Oct / Nov 2014 Volume 16 Issue 5
CEO GUIDE
■
TALKING HEADS
Cisco Systems' Phil Harris explains why NFV benefits can be real ■
NFV PREPARATION
TO NFV PART1
Should CSPs start with the easy steps?
NFV isn't just another network technology upgrade
Can the testing overhead be reduced?
T
H
E
G
L
O
B
A
L
V
O
I
C
E
■
F
VIRTUALISED TESTING
O
R
B
/
O
S
S
C O N T E N T S
Introduction Network functions virtualisation (NFV) is an elegant concept that understandably has gained a lot of CSP attention. The idea of decoupling the function of network equipment from the hardware itself to enable one piece of hardware to perform multiple, virtualised functions is of great appeal because it will eliminate sub-optimal use of capacity and remove the need for CSPs to deploy specific equipment to perform specific functions
S
Orchestration will be critical but so will the ability for CSPs, to prepare for and understand the cultural impacts of moving from deploying, managing and operating functions specific hardware to the virtual equipment arena.
o far, so good. However, much of the benefits will only be realised once pure NFV environments are deployed and most commentators foresee this taking between five and ten years. With this in mind this first part of the VanillaPlus Guide to NFV will focus on what CSPs need to do to plan their NFV deployments. The following pages feature interviews and features that explore the challenges associated with transformational NFV deployments.
The scale of this technological shift cannot be underestimated, and we will explore this further in the second part of our Guide to be published in December 2014. In the meantime, we hope you enjoy this first part and the insights it provides into a technology that truly has the potential to reset the cost base of CSPs and enable them with greater service and network flexibility than ever before.
Our contributors assess how these will be overcome and how the industry will manage the hybrid environment that will inevitably exist composed of virtualised and non-virtualised hardware. Part of that impact will be keenly felt in O/BSS which will have to enable interoperability between virtualised and traditional hardware, as well as cope with transition between.
George Malim
IN THIS GUIDE C4
TALKING HEADS Cisco Systems’ Phil Harris explains why the benefits of NFV are real but warns CSPs must focus on business and operations transformation in addition to the technical change if virtualisation opportunities are to be maximised
C7
NFV PREPARATION Nick Booth argues that CSPs should start with the easy steps in order to get underway with their migrations to NFV
C10
EXPERT OPINION The transition to NFV isn’t just another upgrade to a new network technology. It changes how networks are designed, built and managed, writes Ronnie Neil
C14
VIRTUALISED TESTING Steve Jarman advocates that CSPs take a virtualised approach to testing to keep pace with NFV cost savings
TALKING HEADS Phil Harris
C4 NFV PREPARATION
C7 VANILLAPLUS CEO GUIDE TO NFV PART 1 / OCT/NOV 2014
C3
TA L K I N G
H E A D S
Organisation, operation and orchestration provide the foundations for successful NFV roll-out hil Harris is vice president and chief technology officer in the Chief Technology and Architecture Office (CTAO) at Cisco Systems. As communications service providers (CSPs) continue to prepare for network functions virtualisation (NFV) deployment, he tells VanillaPlus the benefits are real. However, CSPs must focus on business and operations transformation in addition to technical change, if they are to maximise on the virtualisation opportunities
P
What benefits do you see CSPs and their customers achieving? Phil Harris: Primarily, the ability to rapidly deploy new and highly differentiated services. As we see the need for differentiation, at the provider level, becoming an even greater driver in the incredibly competitive CSP market and with non-traditional over-the-top and non-incumbent players rapidly entering the market, I think the notion of rapid new technology adoption and new service delivery capabilities are going to be among the key benefits that are derived from network functions virtualisation (NFV) deployment.
The ability to provision capabilities closer to the customer becomes an interesting opportunity with NFV
L
It’s simply because the rate at which we can commission, deploy and activate new services becomes a matter of days – if not less, as opposed to weeks and months as it is today.
C4
VANILLAPLUS CEO GUIDE TO NFV PART 1 / OCT/NOV 2014
We will see the proliferation of not just traditional network functions but a new generation of functions and capabilities we can exploit, that quite frankly either aren’t viable in a pure hardware play or because traditional deployment scenarios have become so complex that their value couldn’t be easily realized by CSPs and/or their customers. I think we’re going to see some other interesting dynamics: As we’ve discussed, NFV promises increased density and richness of capabilities, and with this new types of technology services can be created and innovative business relationships can be contemplated: Companies like Netflix, for example, want to deliver optimised content to their customers, so the ability to provision capabilities closer to the customer becomes an interesting opportunity for traditional infrastructure players to cooperate with over the top players to optimise their respective businesses. What are the challenges you see at the moment in getting to even a partial NFV environment? We’re seeing lots of proofs of concepts at CSPs but how far away are we from widespread rollout and what steps does the industry need to go through to achieve that? PH: It comes down to the three Os – organisation, operation and orchestration. When it comes to organisation we see traditional operators that are relatively siloed in infrastructure based business lines such as fixed line or wireless services needing to, in effect, build internal OTT operational groups to address new service deployment and operational opportunities. To effectively take advantage of NFV, it necessitates an organisational transformation where the underlay and the overlay can work in cooperation as opposed to being ships in the night passing unnoticed. Some CSPs are ahead of the curve in this, some are not, many are just realising that you cannot put a new technology and a new business model into an existing operational model so easily. Those that do realise this are moving quickly but the types of skillsets required are different. Orchestration becomes the key here. We have to orchestrate many of the things that used to happen in the factory before the box was delivered to the CSP or things that are usually done relatively statically. We have to think about not only how we orchestrate the NFV capability, but the NFV environment because, for example, it doesn’t come with a set of RJ45 cable inputs that can simply string virtualised appliances together. We need to create Day 0 operational state, Day 1 configuration and service chains linking services and applications to customers. We have to build this in real-time and the orchestration requirements are fundamental and advanced beyond the abilities of traditional OSS framework.
We’ve got to learn on top of what we knew before, as opposed to dispensing with old expertise. What we need is a hybrid engineer who understands the network but at the same time can programme in Python or model in Yang and can deliver a set of capabilities on top of the network infrastructure they were responsible before.
There is some significant consolidation going on in the infrastructure space and some of the heavier weight telecoms infrastructure providers are struggling
How does Cisco Systems help and what sets you apart from the other providers? PH: There is some significant consolidation going on in the infrastructure space and some of the heavier weight telecoms infrastructure providers are struggling. For example, the consolidation between optical and IP allows us to have a greater set of control points that we can build into our technology. One thing to remember is we have a very broad portfolio from the enterprise right through to the CSP. This enables us to have a great understanding of where the control points are and how to optimise between the virtualised and remaining physical infra plays. When we get to high-speed 40G or 100G interfaces for data centres and core aggregation you’re still going to need a set of physical devices that can handle this scale of transport, but we then need a set of very clean and standard interfaces and methods linking this into the virtual space. We have the products, services and expertise to manage this scale and operational challenges because of the scope of our portfolio of solutions. The key is not about polarising around any one specific technology layer , but about having a range of options. How complex do you see the service chains being in reality and how will CSPs manage them? PH: When the winter storms hit and the kids on the east coast of the US go home for a snow day, the first thing they do is watch TV or hit some massively scalable online game; infrastructure loading becomes asymmetric to the normal patterns. How we handle this comes back to orchestration. We have to recognise the notion of ad hoc infrastructures where we might change the state of an infrastructure based on consumption demand or some other criteria. Service chains could be deployed over alternate infrastructure to balance load or ensure resiliency in the event of, say power shortages or other service affecting events and the complexity could extend to provide more capacity for a given capability
L
How great a challenge is that shift in skillsets for CSPs as they move from a network environment into an IT one?
PH: Every challenge is an opportunity. We have a breed of telecoms or network engineer that has been somewhat layer-specific and within that often hardware specific. We are looking at the IT specialist becoming a more relevant requirement. That’s not to say we don’t have to understand how complex network topologies are built. In fact they become more complex now because they become effectively ad hoc and more variable.
This potentially constant recalibration of the
VANILLAPLUS CEO GUIDE TO NFV PART 1 / OCT/NOV 2014
C5
TA L K I N G H E A D S
It’s true to say that much of the initial impetus for application/operating system level virtualisation in the typical data centre was driven by cost optimisation
infrastructure will be driven more from the consumption side than the supply side and the ability to orchestrate in real-time involves modelling a set of intents or objectives that are late binding in terms of how they get deployed.
others. We’ll see some areas where it’s pure NFV and some areas where we start to decouple some of the control and forwarding plane functions through NFV.
The other side of this is in the past, our management systems have been fairly monolithic between what we provision and configure and what we assure and what we understand from a performance perspective. I think we need to have more complex assurance systems that allow us to look at criteria outside the infrastructure/technology, to make decisions that could be everything from weather patterns to power availability to economic indicators. That’s going to be the critical difference because of the complexity of the service models and their ad hoc nature.
The industry itself, as opposed to just the technology, is at an inflection point because we are seeing how the consumer and the service provider have less of a rigid delineation between who provides what to whom.
PH: It’s true to say that much of the initial impetus for application/operating system level virtualisation in the typical data centre was driven by cost optimisation but when I look at CSPs, most of the decisions are being driven by revenue generation opportunities. I think it’s product managers within CSPs that are starting to drive decision making as opposed to network engineering groups in terms of the types of services they need to offer and even the methods by which they are deployed
The finite demarcation point that exists between the provider and the consumer will soon be a thing of the past because individual services and how we realize the relationship between consumer and provider is no longer physical. That transition to a service oriented infrastructure model is now an opportunity that faces us as a huge opportunity. We can now contemplate how we arrange service relationships between consumption and supply in a much more efficient and expansive way.
How important is it that approaches to NFV are standardised?
While we are rightly focused on the technology transitions that need to happen, there are also a set of business transitions that need to occur quickly. Those business transitions are critical to ensuring the maximum benefits of the technology transition are realized and they really need to come first if the new CSP business models of the near future are to be fully realized.
How long do you see the period of migration to NFV taking?
C6
In general there will be a simultaneous set of actions across all three of those environments.
Is the real business case for NFV the new services and revenue generation rather than the cost saving cases being put forward now?
PH: Standardisation activity is very important and there is a lot of standards body activity going on in the NFV space. We’re heavily embedded in that movement, especially within ETSI and IETF. We’re focused not only on developing standards but also adhering to them. That’s key because we need interoperability across the orchestration components, the NFVs and the supporting virtualised infrastructure and the underlying physical infrastructure. We need a new level of standardisation to achieve that and enable the traditional BSS side of the equation to be integrated into these new environments.
www.ciscosystems.com
One of the things we can start to do is separate the control plane from the forwarding plane for many applications. The forwarding plane is where much of the capex is spent on equipment such as video head end plant or broadband aggregation infrastructure. The control plane, which manages the infrastructure in question, can start to be centralised in network function virtualised appliances. Of course some things will just stay as they are.
PH: I think there will pockets that will be faster than
Deployment cycles will vary among CSPs based on the current state of their infrastructure and their business models. We expect widespread NFV deployment to be achieved during the next five to ten years across the globe. NFV is a reality today and at Cisco we have reorganised our portfolio, our engineering groups, our customer facing organisation and our advanced services teams to help customers rapidly adopt NFV as a key asset in how they evolve their own business and technology areas of focus.
VANILLAPLUS CEO GUIDE TO NFV PART 1 / OCT/NOV 2014
N F V
P R E PA R AT I O N
Can CSPs dance to a faster rhythm when NFV is deployed? Research says practically all CSPs plan to virtualise but they are paralysed by the usual fears of pitching into the unknown. It’s a question of putting your best foot forward and starting with easy steps, writes Nick Booth
T
embark on the exercise too enthusiastically at first, they risk doing themselves an injury. The problem is that something must be done quickly. Business support service (BSS) stacks were built to support voice over circuit switching but, as Martin Morgan, Openet’s marketing vice president points out, the record has been changed to something a lot quicker, so the old BSS body needs to become a lot more supple.
VANILLAPLUS CEO GUIDE TO NFV PART 1 / OCT/NOV 2014
L
he problem for may communications service providers (CSPs) is that they are trying to trying to perform new moves, with an old body, to an upbeat modern tempo dictated by demanding young subscribers. Metaphorically speaking, the CSP tries to throw some shapes but it looks like a dad dancing badly at a wedding. Virtualisation might invigorate them and make them more agile, but caution is necessary. If they
Mauro Macchi: The biggest mistake is thinking that you can virtualise everything
C7
N F V
Gunther Ottendorfer: The culture of interdepartmental finger pointing cannot be supported in a virtualised world
P R E PA R AT I O N
At the same time, Voice over LTE (VoLTE), which is CSPs’ next technical step, not only has a frenetic pace but is complex too. Meanwhile, the poor CSPs are under continuing peer pressure to come with an answer to the over the top (OTT) competition provided by the Facebook generation of content providers, with their agile and fast moving data centers and supple integrated backbones.
applications that require massive computing power because though the x86 processor is improving massively, that progress won’t compete with the pace of growth of data volumes.
Which is why 13 of the world’s top CSPs held their defining meeting, under the auspices of the European telecommunications Standards Institute (ETSI), in October 2012. Though they all agreed that network functions virtualisation (NFV) would be the way forward, the methods for virtualising their BSS are untested.
It’s not just the technology change that’s going to be hard, warns Gunther Ottendorfer, CTO at Telekom Austria. The personnel management will be even more complex.
The concept of building virtualised infrastructures by increments has been honed by the data centre industry, where vendors like Cisco Systems and Nutanix are creating modules in which many of the elements has been pre-configured and integrated. According to research company Heavy Reading, 66% of CSPs plan to run on virtualised BSS by 2016. If so, they will need a strong re-organisation, says Mauro Macchi, Cisco’s business development director.
Jennifer Kyriakakis: It’s important to ask what virtualisation looks like
“It will be hard to migrate from multiple teams siloed in different technologies into a convergent team,” he says, “but service providers will be forced to follow this path as they strive to optimise opex and activate new services quickly.” If they can’t they won’t be able to compete with the OTTs, he says. This calls for streamlining processes and simplifying the complex service provisioning chain. Partial deployment is a sensible plan, says Macchi. A transition programme should be put in place. For example, some service providers are planning separate networks for their more daring and innovative systems. “The biggest mistake is thinking that you can virtualise everything,” Macchi warns. Initially CSPs should choose the services that are naturally easy to virtualise, like managed customer premise equipment (CPE). Avoid trying to virtualise
C8
CSPs must be selective about their initial virtualisation if they want to avoid a net loss, says Macchi. So venture carefully onto the floor and chose an easy number at first.
Currently, NFV applications are the domain of the network team in many CSPs, but the scope of their responsibility needs to change. The IT infrastructure team should take on the mantle of providing the 99.999% availability, because virtualisation is a data concept. “This will require education,” says Ottendorfer. The IT will need to build its telecoms competence and operational experience. Both the IT infrastructure team and NFV application team will need to learn to work together much more closely, as quick troubleshooting will be called for. The culture of interdepartmental finger pointing cannot be supported in a virtualised world, says Ottendorfer. Like all things virtualised, the CSP’s strategy has to be flexible too, says Jennifer Kyriakakis, founder of management system vendor Matrixx Software. Map out a strategy that takes account of issues that arise as real-time technologies are virtualised. Some functions straddle network and IT domains, so virtualising them is not straightforward, says Kyriakakis. “As real-time charging, billing and policy management converge it’s important to ask what virtualisation looks like,” she adds. While NFV offers lots of opportunity, there are architectural and geographical challenges involved in separating the elements out – the control and data planes, the IT and network teams – before they can be orchestrated again more fluently, says Robin Kent, director of European operations at Adax. Let’s hope they all dance to the same tune.
VANILLAPLUS CEO GUIDE TO NFV PART 1 / OCT/NOV 2014
EXPERT OPINION NFV is more than just a network change Network Functions Virtualisation (NFV) completely changes how communication networks are designed, built and managed by pulling the functions necessary to run networks off proprietary hardware and placing them on open-architecture-based servers. Network element functions can then be deployed where required, when required, and scaled on demand, thus greatly improving infrastructure utilisation, significantly improving the network cost curve growth, and enabling faster, more flexible service introduction, writes Ronnie Neil
N
FV liberates communications service providers (CSPs) from many existing network constraints and complexities. Networks become more scalable, more dynamic by allowing changes in real-time, and more cost effective to operate over their lifetime. NFV also makes network management more flexible with the potential to deliver better performance and customer service while providing a more agile environment for the introduction of new services and applications.
The transformation to NFV does create a significant opportunity to cost-effectively and quickly deliver revolutionary services at breakthrough price points
Market drivers and challenges The market for NFV is nascent, although it is projected to surge over the next few years as CSPs look to bolster service offerings and cut costs. Almost all parts of a service provider’s network can benefit from virtualisation including the customer premises – both business and residential, although the mobile core is likely to be one of the first areas where providers will introduce virtualisation. Some of the main market drivers for NFV include: • Optimised equipment and infrastructure costs by consolidating network functions while exploiting economies of scale from the IT industry • Improved scalability by using a single platform for different applications and users • Accelerated time-to-market by shrinking typical innovation cycles for new services and applications • Expanded collaboration resulting from open environments, enabling a wide variety of ecosystems and partnerships
C10
Forcing major OSS transformation Operational support systems (OSS) and their associated processes are at the heart of delivering and maintaining today’s communication networks and services. Most CSPs have – at least – tens of OSS, each performing a specific function in areas such as network inventory, service provisioning, service assurance and customer care. These systems deliver a significant impact on the quality of experience that end-users receive, including service provisioning times, service uptimes and service quality. To perform their function, a large number of these systems interact directly with the network, especially those OSS involved with service provisioning and service assurance. For these systems in particular, NFV will introduce a fundamental disruption to their current mode of operation.
L
To deliver on the benefits of NFV, however, some major challenges will need to be overcome. Networks
of the future will be designed, implemented and managed radically differently than they are today. Network performance, once predictable in a dedicated, hardware-based network appliance, will be affected in many new ways when a virtual network element is installed or reconfigured. Today, physical links connecting network equipment are points of delineation and locations where data can be accessed for monitoring and troubleshooting. In the future, these physical links will be virtual interfaces, connecting functions within software in the same – or different – virtual machines within a physical server. Today’s methods and tools to identify which subscribers are impacted by poor network performance, where and why, will need to be replaced or adapted.
OSS that have traditionally interacted with statically
VANILLAPLUS CEO GUIDE TO NFV PART 1 / OCT/NOV 2014
configured network equipment, for tasks such as configuration, provisioning, performance monitoring, and fault management, will now interact with dynamic, virtual network equipment. In this new NFV world, how and where do OSS connect to the network? How do they cope when the network and their current access points dynamically reconfigure? And, how do they ensure that the NFV processes themselves are not introducing issues that impact the end-user quality of experience?
link into NFV management functions to receive information on how the network has configured itself. Some OSS will also have to link into operational policy systems as network configurations will be adjusted according to specific customer service level agreements, security requirements, and traffic patterns. Open, standards-based application programming interfaces (APIs) will be crucial to the seamless integration of OSS with the NFV management infrastructure.
The transformation to NFV does create a significant opportunity to cost-effectively and quickly deliver revolutionary services at breakthrough price points. However, there are two major obstacles to overcome – the successful introduction of virtualised network infrastructure itself and the evolution of the associated OSS and processes.
Performance-based OSS will be increasingly important in a virtualised environment including playing a vital role in the management of the NFV infrastructure itself. NFV infrastructure enables automatic reconfiguration of a network, but to be a success, the reconfiguration needs to deliver either better customer experience and/or lower operational costs. Real-time performance data will be a key input into the operational policy systems formulating the NFV reconfiguration decisions.
Evolution not revolution The limitations and shortcomings of legacy OSS will be a major barrier to the wide-scale adoption of NFV. But how do these systems need to change? Firstly, as service providers virtualise their networks and adopt cloud-based architectures, it is vital that the OSS support virtualised environments. That is, the OSS software must operate in virtual environments, support emerging virtualisation protocols like OpenStack, and be deployable in a variety of public, hybrid and private cloud environments. To keep pace with the dynamic nature of virtual networks, OSS must be able to adjust on a real-time basis in terms of capacity, resources and configuration. To know how to adjust, the OSS must
For all CSPs, NFV will drive a profound change to their network and business processes. So, while NFV can unlock dramatic new operational efficiencies and service delivery benefits, these gains may be limited unless the associated OSS are appropriately transformed and integrated with the NFV infrastructure. As the transition to NFV will take some time, the impacted OSS will be forced to support heterogeneous environments spanning both virtual and legacy networks. So the OSS changes, while major, will need to be an evolution, not a revolution. In a future article, we will examine in more detail how service management and assurance solutions must transform to meet the needs of an NFV-enabled virtualised environment.
VANILLAPLUS CEO GUIDE TO NFV PART 1 / OCT/NOV 2014
The author, Ronnie Neil, is a strategic marketing manager for Mobile Assurance and Analytics at JDSU
www.jdsu.com
C11
C E O I N T E R V I E W
Software is virtually eating the network Telecoms as an industry has ridden the wave of the electronics hardware revolution for over half a century. The industry is now entering the next phase of its evolution, based on a network composed of photonic transport and flexible network and operations software running on a virtualised, commoditised computing infrastructure – a cloud. The changes will be profound. Vendors such as Metaswitch, with a strong history of software portable to multiple computing environments, will help lead the charge. Here, Dr Mark Mortensen, the practice head for BSS at Analysys Mason, interviews John Lazar, CEO of Metaswitch Networks about the history and the future of the cloud-based, virtualised network
M
ark Mortensen: The last 50 years of networking technology has brought us from a mechanical/electrical infrastructure through electronics, digital, and, in the last decade, packet technology. At the base of the current network is specialised photonic and packet switching hardware for transporting voice and data with packet and legacy digital switching for voice. Above this physical network sits an agile, centralised service layer of service delivery platforms (SDPs) for hosting new services, policy management systems for controlling the services, and IMS systems for billing services and controlling the network. This next-generation network architecture, implemented only in the last decade, has brought tremendous benefit to CSPs. What benefits has this brought us so far?
C12
Mark Mortensen: So why is this unprecedented flexibility important to CSPs? John Lazar: New driving forces require that CSPs move quickly to the next generation architecture to gain business agility. New over-the-top (OTT) players are eating CSP’s high margin messaging and voice services, while offering new services at faster and faster rates. Meanwhile, the rate of new customer growth for CSPs has slowed worldwide, requiring CSPs to compete more vigorously with each other for
L
John Lazar: It has decreased the time to deploy myriad new data and voice services from two or more
years to weeks – or even days. But we find ourselves at the limits of business speed and service agility since the service layer is implemented such that much of the logic is embedded in software that is tightly tied to the hardware on which it runs. New software means new hardware. New hardware to replace outdated hardware – that is expensive to maintain – usually requires new software. This is limiting the rapid evolution of the software within the networks and decreasing the business agility of the industry.
VanillaPlus CEO Supplement
revenue growth. These are requiring that CSPs lower the costs, innovate faster, and scale successful new services quickly. Mark Mortensen: Hardware vendors have traditionally led the charge on new network equipment, but the new technologies that are emerging to meet these new hurdles seem to all be software-based. John Lazar: Indeed they are. The basics come from the enterprise market, where data centres are already in the process of virtualising – adding a virtualisation software layer within and across the computers in the data centres. This layer allows the computing and storage to be managed as a single resource cloud – but chopped up and allocated to whatever application requires it. When applied to telecoms, this breaks the connection of a software application – be it an SDP, IMS, BSS, or OSS – to a particular piece of hardware, also allowing the application and the hardware it runs on to evolve separately, with expected costs savings and a radical increase in speed of innovation. Mark Mortensen: Reductions of up to 30% in total cost of ownership have been widely reported in the enterprise market. With the introduction of a robust way to dynamically connect the applications, whether within, or between, data centres, applications will be able to be run anywhere, anytime, as required. John Lazar: Yes. But the first step is to virtualise the functions. Voice processing, voice signalling, session border control, routing protocols will all be run as applications inside a virtual machine on top of a hypervisor. The further introduction of an overall orchestration function will mean that the applications will be dynamically created, configured, connected, put in service, and eventually decommissioned, as needed, without changing the underlying hardware. This will greatly speed up the process of scaling up the capacity of new, popular services to meet the demand. Mark Mortensen: Today, CSPs are virtualising their data centres and their enterprise IT applications such as payroll, human resources, and financials – turning their data centres into private computing and storage clouds in which the applications run. They are also early in the process of deploying the first of the virtualised network functions into these clouds. A set of network vendors are working with these CSPs, providing security, policy management, VoIP, and IMS software that has been designed to run on general purpose hardware and have been tested to work with the leading hypervisors that provide the virtualisation layer, as well as on virtualised public cloud computing environments. How does Metaswitch plan on helping them in their journey? John Lazar: Metaswitch, has sold virtualised session
border controllers to BT OnePhone and was named as one of the first four vendors in AT&T’s Domain 2.0 programme. But Metaswitch’s experience in providing these virtualised network functions is no overnight success – it has been writing portable, network software for CSPs and TEMs for more than 30 years. It has products that are already running in virtualised environments and strong technical competency – along with a commitment to progressing this technology. Its products have always been built on a portability layer to allow them to run on different hardware – or a virtualisation layer. Hence, CSPs can deploy a number of its key network components in a fully virtualised environment, running on private clouds, but some CSPs are even running them on the Amazon public cloud. These include: • Class 4/5 switch alternatives • Hosted business services • Session border controllers • Multi-layer transport SDN solutions • Core IMS technology.
John Lazar: : Virtualised network functions can provide more flexibility as they can be spun up and spun down as required
As more network functions are virtualised, the company has been adamant that CSPs will benefit from a new wave of software innovation. NFV opens up the promise of accelerated innovation, enabling service providers to engage in a much more flexible, agile way with software vendors, like Metaswitch, or their own software resources, to allow fast prototyping and rapid scaling of new services on top of a much more flexible platform. Mark Mortensen: Analysys Mason has opined that the journey will take at least a decade. There are some very good reasons for certain types of network functions to be paired with built-for-purpose hardware – high-speed core routers, for instance, used in high-volume applications or photonic transport systems. John Lazar: Perhaps, but for many cases, virtualised network functions can provide more flexibility as they can be spun up and spun down as required, running on a common, virtualised computing environment. This decreases the amount of computing hardware required – decreasing both opex and capex. The benefits mean that vendors will create and execute aggressive roadmaps for virtualising their network equipment, where possible and economical. Towards the end of the journey, the compelling benefits of virtualisation will mean a simpler, incredibly flexible and agile communications network with massive computing and storage capability. It will run on a minimal amount of general-purpose computing hardware, with the resources allocated dynamically to support whatever services consumers require at that moment. Services will be able to be created quickly by installing and configuring software only, scaled up to match actual consumer requirements – not forecasts, and decommissioned quickly, without requiring equipment be removed from CSP offices.
VANILLAPLUS CEO GUIDE TO NFV PART 1 / OCT/NOV 2014
www.metaswitch.com
C13
V I R T U A L I S E D
T E S T I N G
Virtualised testing keeps pace with NFV cost savings The growth in IP traffic presents both an opportunity and a challenge for network operators – revenues continue to grow, but are being overtaken by the costs of handling the growth in traffic, writes Steve Jarman
E
ven in a single metropolitan area, communications service providers (CSPs) deploy dozens of routers and appliances with custom ASICs, running millions of lines of code and performing specialised functions. The costs involved in increasing capacity raises the cost per gigabyte, and it is increased still further by operational expenditure for ongoing management of services, power consumption, and the cost of employing the increasingly rare but necessary skills. Given the complexity of today’s networks, the provision of a new service, even a simple one, to meet customer demand can take months. The success of virtualisation methodologies that have provided flexibility in IT environments is encouraging network operators to adopt similar techniques. network functions virtualisation (NFV) has become a hot topic as a potential solution.
The benefits of NFV NFV adapts IT virtualisation technology to provision many of the functions provided by traditional network equipment on industry-standard high-volume servers, switches and storage located in data centres, network nodes or in end-user premises. Instead of deploying separate boxes dedicated to specific network functions, the same functions are run as virtual machines in ordinary high power servers. This offers immediate benefits. Fewer boxes means lower capital expenditure and less power consumption, while software flexibility allows faster roll-out of new services with less risk.
The challenges of testing NFV
C14
What are we testing for? The end user is mainly interested in performance and quality of experience – they want the service they have paid for to achieve its SLAs. They don’t really care whether functionalities are virtualised or running in purpose-built appliances. For CSPs, performance and quality are also important, but there are additional concerns regarding the control plane and data plane scale, and whether, for example, the number of PPPoE sessions, throughput and forwarding rates, number of MPLS tunnels and routes supported are broadly similar between physical and virtual environments. The new, virtualised network may be reducing costs, but it must not in any way deliver worse service than the corresponding physical environment. CSPs and users accustomed to 99.999% availability will expect the same for virtual environments. So node, link and service failures must be detected within milliseconds and corrective action taken promptly without degradation of services. When virtual machines (VMs) are migrated between servers, any loss of packets or services must not break the relevant SLAs. Instantiating or deleting VMs can affect the performance of existing VMs as well as services on the server, so new VMs must be assigned the appropriate number of compute cores and storage without degrading existing services. It is also critically important to test the virtual environment, including the orchestrator and cloudmanagement system. Pre-deployment and turn-up testing of the virtualised network functions mean that they can be rolled out with confidence, but it is also important to monitor
L
The author, Steve Jarman, is business development manager, EMEA, at Spirent Communications.
The benefits are significant and highly attractive, nevertheless replacing thousands of specialised routers and appliances with NFV-based servers presents quite a challenge. For a start, no-one likes to junk costly equipment that still promises years of good service before it becomes obsolete, so migration to NFV will be subject to depreciation of legacy equipment. In addition, staff skilled in traditional network deployments, Command Line Interfaces (CLIs) and element management systems will need to be retrained to work with cloud- management systems.
Above all there is the challenge of the unexpected when making any change to a complex system. However good the virtualisation process looks on paper, what matters is how it works in practice. If the upgraded system fails or performs below expectation, it will result in angry customers, lost revenue and ultimately discourage the adoption of NFV. So everything – virtualised functions, virtual environments and end-to-end services – must be thoroughly validated by testing before deployment.
VANILLAPLUS CEO GUIDE TO NFV PART 1 / OCT/NOV 2014
services and network functions on either an ongoing, passive basis or an asneeded, active basis to make sure that the system can cope with evolving demand and unexpected surges. Monitoring virtual environments is more complex than their physical equivalents, because CSPs need to tap into either an entire service chain or just a subset of that service chain. For active monitoring, a connection between the monitoring endpoints must also be created on an on-demand basis, again without degrading the performance of other functions that are not being monitored in that environment. Network operators are looking for effective ways to scale their services while keeping down costs to maintain revenue growth. NFV has emerged as a promising solution, enabling service providers to deploy large numbers of virtualised network functions on industry-standard high-performance servers. As well as cutting both operational and capital expenditure, CSPs benefit from shortened time to market. Other advantages for the industry include the growth in independent software vendors, and easier development and deployment of new revenuegenerating features. In theory, virtualised network functions perform exactly as their physical counterparts. In practice such complex systems always require thorough testing under simulated but realistic operating conditions to eliminate unexpected phenomena arising out of complexity. So NFV functions and infrastructure must be stringently tested in the lab before turn up of services, in order to minimise network outages in the final network. It is also necessary to monitor performance on an on-going basis to ensure continued good service. With the massive scaling of today’s networks, testing at this level could add considerable cost – although, of course, the long-term cost of not testing might be far greater. The use of virtual test appliances, however, greatly reduces this overhead, both in terms of capital outlay and labour. This approach is recommended for all but the most critical functions, where a certain proportion of physical testing would still be recommended.
VANILLAPLUS CEO GUIDE TO NFV PART 1 / OCT/NOV 2014
C15
Transform the Business Transform the Experience Transform the Architecture
Business Transformation Through Innovation With the Internet of Everything driving massive changes in how you build, operate and leverage your network, you need a trusted source to help drive profitable business outcomes. With Cisco, we have a track record of helping Service Providers grow top line revenue while helping lower operating costs and increasing agility to turn up services faster. We do this by offering a comprehensive set of solutions which leverage the Cisco Evolved Services Platform, the industry’s most comprehensive virtualization and orchestration software platform, which creates, automates and provisions services in real time across compute, storage and network functions.
Learn more about Cisco’s Evolved Services Platform: http://www.cisco.com/go/esp