Telecom IQ
Showing results for 
Search instead for 
Did you mean: 

NFV: Supporting the Present to Insure the Future

Supporting the Present to Insure the Future


Even in today’s networks, operations costs make up about half of total cost of ownership for infrastructure.  As we add computing elements to networks, both to host network features and to provide cloud computing services, we risk increasing infrastructure complexity and operations costs further.  Uncontrolled opex growth could overwhelm other savings sources and stall NFV progress, so we have to control it.  More than that, we have to reduce it.


NFV adds components to services.  When you replace purpose-built appliances with software and hosting, you’re adding servers, virtual switches, hypervisors, operating systems…and you’re still supporting the familiar access and transport technology that moves traffic.  Opex used to be about a quarter of network total cost of ownership (TCO); by the end of the decade it will be three-quarters of TCO unless we can do something.  Even today, operators say their revenue-per-bit curve and their cost-per-bit curve will cross over in 2016 or 2017, making further infrastructure investment impractical.  More opex would only make it worse.


NFV shows us what that “something” we must do should actually look like, with the most powerful concept to be introduced to networking since packet switching.  Management/Orchestration or “MANO” is a set of automated processes that can deploy virtual network functions (VNFs) based on models and instructions, and establish a framework to link these deployed VNFs with current management and operations processes.  Properly implemented, MANO controls the management complexity of NFV to prevent that opex explosion.  Without it, opex increases would overwhelm capex benefits and we’d never see NFV at all.


The problem is that NFV impacts only a small part of service infrastructure, particularly in the early days of NFV adoption.  How will operators prove out NFV when early projects simply hold their own in operations costs because the scope of NFV is limited?  There’s an easy answer, which is to expand the scope of MANO to cover not only the “cloud” part of the network of the future where VNFs are hosted, but the legacy network.


We know the cloud, from the systems it runs on to the OpenStack and OpenDaylight elements that deploy and connect virtual functions and applications alike.  We also know the network, from our long history as a provider of network technology and our roots in operations and business support systems (OSS/BSS).  When MANO emerged from the ETSI NFV process, we recognized how powerful it would be if it were spread over all of the infrastructure in the network of the future.  We saw that operations could be the unifying element for that future network, and that savings in opex could not only protect the benefits of NFV and even SDN, but also could actually fund early NFV deployment to improve the business case and accelerate realization of benefits.


We also realized that this expansion of MANO was outside the scope of the ETSI NFV Industry Specification Group, likely outside the scope of all of the standards bodies working on aspects of the challenge and opportunity represented by the next-generation network.  We’ve launched an initiative to extend NFV and preserve the benefits of standards through reliance on open-source software and a community of interested partners built around open interfaces.  We called this “OpenNFV” and we think it’s the model for how to move to the next-generation network-cloud we all know must come, not by taking risks but by leveraging benefits.


OpenNFV: Step by Step to the Cloud


NFV can be the catalyst that joins the network to the cloud, joins present services with the future of services.  To make that happen you need three things:

  • A rich set of VNFs that expands as opportunities expand.  These have to be drawn from all possible sources today to create a full repertoire of functions and features to support early sales and deployment.  They then have to be continually expanded to reflect new opportunities in mobile “decision fabric” and IoT services and beyond. 
  • The ability to support the widest set of resources, both to exploit currently deployed infrastructure and to offer a wide choice of platforms to optimize cost/performance in the future.  What should be a part of Network Functions Virtualization Infrastructure (NFVI)?  The simple answer is “everything!”
  • A universal orchestration model that will orchestrate not only deployment but all of the aspects of service lifecycle management and provide integration with OSS/BSS systems both for “new” NFV infrastructure and for legacy.  MANO principles must apply over all VNFs, all NFVI, and all of the legacy elements of the network, and within any operator or across a complete global federation of operator partners.

VNFs are not only the heart of NFV, they’re the heart of the cloud and of HP’s OpenNFV architecture and community.  We believe that like MANO, the notion of abstract or “virtual” functions has to be spread beyond the explicit boundaries of NFV to incorporate not only physical network functions (PNFs) and also cloud applications and components.


In OpenNFV, functions from all these sources are enveloped in a model that describes how they can be deployed, connected to form services, and connected to management/operations processes.  This model lets us incorporate virtually anything that can be run in the cloud as a virtual function, and in fact we can mingle network virtual functions and application components to create the integrated cloud/network services that many operators and users are already contemplating.  Because we impose no on-boarding changes on function resources, we don’t lock operators into a specific NFV platform and we don’t require function providers to make special changes just to support OpenNFV.  We are the most inclusive VNF framework available.


There’s a tremendous value in starting an NFV project with the widest possible set of service features, but we recognize that today’s network and application software wasn’t designed for the cloud.  Cloud computing is already demonstrating that while there are great benefits to be had by simply migrating current applications to the cloud, even greater benefits can be attained by writing applications that take advantage of the cloud’s elasticity in scale, geographic scope, and application composition.  To insure we can offer the full range of cloud benefits to VNFs and other functions, our VNF model also allows for access to management/orchestration features to fully exploit the flexibility of cloud and virtual network infrastructure.


And just what is that infrastructure?  We think people are paying too much attention to the specific things that are, or should be, part of NFVI.  We at HP know the power that virtualization brings, based on our experience with OpenStack and OpenDaylight.  In OpenNFV, we didn’t start our infrastructure planning by trying to list all the kinds of processors or disk storage or network equipment that we’d have to include in NFVI.  We didn’t even restrict our NFVI vision to our own equipment.  We believe that NFV should be completely transparent to infrastructure.  OpenNFV generalizes the ETSI NFV concept of a “Virtual Infrastructure Manager” or VIM to an Infrastructure Manager, following the same principles of “plug-ins” already applied in OpenStack and OpenDaylight.  With the proper IM, OpenNFV will deploy and connect through any hardware the network operator thinks fits their needs.


Of course, as a premier provider of servers and the cloud, we offer our own NFVI solution.  HP’s Integrity, ProLiant, and Cloudline servers make optimum use of top-tier microprocessor chips from Intel to deliver the best hardware platform price/performance and reliability available in the industry across a complete spectrum of hosting demands.  Our OpenNFV community includes partners who augment our hardware platform with five-nines virtual-machine hosting, OpenStack mapping to everything from containers to bare metal for function deployment, and management integration.  Our own work in OpenStack and OpenDaylight means that OpenNFV leverages these projects’ open architecture to embrace a variety of standard and commonly used management and control interfaces.  Operators can be assured that OpenNFV will support whatever infrastructure they find valuable, present or future.


Management and control is the third of our three NFV requirements.  We’ve already noted that the ETSI MANO concept is a revolutionary advance in the journey to software-automated operations if it’s applied in its broadest sense to evolving infrastructure.  The ETSI ISG has focused its attention on the deployment of VNFs, and HP’s OpenNFV broadens that focus to include deployment of cloud applications, the control of legacy and SDN infrastructure, and the “federation” of service components and NFVI across operator boundaries to create global confederations and services.


We’ve already mentioned one part of our MANO-generalization story, the expanded modeling of functions.  OpenNFV uses abstract models of functionality to represent VNFs as the ETSI specifications indicate, but we also use these models to represent physical network elements and collections of elements controlled by a common management interface.  Further, the same models can describe cloud application components, ranging from security elements (firewall, virus scan) that are essentially high-level network features to unified communications and collaboration (UC/UCC) and pure application components for areas like CRM and ERP.  A single set of models can thus describe any service offering, end-to-end.  Since each object in this set of models is decomposed by OpenNFV and assigned to an appropriate Infrastructure Manager (IM) for deployment, we can assign resources and deploy functions across carrier boundaries.


To jump-start this process, HP has provided its own VIM that is flexible with respect to the deployment of both virtual network functions for NFV and application components for cloud computing.  HP’s implementation of MANO, the OpenNFV Director, is derived from our OSS/BSS Activator framework that has built-in capability to control most popular network devices.  Activator’s IM extends MANO-like control from virtual functions out into the rest of the network, covering every service from the access points inward to the cloud and to legacy VPN, VLAN, and tunnel services as well.  Since Activator was built on an OSS/BSS-integrated platform, operations integration is native to OpenNFV, not something that has to be added on as an afterthought.


These three OpenNFV capabilities offer operators two unique opportunities.  First, it allows them to spin an operations-automation web over both their emerging NFV/SDN deployments and their existing infrastructure, to establish a set of common, efficient, management practices that will insulate operations and management personnel from the gradual changes in infrastructure that next-generation networking based on NFV and SDN will bring.  Second, it creates immediate operations benefits that will reduce opex overall, driving down that cost curve shown in Figure 1 and providing additional time to exploit the agility of their new network-cloud to address new business opportunities through new service offerings.


The most critical step for any new network operator technology is the transition from the lab to the field.  The most critical issue for such a technology is managing the risks and “first costs” of a deployment.  OpenNFV builds an operations bridge from the network of the present to the network of the future.  This bridge leads operators through a process of growing service lifecycle agility while preserving operations practices, tools, and skills.  The early successes OpenNFV creates can even help fund subsequent steps on the migration to next-generation infrastructure and services.


NFV is not driving operators to change; changing market conditions demand changes in the infrastructure and practices of the operators.  NFV is providing the best mechanism suggested so far to address those infrastructure changes.  We believe that NFV, applied to its full advantage, can build a new model for infrastructure, a model that operators can transition to at a pace justified by benefits and that operators can then sustain for decades to come.


For a hundred years or more, network operators have created valuable services by leveraging major infrastructure investments.  The principles that have made telecom one of the most successful and long-standing of all industries haven’t changed, only the infrastructure needs have changed.  OpenNFV bridges network infrastructure from the age of the packet to the age of the cloud.


The Power of One


One revolution is enough for any industry, particularly one like telecom whose capital cycles are measured in decades.  OpenNFV combines the revolutions of the cloud, SDN, and NFV into a single revolution, one that climbs to the valuable higher service layers a step at a time, with each step funded by savings obtained at the prior step, and with each step expanding the scope of opportunity.


The future will demand more than connectivity.  That’s already being proven by the success of the OTTs who have leveraged the advances in information delivery to provide new experiences.  Over-the-network experiences have brought major changes in our lives and in the technology industry, but the fusion of the network and the cloud will bring even greater changes.  We have the ability, through this fusion, to make network-delivered information such an integral part of people’s lives that they’ll think of their technology aids as almost-human in its ability to support their needs.


Those OTTs proved something else in the Internet revolution, and that is that revolutions are too big for a single party to control.  OpenNFV is an ecosystem, built on a fully capable platform and extended through open-source, open interfaces, and open partnerships.  HP has provided the tools to take those critical early steps toward the future, and the OpenNFV ecosystem will expand to include the things we’ll find are essential as we embark on that journey.


The new network infrastructure now evolving is a new ecosystem, a new partnership.  Despite the widely heard statements to the contrary, it’s not a “new business model” for the operators, one that consigns them to forever providing wholesale elements to someone else’s services, further submerged and disintermediated with every passing year.  Because network infrastructure will increasingly include software and servers, it will have the inherent agility to exploit what it has done all along, and to do more for its service users in the future.  NFV will be a big part, perhaps even the biggest part, of the transformation of infrastructure.


NFV is a journey, and as critical as it is it’s not the destination.  What we are transitioning to is not NFV, it’s the cloud.  Agile virtual hosting is the logical extension to agile virtual connectivity, the logical way to exploit declining cost of connectivity.  The cloud, combined with traditional connectivity, will build new processing, decision-making, and content-serving relationships with users, and it is these relationships that will add a trillion dollars per year to service revenue opportunities.


The cloud is the “senior revolution” of our time; SDN and NFV are means of exploiting cloud technology to improve network cost and agility.  We at HP believe that planning for the future of the network is planning for the unification of networking and the cloud.  NFV and SDN are the tools to bring about that unification; paths to that cloud goal.  By unifying NFV, SDN, and cloud planning, deployment, and management around the common framework of NFV’s revolutionary MANO concept, we believe OpenNFV proves that one revolution is all operators will need.

0 Kudos
About the Author


Online Expert Days - 2020
Visit this forum and get the schedules for online Expert Days where you can talk to HPE product experts, R&D and support team members and get answers...
Read more
HPE at 2020 Technology Events
Learn about the technology events where Hewlett Packard Enterprise will have a presence in 2020
Read more
View all