Charlie Ashton, Senior Director of Business Development for Networking Solutions at Wind River, discusses NFV.
Eurocomms.com: What types of network functions lend themselves best to virtualisation in your experience?
Charlie Ashton: Many of the early use cases for NFV are virtual CPE applications, for both enterprise and commercial customers.
These include functions such as virtual routers, firewalls, Application Delivery Controllers (ADCs) and WAN acceleration.
The opex reductions are significant and any reliability risks are contained a to specific set of customers.
We also see early NFV deployments in Cloud RAN or vRAN applications. In both these cases, it’s not necessary to have all the high-level orchestration in place that would be required for core functions like virtual EPC.
What are some of the major challenges that operators have to overcome as they test and deploy NFV?
From a business perspective, high service reliability becomes a key issue as service providers look beyond the early trials and make plans for actual deployments.
The number one driver for high reliability is protecting service providers’ top-line revenue.
Service level agreement (SLA) penalties present a significant risk to service providers’ top-line revenue that could negate the ROI benefits of moving to NFV in the first place.
Of course, service providers don’t guarantee carrier grade reliability for all their customers or all their services and as regular consumers we can all confirm that.
But some core networking services are mission critical and many enterprise customers sign SLAs that guarantee them significant refunds if services that they purchase aren’t delivered with five-nines uptime.
NFV presents new challenges for service reliability, such as: new, virtualised network applications; dynamic reallocation of VMs across servers, racks and data centers; complex, hard-to-debug traffic flows; and off-the-shelf servers.
The key to achieving five-nines reliability for services is to deploy them on an infrastructure platform that actually delivers six-nines uptime, which means an average of 30 seconds downtime per server, per year.
What are the key differences between providing services to the residential and enterprise markets?
Residential customers are accustomed to sketchy reliability and they don’t sign SLAs that guarantee five-nines reliability.
Enterprise customers contract for five-nines reliability.
There has been a lot of talk that operators lack the requisite skills sets in their organisations to make a success of NFV. Is this your experience?
The toughest challenge for service providers deploying NFV may well be the skill sets involved in successfully deploying and operating an NFV-based environment.
Staffers currently responsible for network infrastructure are not necessarily proficient in dev/ops or the operation of large, complex clouds.
Despite all the good work done by ETSI and the growing pool of NFV solution vendors, integrating, operating and maintaining these technologies is by no means a trivial task.
It’s critical that service providers maximise the synergy between IT organisations responsible for traditional cloud data centers and network infrastructure teams that understand the carrier networks.
We are still to see any large-scale NFV deployments; do operators need to move more quickly?
We see service providers moving from lab trials to field trials of NFV this year.
We can expect live deployments to start in the second half of this year or early 2016.
Undoubtedly, service providers would like to be able to move more quickly, but they need to be satisfied that complex issues are resolved first, especially around reliability, scalability, operational processes and interoperability with legacy infrastructure.