There’s long been speculation about how Network Function Virtualization (NFV) can improve the networks of communications service providers (CSPs). In a recent Telecom TV conference, 83 percent of participating CSPs agreed that NFV is ‘very important’ for 5G. All five members of the panel, comprised of ranking industry professionals, concluded that without virtualization, there cannot really be 5G. There’s also plenty of optimism surrounding things like network slicing, which could allow CSPs to better customize their networks through NFV, once 5G has arrived. While both propositions may yet bear fruit, what about today? Can NFV benefit CSPs in the here and now?
Expedient Implementation of Services
We at ECT develop and implement value-added services, like cloud unified communications and collaboration (UCC), cloud contact centers, etc. They are provided in the network of the CSP on our Telephony Application Server (TAS) and Media Resource Function (MRF). By implementing the latter as Virtualized Network Functions (VNFs) running in the CSP datacenter, we are reducing the time-to-market of new services drastically. The VNFs are installed and configured using a VNF Manager, such as Nokia’s CloudBand Application Manager; a telecoms engineer at the CSP can do this on his/her own, making CSPs less dependent on their vendors. This is one of the biggest advantages offered by NFV, as it allows CSPs to react more quickly to the needs of their business customers.
This also means that CSPs no longer need to integrate new hardware for bare-metal solutions into their Network Operations Center (NOC), thus eliminating a large amount of attendant hassle. With NFV, integration of the datacenter with the NOC is standardized and performed via a single interface, making everything much easier. Every bare-metal network function, on the other hand, comes with its own operations manual and has to be integrated separately with the NOC, increasing the number of interfaces in use.
Beyond that, as CSPs surely don’t need to be reminded, bare-metal solutions come with a wait-time of between eight and twelve weeks. To say nothing of the delivery, burn-in, and installation times, opting for vendors of such legacy solutions comes with other difficulties.
Something I have learned as ECT’s Chief Network Architect is that nobody wants a human being actually sitting there and doing all the integrations by hand. There’s the huge potential of human error; they could change something unwittingly, causing a snowball effect that can impact the whole network. Time-to-market is drastically reduced when everything is pretested and automated using a VNF Manager. Thanks to the automation, the length of projects in their entirety can be reduced from months, to mere days; with the actual installation possible within minutes. This is great news for CSPs and their business customers.
CAPEX and OPEX
When considering all this in light of the longer term development of telecoms networks, NFV can dramatically reduce operating and capital expenses. When a CSP purchases a bare-metal solution from a vendor, they usually have to purchase new hardware. Yes, for the most part, thanks to hardware standardization, it doesn’t need to be proprietary. However, networks containing bare-metal solutions are often comprised of many different types of hardware. And CSPs have to add more dedicated hardware every single time they expand capacity or add new components. It is a strategy that is obviously very capital intensive and which, with time, exponentially increases the network’s physical complexity.
But purchasing, installing and integrating all that hardware are merely the first steps. It needs to be housed, maintained, powered, and cooled, which requires constant financial upkeep. This is the OPEX ‘kicker’. The more hardware there is, and the broader the range of it, the higher the upkeep becomes from a maintenance perspective.
This is true particularly when it comes to any required repairs.
To repair or expand a bare-metal solution, you generally have to use exactly the same hardware as in the original installation. Depending on how old the system is, that may not be so easy. To ensure guaranteed repair times of critical systems, the CSP often purchases spare systems or parts storing them on or near the site and then pays a maintenance fee to have vendor personnel on standby 24 hours a day, seven days a week, 365 days a year. I would suggest that this is as cumbersome and expensive as it sounds.
With NFV, however, this problem completely disappears. Because the datacenters into which VNFs are installed always have reserve capacity of at least 20-30%, should any technical hardware problems arise, CSPs can run their networks with the extra capacity the second they need to. The faulty hardware can then be replaced at the next opportunity with absolutely no downtime. Hardware manufacturers usually take care of the physical maintenance of the datacenter, not the CSP; and the underlying hardware is generic and purchased in high quantities, so it is cheaper in the first place.
Flexibility for the Future
Lifecycle management is one of the most important aspects of NFV and one of the most useful to CSPs. The VNF Manager covers the complete lifecycle of VNFs from instantiation to configuration; from healing to auto-scaling. It is essentially the brain of the whole operation, analyzing the performance of VNFs in real-time and taking action to prevent problems before they occur. It can automatically heal virtual machines if problems do arise, and the level of automation of VNF Managers is only set to increase. This is really why CSPs can do away with the baggage of bare-metal maintenance: the operation becomes much more preventative rather than reactive.
NFV also makes networks more elastic and agile: the VNF Manager monitors the capacity and scales in and out resources as needed; or even evacuates virtual machines from one server to another. Let’s say a CSP has a televoting system in place for large televised events. If it is not virtualized, then the CSP needs to have set aside sufficient hardware resources ready to be used the few times a month, or year, that the capacity of their televoting service will be pushed to its limit. With NFV that need completely disappears because the Manager knows how to calibrate the virtual resources to match the demand, expanding and contracting as needed.
So, regardless of what it will allow in the future, the answer to the question of what NFV can do for CSPs now, is this: CSPs can streamline the entire lifecycle of new and existing network functions via automation and virtualization, reducing time-to-market of new services and eliminating downtime for maintenance. They can run their networks far more efficiently and intelligently due to the automation provided by VNF Managers, scaling in and out the right amount of resources to the right place, as needed. CAPEX and OPEX are significantly reduced, since CSPs no longer need to regularly and heavily invest in a lot of different hardware or pay to maintain it. These benefits and more are waiting for you to come and take them!
At ECT, we are today realizing all of the above advantages for our customers. Our VNFs are automated, redundant, and ETSI compliant. We are integrated with Nokia’s CBAM VNF Manager and we are also certifying our VNFs with other major VNF Managers.