Posted by Mae Kowalke on Monday, May 25, 2015 with No comments
Just in case you have a few minutes of downtime amid the activities of Memorial Day, here are some reflections on the latest developments in Network Functions Virtualization (NFV). Enjoy.
As providers look for ways to deploy NFV while minimizing risk, they're likely to first consider innovative services with fairly relaxed quality of service (QoS) requirement, like machine-to-machine (M2M) communications, Light Reading predicted.
Of course, that's not the only motivator for NFV adoption, nor the significant involvement telecom carriers and cable providers are taking in the development of NFV, SDxCentral pointed out recently. In fact, it's probably fair to describe NFV as really a networking initiative driven by 'customers' (carriers and providers).
Common NFV goals among carriers and providers, SDxCentral said, include:
- Drive down the cost of network equipment
- Transform network equipment into a true commodity so gear is truly interchangeable
- Increase their agility by being able to turn network functions on and off instantly and automatically
- Ensure that NFV is built to provide carrier-grade reliability, performance, security, and policy management
That last one is a biggie, but not insurmountable. For example, several vendors recently teamed up on an NFV performance benchmarking test targeting the use of NFV for virtualized customer premises equipment (vCPE) at the service provider network edge, SDxCentral reported.
Their findings indicate that low cost, high performance NFV-based solutions at the network edge are possible, allowing network architects to deploy multiple virtualized network functions (VNFs) with little or no adjustment to performance. Good news, indeed.
Now, let's turn the tables to the businesses carriers and providers serve: what do enterprises need, and are they ready for NFV?
It's true that NFV essentially converges IT and network operations, reducing service providers' costs and leading to higher-value services, brought to maker faster, RCR Wireless noted. Selling the resulting value-added services to enterprises can be tricky, though; CIOs and network admins are reluctant about handing more control over to carriers, even if there are clear benefits.
Convincing them to take that leap is possible, though, RCR Wireless said. Compelling reasons include the ability to completely automate service delivery, and the ability to dynamically provision application-aware networks wherever and whenever they're needed.
The upshot of all this is large-scale NFV deployments could be just around the corner (say, next year--2016), SDxCentral reported, citing a report from Dell'Oro group predicting the tipping point for telco NFV is imminent.
Vendors like Alcatel-Lucent, SDxCentral said, are reporting a quickening pace of people moving from proof-of-concept to production, and there's a general sense that telcos think NFV is inevitable, although they'd prefer someone make the leap first.
One big question telcos and vendors need to figure out is the best way to implement NFV and where VNFs should be hosted, noted SDxCentral. Topology options proposed by ETSI include 'Centralized' (VNFs hosted at central location; lowest cost implementation), 'Service Edge' (virtualized resources and VNF hosting pushed to the service edge, creating a more scalable and service rich implementation), and 'Chained' (where services are built by combining distributed resources).
Whatever topology option is chosen, and however quickly NFV deployments get off the ground, migration to NFV is likely to unfold one step at a time, Light Reading predicted. The process starts with one one VNF and the subset of NFV architecture components needed to support it--and grows from there. As the saying goes, a journey of a thousand miles begins with a single step.