Vikram Saksena: Software-centric migration is still at the
edge of the network, and it will move into the core over time.
NetScout is focusing on technology to monitor virtual and
non-virtual parts of the network
Application agility, service velocity and network elasticity are the three key factors that are helping to encourage virtualisation of networks as operators look for ways of bringing new services to their customers with lower capital investment.
But network functions virtualisation and software-defined networks will also change the way that networks are thought of from the ground up.
“We won’t always be delivering applications from deep inside the networks,” says Dr Vikram Saksena, the chief solutions architect in NetScout’s office of the chief technology officer. Saksena is the executive responsible for the company’s NFV/SDN strategy.
“Latency is a big factor for operators,” he says. “Operators want to push application delivery closer to the network edge in order to cut the latency and improve user experience.”
The arrival of NFV and SDN means a complete change in how networks are designed, built and operated. There is concern over how operators will monitor service quality within and through virtual environments. NetScout provides service assurance technology to mobile, fixed and cable operators around the globe, and so has an excellent view of the implications of the new technology.
“Customers want to be able to set up services on demand, and on the fly,” says Saksena, who has been at the forefront of innovations in the wireless, internet, cloud and data analytics for many years. That’s why he lists service velocity as one of those requirements. “Customers need to be able to set up VPN connections, bandwidth and quality-of-service profiles on demand. Operators need an environment where they can set up services rapidly in response to customer demands.”
And elasticity? That’s because when a new service is set up, it’s hard to know exactly how much it will be used. “You may have to scale the network up or down in order to match demand.”
But in the conventional way of setting up networks, capital equipment has to be allocated according to the expected requirements. “You can never predict accurately how much a service will be used,” says Saksena. Do you over-provision and risk having equipment that is under-utilised, or under-provision and then have to struggle to meet better-than-expected demand?
“With virtualisation, operators will be able to provision a network based on what users actually want to do,” Saksena explains.
He’s watched the industry evolve over the decades since he started at AT&T. In 1997, he became an AT&T Fellow before being part of two start-up teams. Later he was CTO of Sonus and then Tellabs.
“We’re moving from hardware-centric networks to software-centric networks,” says Saksena. “That’s the starting point. But to actually deliver this complex goal there are a number of key requirements,” he adds.
In order to virtualise a network, “you need to be able to abstract services away from the hardware,” he explains. “That’s a substantial design task and once that is done you need to be able to program them into the network — that’s where software-defined networks come into play.”
Two further capabilities are needed. “First, orchestration is the ability to program a service across the network. You don’t want to have to configure different parts of the network in different ways, which takes time and can be error-prone.”
“This is where we at NetScout come in, once you’ve done all that you need to assure the service so that it meets users’ expectations,” says Saksena.
“We are in the business of making sure these software-defined networks provide service assurance in real time,” he adds.
That means providing operators with the ability to know what is going on as it happens, so that they can check that the key performance indicators they promise to their customers are being met.
“Previously you could collect the data and then analyse it off-line and figure out what’s going on.” That takes time. If virtualisation means that services can be created and launched rapidly, service assurance needs to work at the same sort of time scales.
“Now we’re shrinking the time.,” says Saksena. “Service assurance has to be more real time. That is what NetScout is focused on — we provide intelligent systems that can process KPIs in real time.”
NetScout has 30 years of experience of this field. “We started with enterprise and data centres and monitoring local area networks. We have mastered this art. The focus of the company hasn’t changed, but the landscape of this industry is changing.”
And 30 years of experience perfectly suits NetScout to take on the challenges of virtualisation, he adds. “Overall this transformation is going to take a decade,” he forecasts. He likens the change to the switch — still going on — from traditional TDM networks to IP-based networks.
Why should operators set about this transformation to virtualisation? “Services were getting commoditised,” says Saksena. Services were being baked into the hardware that operators were buying from a limited number of vendors.
“Now, moving to a software-centric approach will give operators the ability to differentiate. Operators can again be in control of the services they provide, and different operators can spin out services for their own customers,” he explains.
There’s a long transition ahead, but where is virtualisation taking hold first? “One area that is getting early traction is the evolved packet core,” he says. “This is where we are finding some of the largest operators moving to NFV. They are using a virtualised structure and putting EPC into the stack.”
But it’s clear that operators are still moving carefully into this new virtualized world until they have enough confidence. “It’s something like a sandbox,” says Saksena. “Operators are starting off by looking at things like the internet of things and connected cars, services for which they can carve out a virtualised part of the network.”
Until their confidence builds up with experience, they are still a touch wary of moving existing mainstream users on to virtualised services. “The sandbox is for applications where the risk is not high. The internet of things is a prime area,” he says.
“We’re seeing virtualisation being introduced mainly for newer, non-mainstream services where operators have the opportunity to work through the issues before scaling the deployment.” They are more comfortable starting off where “risks are lower”.
Saksena highlights another area where operators are looking to virtualization in order to start new services. “Cable operators in the US are creating their own WiFi networks, and they need a system for authentication. All that can be virtualized.”
Small and medium businesses are likely to be early in the line for virtualised services, suggests Saksena — services such as hosted voice for small and medium businesses, plus virtual customer premises equipment, “instead of using a fixed router. You can even do some of the security functions on a virtualized basis in the cloud,” he adds.
“What we at NetScout are focusing on is to create a virtual probe technology that can run on a virtual machine in order to monitor east-west traffic,” he says. “And in addition it has to probe the north-south traffic. We use the same software on both the physical and virtual platforms.”
It is important to provide “exactly the same functionality for end-to-end monitoring and service assurance,” he says. “The complexity of physical and virtual environments have to hidden from the operators so services can be assured to provide a seamless, disruption free evolution.”
NetScout has developed and patented its ASI — Adaptive Session Intelligence — technology so that operators with virtualized networks can provide real-time KPIs almost instantaneously. “Customers need information about packet loss and dropped calls without the time lag, so that they can take action right away — or put remedial action on the priority list,” he says.
But the key challenge during this transformation is that operators will need to manage hybrid networks, where only parts are virtualised.
“You need to make sure that your non-virtualised networks can be managed seamlessly alongside the virtualized networks. Our ASI technology works on virtualised networks as well as on the non-virtualised networks.”
It will take time, he emphasises. “We recognise that operators will be in a long period of transition and that they will need full visibility throughout.” But that’s what NetScout is here to provide.