to virtual machines and do not require the entire guest OS to get up and running. This results in more instances per server and faster implementation time; aspects that can become a factor in being able to respond better to bursty changes in demand.
- NFV and SDN are inextricably linked, and the full set of NFV benefits comes from integrating functionality in the application layer that can impact what’s occurring in the packet forwarding plane. The ETSI NFV ISG expressed their belief that NFV and SDN are highly complementary efforts and will increasingly be intertwined into a broad, unified, software-based networking paradigm.
You can download the entire white paper - “NFV Applications: Key Considerations for Profitability” - for free by clicking on the button below. What other requirements are there for profitable NFV deployment? Let us know by tweeting them with the hash tag #NFVprofitability to @Dialogic.