Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

HP Network Function Virtualization e-zine

Hewlett-Packard HP NFV Network Function Virtualization E-zine

  • Login to see the comments

  • Be the first to like this

HP Network Function Virtualization e-zine

  1. 1. 1Communications Edition Industry EdgeNetwork Functions Virtualization Edition Feature stories The need for speed: CSPs look to NFV to innovate faster Freedom of application choice—HP’s OpenNFV Monetizing the new wave of technical innovation: the business case for NFV CIO and CTO NFV priorities HP • Issue 021 • 2014
  2. 2. The new business of the network Few business enterprises have the reach of communication service providers (CSPs). Whether via the Internet or mobile phones, we are increasingly delivering market information, entertainment, financial services, and health services to remote areas, helping to change people’s lives in unprecedented ways. But CSPs face a looming economic challenge. They must continue meeting customer needs while fending off the emergence of new competitors, especially over-the-top (OTT) providers that can innovate quickly and experiment with new offers with less risk than CSPs tied to traditional infrastructure. It all starts with the network. Network functions virtualization (NFV) enables CSPs to introduce new, compelling services quickly and cost- effectively. NFV radically reduces both the costs and the risks of introducing new services by eliminating the need for specialized hardware and building on the standard building blocks used in enterprise IT. And NFV offers the freedom and agility to respond to market demand faster and trial new ideas with limited risk and capital investment. In this Network Functions Virtualization edition of Communications Industry Edge, we share highlights on: • How NFV reduces risk, giving you the ability to trial new services in the market • Making the business case for NFV • HP OpenNFV: freedom of application choice • Primary research on CSP CIO and CTO decision making • New NFV standards established by global standards bodies • How to manage NFV infrastructure • The exciting new role of microservers HP wants to work with you to create new levels of consumer experience, engagement and innovation in this fast-paced mobile world. We’ll help you open new markets, capitalize on emerging ecosystems and architect the future while anticipating and enabling the experiences your increasingly demanding customers expect. For more information, please visit Werner Schaefer Vice President Network Functions Virtualization Hewlett-Packard
  3. 3. In this issue 4 The need for speed: CSPs look to NFV to innovate faster 6 Freedom of application choice—HP’s OpenNFV 10 Monetizing the new wave of technical innovation: the business case for NFV 14 The reality of cost reduction 18 CIO and CTO NFV priorities 24 Carrier-driven standards bodies pave the way for NFV 28 Intel is enabling communications infrastructure to implement NFV 32 NFV orchestration: efficiently manage the NFV infrastructure 36 Microservers, an essential addition to your NFV toolbox 40 SDN and NFV together: Q&A with Dave Larson
  4. 4. 4 Communications Edition The need for speed: CSPs look to NFV to innovate faster Every day, the average U.S.- based network provider moves 19 petabytes of data through its network, an enormous increase from previous years. (For some perspective, James Cameron and Industrial Light and Magic only had to move 1 petabyte of data to render the movie Avatar, and they had three years to do it.) The unfortunate part of this story is that all this increase in data traffic isn’t translating to more revenue for the communication service providers (CSPs). Instead, they are losing ground between two important trends: the rise of over-the-top (OTT) social messaging applications that cost CSPs in revenue, and a rise in capital expenses (CapEx). In a January 2014 survey of service provider CIOs and CTOs, 68% of respondents said that the convergence of the IT datacenter and service network is the only real driver for CapEx reduction going forward. Finding the same agility that OTT providers use to seek new revenue streams Today it is almost impossible for carriers to look for new revenue streams by experimenting in the market. Network functions virtualization (NFV) promises to help CSPs radically reduce infrastructure capital and operational costs— and provide a more agile environment to introduce new services more quickly at far less cost than previous network infrastructure. While at TM Forum’s Digital Disruption Conference in San Jose last October, I monitored a few NFV panels, where participants were asked about the biggest motivators driving NFV transformation initiatives. Was it CapEx reduction? Or a faster path to revenue? The two panelists representing Sprint and CenturyLink definitely felt that it was more the latter. Buzz stirred throughout the room. In another session, when a “show of hands” was requested, half dsdsd
  5. 5. 5Communications Edition of the carriers said that CapEx savings was the primary goal, while the other half indicated it was about a faster path to revenue. In the survey of service provider CIOs and CTOs, 78% of respondents said that driving new revenue is a key business driver for the adoption of NFV while 68% that said reducing CapEx is a top priority. As we go through this massive change, I hope and believe the final answer is that NFV will finally loosen some of the restrictions that hamper the CSPs’ ability to trial and act with agility—the way their formidable OTT competition does. You cannot innovate without taking risks. NFV helps you reduce the risks of innovation by moving to an environment where you can set up a network in a minute instead of a year and roll out new services in almost immediate response to market demand. Interoperability, partnership with new, emerging partners and an open ecosystem will prove to be key ingredients to ensure success as they radically reduce the cost of trying out new services in the market. For the trials that work well, scale up rapidly. For the things that do not work well, retire the service knowing that you have contained the cost. We expect that the speed and agility of NVF will enable more innovation, which means that CSPs can stop being disrupted and once again become the disruptors in the industry. For more information, please visit Julia Ochinero WW Communications Industry Marketing and NFV Marketing Solution Lead Hewlett-Packard
  6. 6. 6 Communications Edition Freedom of application choice—HP’s OpenNFV At HP, we are combining our proven network functions virtualization (NFV)-ready cloud platform capabilities, experience in long-term technology partnerships, and deep domain expertise in telecommunications to help our carrier customers achieve their business objectives with NFV. HP OpenNFV is not just a technology architecture; it is an open and proven approach to NFV that incorporates industry intellectual property (IP), services, a partner program for independent software vendors (ISVs), and relationships with network equipment providers (NEPs) and other applications developers. In addition, the OpenNFV Labs ensure the onboarding of partners for validation of the interoperability of complex multi-vendor applications. Taken together, HP ensures that communication service providers (CSPs) are not alone as they make the transition to NFV, and also supports them every step of the way in a proven, open environment. Read on to see how each of these elements contributes to an open, safe, and flexible option for deploying NFV. Freedom of application choice CSPs want an NFV platform that will allow them the freedom to choose the applications they want and need regardless of vendor. In a recent survey of service provider CIOs and CTOs, 85% of respondents see portability/interoperability as a key challenge with respect to NFV. CSPs need speed and agility, which means their network environment must be open and flexible. And they need to be able to collaborate on technology with third parties for rapid delivery of proof of concept (PoC) projects. To support this degree of flexibility and openness, adhering to standards is critical. Through its leadership in standards, reference architecture, and a partner program that embraces newer, smaller ISVs, as well as traditional players, HP is offering CSPs “freedom of application choice.” This helps CSPs transform in a way that best suits their business and IT. HP OpenNFV Labs In HP OpenNFV Labs, we test applications from partners to validate the interoperability of complex multi-vendor applications, accelerating innovation for CSPs. We provide a testing environment where we test a set of
  7. 7. 7Communications Edition critical applications simultaneously so they are ready to be deployed for rapid delivery of CSP proof of concept projects. In addition, HP provides a catalog of proof of concept projects that have already been completed and are ready for deployment. This helps carriers rapidly test the feasibility of new NFV applications to dramatically reduce deployment time and risk, accelerating applications innovation. Proven, NFV-ready cloud platform Consumers use approximately 60% more bandwidth every year (due to video and other apps), but they don’t pay any more for it. CSPs need to continually increase the capacity of their networks without gaining any additional revenue to pay for it. Carriers are aware that on the IT side of the house, the price of servers drops and capacity increases every 18 months, thanks to Moore’s Law. On the network side, hardware prices do come down over time, but not as rapidly as they do for commercial-off-the-shelf (COTS) enterprise IT infrastructure. Thus, the network side of the house would prefer to use the same equipment as the IT group to reduce costs and increase buying power. HP provides a layered architecture to make this possible, including: • HP Converged System: We take a systems- based approach to converged infrastructure. We provide preintegrated, workload- optimized systems to run virtualized functions on the carrier network, based upon the latest technologies in easy-to-buy, manage, and support models. We support various deployments of hypervisors with SRVIO/DPDK, providing ISVs and CSPs a way to achieve line rate networking performance on HP servers. This allows carriers to run their networks with the same hardware that IT uses. • HP CloudSystem: HP delivers the most complete, integrated and open solution for building and managing NFV cloud platforms. It helps automate service lifecycle management—from initial provisioning, through ongoing scaling, updating and monitoring, to final service retirement— for any cloud service, from infrastructure to application. • HP NFV Director: This is the HP implementation of the NFV orchestration functionality in the emerging model from the ETSI standards organization. This orchestration function is critical to NFV
  8. 8. 8 Communications Edition operation. A key differentiator for HP is ensuring the capability to manage NFV in heterogeneous, multi-vendor, and geographically distributed environments. HP brings our deep experience and expertise in both telecom and IT to ensure that HP NFV Director addresses all the key functionality required by CSPs. Long-term reliable partner HP supports structured partner relationships with key industry players to reduce risk and maximize innovation in NFV deployments, creating a partner ecosystem that includes original equipment manufacturers (OEMs), ISVs, NEPs, and CSPs. Combining partners’ technology, applications, and expertise with HP infrastructure and services, we can provide a complete NFV solution. With our partner program, CSPs are not going it alone. As we continue to work with standards bodies, ISVs, and trusted equipment providers, we make it possible for carriers to deploy new services quickly and safely to gain competitive advantage. And our partner program goes way beyond a simple application programming interface (API) or software development kit (SDK). We also provide comprehensive services to help CSPs deploy new technology from throughout the ecosystem. Carriers and partners can rely on our three-year NFV technology roadmap with the confidence that HP understands what it takes to deploy, manage, and support solutions through the typically long network lifecycle. In addition, HP services provide a safe way for CSPs to implement NFV and get up to speed quickly, including: • Integration and support services • CMS consulting services • HP financial services Deep domain expertise in telecommunications and industry IP With the broadest and most widely deployed portfolio of telecom-certified servers, HP understands what it takes to bring IT technology into telecom networks. HP provides NFV-specific IP at both the platform and application levels, including platform support for packet
  9. 9. 9Communications Edition processing, applications from HP Networking, HP CMS, and ISV and NEP partners. With dedicated software and services for telco, HP uniquely bridges the understanding of IT requirements with network needs. With more than 5,000 experts in the communications industry deployed around the world, we are well suited to support CSPs in this NFV transition. Built on a strong foundation HP is uniquely positioned to lead in NFV. Not only does HP lead in all the critical technologies required to deploy NFV, but it also brings the deep domain expertise in telco that customers need and expect. HP is ready to help you on your NFV journey—creating a path to agility, innovation and efficiency. To learn more about OpenNFV from HP, visit HP in the communications industry • 500 HP OSS customer deployments worldwide • 170 customers using HP Real-time Mediation solutions with over 2 billion transactions per day • 400M subscribers across 38 service providers depend on HP HLR/HSS and its 6 9s of availability • 150 Media Server deployments with 600K VoiceXML ports deployed in more than 100 countries • 90 carriers worldwide depend on HP CentralView to protect revenue streams and customers Werner Schaefer Vice President Network Functions Virtualization Hewlett-Packard
  10. 10. 11Communications Edition Monetizing the new wave of technical innovation: the business case for NFV After a decade of amazing growth and innovation that’s changed billions of lives, communications service providers (CSPs) face an impending economic challenge. They must continue meeting customer needs while fending off the emergence of new competitors, especially over- the-top (OTT) providers like Skype, Google®, and Facebook. OTTs can innovate quickly and experiment with new offers with less risk than CSPs. Also, CSPs need to address the proliferation of mobile devices that generate big amounts of data on the network. The fast growth of ubiquitous mobile access places new demands on infrastructure and calls for greater agility in network sizing. CSPs need a flexible and cost- effective way to respond. The root challenge for CSPs is that much of their network architecture is built with vertically integrated hardware and software. To add capacity or support new service offerings, network elements must be provisioned and updated as customized appliances. That’s inflexible, slow, and expensive. In the face of mounting pressure from diminishing margins, network functions virtualization (NFV) is a game changer for CSPs. Now, providers can introduce new, compelling services quickly and cost-effectively. NFV radically reduces both the costs and the risks of introducing new services by eliminating the need for specialized hardware and building on the standard building blocks used in enterprise IT. In a survey of network and IT leaders from service provider organizations, 79% of respondents said their NFV plans will allow them to launch and tear down new services at a rate that enables them to improve their competitive position vs. OTT providers.
  11. 11. 12 Communications Edition Virtualize the network Enterprise IT shops have long enjoyed the benefits of virtualizing their servers, storage devices, and other IT infrastructure. Virtualization enables the enterprise to run more efficiently by ramping up capacity only when demand warrants it. IT groups no longer have to build capacity in advance through heavy, stair-step capital expenditures (CapEx) to be ready for spikes in demand. Instead they can treat capacity increases as operational expenses (OpEx) and map them more closely to demand. And they can pick the vendor that provides the best cost-performance advantages. CSPs are beginning to benefit from the same approach to upgrading their networks with NFV. This type of architecture enables them to increase revenue by: • Deploying software rapidly to deliver new services • Gaining agility via the ability to rapidly scale applications and bandwidth • Opening a new independent software vendor (ISV) ecosystem to provide differentiating services on open platforms. Implement quickly without disrupting service NFV is based on open standards. That means CSPs don’t have to purchase new and proprietary infrastructure from a single vendor when they need to increase network capacity or introduce new services. Rather, they can select from a variety of vendors to find the right solution that fits their needs at the best cost. Then they can implement it quickly without disrupting service. The benefits of using custom off the shelf (COTS) network elements can be dramatic. Through proof of concept (PoC) projects with leading CSPs, HP has found that using COTS provides two big cost levers: • Hardware—COTS can reduce the capital intensity of network appliances by 30 to 35%. • Software—Moving to open source software can save another 30 to 35%. What CSPs are doing now with NFV HP is a driving force behind the NFV initiative. We’re working with the industry to adopt a more virtualized approach utilizing storage, networking, software, and high-volume, industry-standard servers for telecom appliances virtualization in the near term. Example new service: network-based content filtering A real use case for network-based content filtering from a current HP PoC provides a good example of the benefits of NFV. A common means of keeping adult content from entering the home over the telecom network is to block the URL at the source. But user-generated content is taking off, and much of it is not suitable for family consumption.
  12. 12. 13Communications Edition Trying to block it within the CSP network with the old architecture would involve several discrete telco elements: deep packet inspection, proxy gateway, serving gateway, app server, and perhaps others. Even though consumer demand is probably quite strong for this type of service, trying to implement it with expensive proprietary network appliances at each point of presence would be prohibitive for the CSP. But with the NFV/SDN approach, content filtering can run as a virtual service on top of a virtualized infrastructure to deliver a new service features and strong return on investment to the provider. Pick your element and get started today NFV represents a significant transformation, but HP believes that NFV will give CSPs a more agile environment in which to introduce new services. To make the transformation successful, it’s important to plan for virtualization early in the capital refresh cycle, and HP can help. We combine our strengths as the world’s largest IT vendor with our network and communications expertise. HP is naturally positioned to apply our leadership, experience, and scale to successfully deliver NFV to our global CSP clients. To learn more about the NFV solution from HP, visit Jeff Edlund Chief Technology Officer Communications, Media and Entertainment HP Enterprise Services HP proof of concept projects Proof of concept (PoC) projects help HP identify key use cases and solution designs that help create the key NFV requirements from which the HP NFV architecture is derived. HP’s direct and indirect participation in a number of PoCs has validated the HP NFV architecture. The topics we’ve studied in detail include: • Broadband remote access servers (B-RAS) • Evolved packet core (EPC) • Content delivery network (CDN) • IPsec termination • IP multimedia subsystem (IMS) • Next-generation IN (NGIN) • Multimedia services environment • Customer premise equipment (CPE)
  13. 13. 15Communications Edition The reality of cost reduction As with any new technology introduction, the decision to deploy network functions virtualization (NFV) by communication service providers (CSPs) will be guided by a realistic business case and not just marketing promises. Fortunately, cost savings can be estimated. And these savings open a new world of opportunity for providers to drive innovation. A framework for decision-making Several things have to be considered when deploying NFV, including accrued cost savings, business models, and architectural options. Cost must be considered over a full five- year lifecycle that would include both capital expense (CapEx) as well as operating expense (OpEx). CapEx (hardware and software licenses) reductions may become apparent immediately, while OpEx (installation, support, and power) will improve over time. Of course, there are additional savings from not having to add new discrete network elements when you want to introduce new services. There may be up to a 25% savings from avoiding training, configuration, and shipping costs. Further, NFV helps CSPs avoid the traditional stair-step CapEx needed to provision capacity in advance of expected demand. Instead, it follows a cloud utility-based business model where capacity is easily and cost-effectively added when you need it. An NFV deployment decision is not strictly related to a product replacement strategy but covers an architectural approach. It may be beneficial to deploy software functionality either on customer premises or consolidate it at network edge. vCPE reduces costs up to 24% The cost benefits of NFV can be illustrated with a virtual customer premises equipment (vCPE) use case, which includes set-top boxes and ADSL routers. Assumptions regarding the costs of virtual and physical appliances, servers, labor costs, and power consumption have been made.
  14. 14. 16 Communications Edition Virtual Routing Cost Model Comparison Options 2 2a 2b 3 3a 3b Branch Size Small Small Small Large Large Large Router type H/W H/W S/W H/W H/W S/W Branch services WAN Accel, Caching, SBC WAN Accel, Caching, SBC WAN Accel, Caching, SBC WAN Accel, Caching, SBC WAN Accel, Caching, SBC WAN Accel, Caching, SBC HW Required MSR930 3 Appliances MSR930 Server Server MSR3012 3 Appliances MSR3012 OAPv2 Server SW Required Included Services vApps VSR1001 + Services vApps Included Services vApps VSR1008 + Services vApps HW Cost $11,100 $2,100 $1,500 $21,500 $6,000 $2,500 SW Cost $0 $7,500 $8,000 $0 $15,000 $16,250 Installation/ config time (hrs) 14 4.75 3 14 2.75 3 Installation/ config costs $1,400 $475 $300 $1,400 $275 $300 Support Costs (3 yr) $4,995 $4,320 $4,275 $9,675 $9,450 $8,438 Power costs (3 yr) $1,440 $720 $360 $1,440 $360 $360 Total Costs $18,935 $15,115 $14,435 $34,015 $31,085 $27,848 As the above cost model shows, a small NFV deployment can reduce costs from $34,015 to $27,828, or about 18%. Even with increased software costs, the reduction in hardware, installation, configuration, and power costs more than compensate. The results for a larger NFV deployment are even more dramatic. Total costs can be reduced from $18,935 to $14,435, or 24%. And in both cases, there will be additional benefits over time from new revenue opportunities that are not considered in this analysis. Proof of concept: save $66.4M per year As illustrated, the vCPE solution benefits are enormous both in cost savings and in the speed with which new, innovative services can be introduced. HP is working with CSP clients to prove NFV concepts in actual trials. For example, “Carrier X,” a large conglomerate, has identified NFV as one of its five strategic programs, with an approved plan to deploy up Typ small virtual appliance – $2,500 Typ small H/W appliance – $3,500 Typ large virtual appliance – $5,000 Typ large H/W appliance – $6,500 Small Server – $1,500 Larger Server – $2,500 Basic HW install hours – 2 Local SW config hours – 2 Remote SW config hours – 0.25 Installation labor/hr – $100 Support cost/year – 15% Power consumption/box (kW) – 0.3 Hours of use/year – 4000 Power cost/kWh – $0.10
  15. 15. 17Communications Edition to five million vCPEs worldwide. By virtualizing most of the network functions on residential broadband CPEs and set-top boxes, the company conservatively estimates $66.4M in savings per year. A foundation for a new era of innovation NFV solutions from HP mark the beginning of a new method of delivering network capabilities that exponentially change the economics of CSPs. This helps create a new foundation for a level of massive industry innovation never before experienced. To learn more about how the broad portfolio of NFV solutions from HP can help you reduce capital costs, visit Ghassan Abdo Senior Director Communication Service Providers HP Networking
  16. 16. 19Communications Edition CIO and CTO NFV priorities There is clear evidence that NFV is going to have a major impact on the Communications Service Provider (CSP) community in the short to medium term. It is being driven by a need to increase revenues and meet the OTT competition head on and will demand a change in the role and KPIs of both the CIO and CTO. The following are findings from a January 2014 survey by Coleman Parkes Research on behalf of HP. Research is based on phone interviews of 50 CIOs and 50 CTOs1 from Tier 0, 1, and 2 CSPs globally. 25 interviews were undertaken in each of four regions globally: Europe, Middle East, and Africa (EMEA), North America (NAM), Asia Pacific (APAC) and Central America/Latin America (CALA). NFV tops the list of important trends When asked to identify important trends, NFV is foremost on executives’ minds. 80% of respondents (including 84% of CIOs and 76% of CTOs) said that the move to NFV is a top trend impacting their role. Additional top trends include advances in mobility technologies (69%) and the move to virtualization and cloud computing technologies (66%). When asked to identify the single most important trend, 36% chose the move to NFV, more than any other trend listed. (See figure 1.) NFV is not only top of mind for these executives, they see it having a major impact in the near term. 93% of respondents say that NFV will be a major player in the CSP market within 3 years, with many seeing its rise within 1-2 years (36%) or even within 12 months (8%). (See figure 2.) 1 We use CIO as shorthand for all titles that indicate “head of IT” and CTO as shorthand for all titles that indicate “head of network operations.”
  17. 17. 20 Communications Edition Driving revenue is primary NFV business driver In the “make more vs. spend less” debate, which one is driving CSPs to adopt NFV? In our survey, 78% of respondents said that driving new revenue is a key business driver; 75% said competitive pressure from OTT players; and 68% that said reducing CapEx. (See figure 3.) When asked to choose the single most important business driver, 21% of respondents chose driving new revenue, more than any other driver. But CIOs and CTOs split their responses in second place. CIOs were more concerned with reducing CapEx (18%) while CTOs were more concerned with increased agility and flexibility to scale up or down services to address changing customer and line of business demands (18%). (See figure 4.) Flexibility and agility are top priorities for NFV implementation as well. 79% of respondents said their NFV plans will allow them to launch and tear down new services at a rate that enables them to improve their competitive position vs. OTT providers. And when asked to choose a single priority, 37% of respondents (more respondents than any other) said that Competitive pressures from OTT players NFV key business drivers – fig. 3 Increased agility and flexibility to scale services to address changing customer and LOB demands
  18. 18. 21Communications Edition their primary priority for the implementation of NFV is to enable the deployment of services that previously would not have been cost effective to do so. (See figure 5.) But cost reduction plays an important role as well. 68% of respondents said that the convergence of the IT data center and service network is the only real driver for CapEx reduction going forward. 24% – For the deployment of new capabilities within the network that are not currently in service 13% – For technologies on customer premises (enterprise and consumer) 37% – To enable the deployment of services that previously would not have been cost-effective to do so Priority area for NFV implementation – fig. 5 Increased agility and flexibility to scale services to address changing customer and LOB demands Single most important NFV business driver – fig. 4
  19. 19. 22 Communications Edition Openness and portability are top technical challenges There are many technical challenges ahead, but with concerns about integrating applications and virtual appliances from multiple vendors, portability and openness top the list. When asked to identify the single most important technical challenge with respect to NFV, 26% of respondents chose portability/interoperability, more than any other technical challenge. (See figure 6.) In addition, 83% of respondents said that integrating multiple virtual appliances from different vendors is a core challenge. And 75%
  20. 20. 23Communications Edition Most important technical challenge – fig. 6 Portability/interoperability 26% 18% Ensuring performance levels are maintained or managed to acceptable levels 13% Achieving high performance virtualized network appliances that are portable between different hardware vendors and with different hypervisors 11% Working in a hybrid network environment composed of classical physical network appliances and virtual network appliances said that the move to NFV will drive the use of a new set of vendors for the company. That’s why we’ve created HP OpenNFV to give CSPs the freedom to choose the applications they want and need—regardless of vendor. We have focused on portability and openness, combining a technology architecture based on standards, industry intellectual property (IP), services, a partner program, and the HP OpenNFV Labs to ensure that applications from various independent software vendors (ISVs) run on the reference architecture. HP ensures that CSPs are not alone on their NFV journey and supports them every step of the way in a proven, open environment. For more information, visit Julia Ochinero WW Communications Industry Marketing and NFV Marketing Solution Lead Hewlett-Packard
  21. 21. 24 Communications Edition Carrier-driven standards bodies pave the way for NFV The role played by global standards bodies has been essential to the development of network functions virtualization (NFV). They help ensure that common standards and specifications are introduced that fully meet the needs of communication service providers (CSPs) and that keep NFV an open architecture. That’s why HP has taken an early and leading role in key organizations. The work here is critical—not only helping carriers define the architecture, but also breaking down barriers between vendors in the emerging NFV ecosystem. While we all work together to resolve definitions of standards, speed is of the essence to ensure we keep pace with market demand. European Telecommunications StandardsInstitute (ETSI) The ETSI NFV initiative was launched in January 2013. This is a prestandardization effort that brings together most (if not all) Tier 1 CSPs worldwide, telecom equipment providers, and IT/ISV vendors. ETSI’s objective is to accelerate the industry’s move to NFV with this community of technical, operations, and business people specifying architecture, requirements, and use cases that are necessary to design and deploy a multivendor, open- standard NFV telecom network. How HP is participating in ETSI HP has been instrumental in the creation of ETSI NFV, having worked for the past two years with the key founders. And HP has contributed heavily in shaping this initiative. HP technical and business experts contribute to most of the ETSI NFV working groups. In particular, HP has been quite active in helping
  22. 22. 25Communications Edition develop four high-level NFV specifications that ETSI has released: • Terminology • End-to-end architecture • Requirements • Use cases Our contribution is based on our many projects with CSP clients and our ongoing product development in the NFV space. HP anticipates that the new specifications will help the industry accelerate NFV implementations. HP NFV architecture and solutions are already compliant with these specifications. Further, HP is working with a number of operators on ETSI proof of concept (PoC) projects. Some of these have started to be communicated internally in ETSI meetings. HP committee memberships • ETSI NFV Steering Committee: HP is a full member. • NFV ISG: HP is co-chair of this new group, comprising 95 companies and 257 participants. • ETSI NFV Working Groups: • Infrastructure: Paul Bottorff, HP Networking and Lloyd Mayhew, ISS • Software architecture: Marie-Paule Odini, HP CMS (co-chair) • Management and orchestration: Marc Flauw, HP CMS • Security: Boris Balacheff, HP Labs • Proof of concept: Marie-Paule Odini, HP CMS
  23. 23. 26 Communications Edition Open Networking Foundation (ONF) ONF is a non-profit industry consortium that is leading the advancement of software- defined networking (SDN) and standardizing critical elements of the SDN architecture. The consortium is standardizing the OpenFlow protocol, which structures communication between the control and data planes of supported network devices, and enables global coordination of network devices. ONF is also working on NorthBound Interface, enabling applications to interact with the network control plane, so that applications can communicate their requirements to the network and get notified of network conditions impacting them. As part of this work, ONF is looking at traffic steering and service chaining. How HP is participating in ONF HP is one of the founding members of ONF, which was created in 2011. HP began working on OpenFlow several years ago with Stanford University and has supported and contributed to this standard initiative since its inception. Switches from HP Networking support the OpenFlow standard, and HP has announced an SDN controller as well. HP is also spearheading an SDN partner ecosystem for third parties to develop ONF-compliant applications on top of an HP controller. HP committee memberships • Chair, Extensibility Working Group: Jean Tourrilhes, HP Labs, is responsible for OpenFlow specification since its inception in 2011; member of the Technical Advisory Group (TAG). • Vice chair, Chip Advisory Board (CAB): Mark Gooch, HP Networking. • Chair, NBAPI group: Sarwar Raza, HP Networking
  24. 24. 27Communications Edition Alliance for Telecommunications Industry Solutions (ATIS) ATIS brings together the top global information, communications, and technology companies to advance the industry’s most critical business problems. Focused on SDN, ATIS drives its viewpoints and standards requests through other bodies such as the 3rd Generation Partnership Project (3GPP) for mobile. How HP is participating in ATIS HP is a member of the ATIS Technology and Operations (TOPS) Council, which is defining the ATIS specification for telecom. HP participated actively on the SDN Landscape Team by providing most of the telecom SDN use cases, which constitute the main part of the team’s deliverable. This effort was then submitted to the TOPS Council that launched further investigation with a focus group on Operations Support System (OSS). HP is also participating in that effort. HP committee memberships • TOPS Council: Jeff Edlund, HP CMS CTO • SDN Landscape Team: Marie-Paule Odini, HP CMS, and Vishwas Manral, HP Networking TM Forum TM Forum leads the industry in defining the frameworks for OSS (both assurance and fulfillment) in the telco space. HP is a major contributor in the Catalyst work that goes on there. Jeff Edlund is on the Executive Committee, and Marc Flauw is recognized as a senior fellow in TM Forum and has written many of the accepted frameworks definitions there. Moving forward together HP continues to work with our colleagues at ETSI, ATIS, ONF, and TM Forum to drive NFV forward. We are now also engaged with the Internet Engineering Task Force (IETF) that has initiated some work on service chaining and SDN. To learn more about NFV solutions from HP, visit Marie-Paule Odini CMS Chief Technologist Office Communications, Media and Entertainment HP Enterprise Services
  25. 25. Guest Author: Rose Schooler, Intel Intel is enabling communications infrastructure to implement NFV At Intel®, we clearly see that software-defined networking (SDN) and network functions virtualization (NFV) are enabling the transformation of networking. In early conversations with telecommunication service providers (CSP), we learned that both data and signaling traffic driven by new end-user devices, primarily smart phones, were rapidly increasing the burden on the communication networks. The costs of managing, securing, and storing this data deluge on these networks would eventually surpass revenues. But, working with HP, we’re taking advantage of the high-volume server economics and IT technologies such as virtualization, increasing asset utilization to reduce service provider capital and operating expenditures. Coupling that with standard, open, and programmable interfaces, we are unleashing broader market participation to deliver new applications and services and helping to evolve the communications network into a platform for innovation. Identifying new architectures The service providers came to us because they saw that virtualization and cloud technologies could help meet their challenges. Intel processor-based servers and storage architectures are ideal for supporting new approaches such as NFV. We have been working with data centers for a long time to help them transition from proprietary IT infrastructures to
  26. 26. 29Communications Edition open systems. Service providers are now going through that same transformation. But the key question was whether we could work with leading IT vendors, including HP, to successfully bring NFV into the unique communications environment. Fortunately the volume, breadth, and long history of Intel’s work with service providers gives us an understanding of communications workloads. We understand which technologies can be brought into that domain and how they can be applied and adapted to meet the performance requirements for communications and storage-specific workloads. Further, we work closely with the Intel CPU development teams internally to achieve tight integration of requirements for communications networks. “Working with HP, we’re taking advantage of the high-volume server economics and IT technologies such as virtualization, increasing asset utilization to reduce service provider capital and operating expenditures.” –Rose Schooler, Intel
  27. 27. 30 Communications Edition 3,500Number of HP information management consulting professionals, averaging 18 years of expertise 1,100HP’s community of 1100+ global data analysts have experience in statistical and data analytics + What we’re doing now for the communications industry Intel has more than 17 pilots or proof of concept projects in progress with a number of major service providers and communications equipment manufacturers, in all major geographic regions. This is part of our systematic effort to apply server and IT technologies with a proven approach of working directly with service providers to test and prove our capabilities in new contexts such as NFV. Our efforts continue to yield critical benefits for providers. Enabling network flexibility The HP Software-Defined Networking (SDN) Open Ecosystem and HP SDN Developer Kit were created to reduce complexity to accelerate the delivery of next-generation services. As an ecosystem partner, Intel is simplifying device programmability across the network via OpenFlow-enabled devices based on the Intel ONP server and switch reference designs. Lower total cost of ownership (TCO) HP is an ecosystem partner supporting the Intel Open Network Platform (ONP) server and switch reference designs, which can run on HP ProLiant Gen8 Servers and HP Networking solutions. The servers are based on the Intel Xeon® processor E5-2600 v2 product family and allow customers to consolidate networking workloads, reduce deployment costs, and shorten provisioning time. Higher performance At the Open Networking Summit (ONS), Intel, together with HP and Verizon, demonstrated the orchestration of wide area network (WAN) bandwidth burst with virtual machine bursting, which can be used by private cloud operators who must move application workloads to public clouds during periods of high load. Running dual Intel Xeon processors E5-2600, the solution dynamically adjusts for changes in compute workload demand and network bandwidth usage patterns, as described in a joint paper. Migration to Intel architecture A large, multinational telecom company worked with Intel to convert processes running on purpose-built hardware in the network edge infrastructure into software-based solutions that can be virtualized and run on Intel architecture processors. The solution enables the company to increase service innovation, reduce development cycles, and lower costs for network operation and evolution. Advancing 4:1 communication processing Over the past several years, the Intel Communications and Storage Infrastructure Group has made tremendous strides with its 4:1 communication processing workload consolidation strategy to run all major communications workloads (application, control, packet, and signal processing) on standard Intel Architecture (IA). The recent
  28. 28. 31Communications Edition agreement to acquire certain assets of the wireless division of Mindspeed Technologies Inc. adds a wireless access team that will deliver small cell system-on-a-chip (SoC) and accelerate Intel’s big cell development. Working together to transform the communications industry ecosystem The thoughtful sharing of inspiration and resources can accelerate the transformation of communications. That’s why Intel and HP have created SPECTRUM, a program designed to accelerate the development and application of technologies for communications solutions builders. SPECTRUM enables developers of hardware and software to deliver solutions to meet the new challenges of the communications market. To learn more, visit You also can download the “Intel/HP/Verizon cloud/WAN Bursting Reference Architecture” and other resources at the Intel Network Builders site, And learn more about NFV solutions from Intel and HP while attending Mobile World Congress in Barcelona by visiting our booth and joining a tour of our NFV demos. Please reach out to Intel and HP to learn more: Lloyd Mayhew WW Business Development Manager Telecommunications Industry HP Servers Hewlett-Packard Ron Whitfield WW Market Development Manager HP Communications Storage & Infrastructure Group Intel Corporation Rose Schooler Vice President, Data Center Group General Manager, Communications and Storage Infrastructure Group Intel
  29. 29. 33Communications Edition NFV orchestration: Efficiently manage the NFV infrastructure Network functions virtualization (NFV) promises significant reductions in capital and operating expenses while increasing business agility, network flexibility, and innovation. However, as communication service providers (CSPs) evaluate the required transformations in their networks, they face a key challenge in how to manage NFV. The recent industry initiative to define and develop NFV indicates ways to address management needs through an orchestrator. But several questions still remain. • How will the orchestrator proposed in the European Telecommunications Standards Institute (ETSI) model meet operational requirements? • How will the operations support system (OSS) adapt to support NFV? • What will the impact be on operational business processes? • How can it adapt to a mix of NFV and non- virtualized network elements? Implementing NFV orchestration While the ETSI specifications for NFV orchestration are still evolving, the facets of an ideal NFV orchestrator should include the following: • The NFV orchestrator should allow the CSP to improve the customer experience. This requires visibility into the performance of provided services in real time. The higher levels of automation and consolidation provided by the NFV orchestrator translate to increased agility and a better understanding of the end-to-end customer experience. This is particularly important for new services aimed at highly segmented target markets so the CSP can test new offerings and either quickly scale or shut them down based on the customer experience and market acceptance.
  30. 30. 34 Communications Edition • The NFV orchestrator should firmly link service provisioning to their assurance. This requires an automated loop that can provide resources on demand and grow the network based on usage, rather than building capacity in the traditional CapEx- heavy, stair-step approach. Automation in the orchestrator can provide configuration and monitoring of the entire loop to support a more agile form of OSS, delivering network setup, provisioning, and assurance as a self- organizing network. • Developing the ideal NFV orchestrator requires alignment and involvement in ETSI to define a widely accepted standard. NFV architecture is inherently open. That’s what makes it cheaper and more flexible than provisioning with proprietary network elements. The future for communications services is multivendor, not mono-vendor. • The NFV orchestrator should be an expression of the convergence between CSP networks and IT architectures. It should fully leverage the strides that IT has made with virtualization and apply them to the CSP network. HP Network Functions Virtualization Director solution suite The HP NFV Director solution suite provides the ability to track and implement the NFV orchestrator as defined by ETSI. The NFV Director maintains the mapping among physical resources, virtual resources, and application instances. It is aware of the application high-
  31. 31. 35Communications Edition availability model and deploys app instances correctly to avoid single points of failure. Further, the application element management system (EMS) for third-party network elements remains 100% responsible for application performance and availability. This assumes that the NFV Director delivers the appropriate instantiation of that service under a virtual environment, according to the service level agreements that are given to HP by the EMS. Other important functionalities for the NFV Director include: • Management of multiple hypervisors and controllers • Connectivity provisioning across software-defined networks (SDNs) and the legacy network • Allocation of resources across applications and with orchestration so that applications don’t have to request resources and step on each other • Resource status reporting to OSS and applications Preparing for NFV NFV solutions proposed by vendors should include OSS software as well as transformation consulting to ensure the CSP OSS systems are ready for NFV. It’s important to combine both advanced telecom and IT management capabilities so the entire infrastructure can be managed both efficiently and effectively. Resource management and an orchestration layer for NFV can be achieved by using the new HP NFV Director, a solution suite that integrates seamlessly with HP and third-party management systems. Essentially, it is an “OSS in a box” that can manage complexity in both NFV and OSS because it was built in an OSS environment. The HP NFV Director is part of the end-to-end NFV portfolio from HP that includes infrastructure, abstraction, orchestration, and applications solutions. To learn more about the HP NFV Director solution suite and the full portfolio of NFV solutions from HP, visit Pierre Lavillat Head of OSS Product Management HP Communications and Media Solutions
  32. 32. 36 Communications Edition Microservers, an essential addition to your NFV toolbox Microservers may provide an alternative approach to alleviating space, power, and latency bottlenecks in network functions virtualization (NFV). Do microserver platforms belong in your NFV toolkit? Read on to decide what applications are right for this exciting new approach. Hypervisors provide a powerful abstraction layer over which multiple virtual machines (VMs) performing different network functions can co-reside on the same physical hardware. This improves overall resource utilization while also improving the flexibility and dynamism of the network service. However, this abstraction layer comes at a price in terms of non-deterministic performance and latency. Communication service providers (CSPs) who are used to delivering highly demanding applications within tight latency budgets are looking for alternative platforms. Developments in microserver technologies may be the essential addition for CSPs to successfully implement NFV for a full complement of current services. The challenge of hypervisors In the context of many applications, the overhead introduced through hypervisors is negligible and the inexactitude of the virtual representation is tolerable. However, the limitations of hypervisor-enabled virtualization of x86 architectures are felt acutely in specialized network applications that cannot tolerate non-deterministic performance or increased latency. The hypervisor itself consumes some resources and adds to packet processing overhead. Further, not all hardware resources, including caches, buses, graphic accelerators, network processors, and digital signal processors, can be sliced infinitesimally through hypervisor- enabled virtualization. Increased latency occurs due to hypervisor- induced overheads in transmitting packets between the wire and the VM. It can also result from the unsuitability of x86 processors for
  33. 33. 37Communications Edition specific applications such as signal processing and voice transcoding. The microserver is an emerging platform that is characterized by a shared infrastructure environment where a number of single-socket servers are packed into a dense, power-efficient ecosystem. This can provide an alternative approach to alleviating space, power, and latency bottlenecks in CSPs by using workload- specific processing units in dense, power- efficient form factors. The microserver revolution Microservers are revolutionizing the server market with a fundamentally new approach to packaging servers with low-power, purpose- built processors in very high densities. This is made possible by system-on-a-chip (SOC) architectures that have revolutionized personal computing, especially in power-sensitive mobile devices. And the continued expansion of mobile device access is accelerating the adoption of these new processing systems. This approach is a welcome development to address severe power constraints in data centers, where the implications for optimizing both IT and network functions cannot be emphasized enough. That’s why the first generation of servers used ARM® processor architectures and were aimed directly at IT functions such as web page serving, query processing, and cache retrieval. Ubiquitous mobile access and other scale-out workloads may further transform server requirements from those capable of processing heavy, computational workloads to processing millions
  34. 34. 38 Communications Edition of relatively smaller workloads, paving the way for microservers. SOC architectures are rapidly evolving to incorporate digital signal processing, graphics, network processing, and other specialized functions alongside multiple cores to make the next-generation “software-defined server” a reality. These drivers of transformation from pure computation to data access make microservers an ideal match for typical CSP applications. Are microservers right for you? CSPs run very specific workloads like transcoding and deep packet inspection that move a large number of packets into and out of a server with specialized, well-defined functions. Such applications also require a control plane driven by software running on x86. Microservers with cartridges that combine specialized silicon (GPUs, DSPs, NPUs) with general-purpose processors (ARM, x86) are ideal for these needs. And standard infrastructure provisioning technologies such as OpenStack are being extended to bare-metal environments, including microservers. Low-latency packet processing has also driven hypervisor-bypass technologies, such as the Intel Data Plane Development Kit (DPDK), that vastly reduce packet ingress/egress latencies at some cost to VM dynamism. Microservers can be combined with hypervisor-bypass to reduce end- to-end latency for the application. But there are some limitations to running bare- metal applications on microservers: • They’re not for all applications. • They entail a loss of some capabilities, such as the ability to suspend/resume or migrate a virtual machine. • They are an emerging technology and not yet appropriate for all customers. • They have a new, immature software tool-chain. • Applications need to be developed. Is a microserver or traditional x86 server the right NFV platform for your application? To find out, review the matrix of application characteristics on the next page. Launching CSPs into a more competitive future CSPs are excited about the application innovations that can be delivered in an optimized and cost-effective implementation based on microservers. And HP is already talking to application vendors about the new opportunities created by microservers. Further, we’ve launched the HP Moonshot System, a second-generation, microprocessor- based server. Moonshot is the world’s first software-defined server. It accelerates innovation while delivering breakthrough efficiency and scale.
  35. 35. 39Communications Edition Moonshot is the result of years of intense collaboration across many organizations within HP including HP Labs, HP IT, HP Cloud Services, HP Software, and other groups within the HP Enterprise Group (network, storage, solutions). This effort goes beyond HP to include joint development with leading silicon and software partners, as well as countless interactions with customers. Moonshot is just one of the initiatives that HP is taking to help make an open network ecosystem possible for CSPs. To learn more about the broad portfolio of NFV solutions available from HP, visit Application characteristic Microservers may suit your application if… Are power and footprint serious considerations in the data center? Data centers with limited power or footprint may benefit from microservers. In many cases, these alternative architectures reduce power and space requirements by up to 50%. Does the function scale out? Is the function parallelizable? If the function can be distributed across multiple servers, then a multitude of low-powered specialized chips may be better than a GPU architecture. Is the application limited in its memory usage? If no more than a limited amount of RAM per server is needed, then a microprocessor platform may be considered. Memory-intensive applications will fit better on traditional x86 architecture, though the exact boundaries of processing and memory capacity will expand over time. Does the application have limited per-server overall and aggregated uplink bandwidth requirements? The current generations of microservers are limited in their I/O. Traditional computing platforms support a much greater diversity of networking technologies (iSCSi, Infiniband, etc.), as well as speeds. Can the application be accelerated through special purpose hardware, such as NPUs, GPUs, DSPs or ASICs? A microserver platform with readily available hardware acceleration may be the right approach. Does the application have a stringent latency requirement? Tight latency requirements for node-to-node communication are better suited for a microserver. Are power and footprint serious considerations in the data center? Data centers with limited power or footprint may benefit from microservers. In many cases, these alternative architectures reduce power and space requirements by up to 50%. Raju Rajan Verizon Account Chief Technologist HP Enterprise Services
  36. 36. 41Communications Edition SDN and NFV together: Q&A with Dave Larson Industry Edge HP Networking is very active in both the SDN and NFV space. Can you tell us what you are doing to help the service providers? Dave Larson, CTO and Vice-President of HP Networking The communication service providers (CSPs) have been struggling with an inflexible network built with vertically integrated silos that are physically chained together and hard to scale up or down, making it difficult to create new services. They have done their homework in the past few years, and they are already convinced that network functions virtualization (NFV) and software-defined networking (SDN) can help them get beyond this problem. I’d like to talk about how. SDN is an important step in helping CSPs become more agile as the size and dynamism of their networks increase. And it’s a key enabler for successful deployments of NFV in the CSP environment. Working together, SDN and NFV provide synergies for creating a complete architecture to help CSPs become more innovative and competitive. Industry Edge Do you see NFV as a new infrastructure model for network functions, creation of new network functions and services, or a combination of these? Or do you see NFV from a different perspective? Dave Once you combine NFV with SDN, the answer is “all of the above.” SDN is a complimentary technology for NFV. In turn, NFV deployments can enable easier deployment of SDN. I see this working in three distinct ways. For NFV to really work, it’s not just about ensuring that the traffic is directed to the appropriate service, but also intelligently routing the traffic across a cloud of virtual service appliances. As an example, any kind of policy enforcement implemented as a virtual network function can be scaled by cleverly distributing layer 3 and 4 policies across the entire fabric so that inefficiencies such as centralized traffic redirecting can be avoided. SDN provides for service chaining, auditability, and insurance of policy enforcement for services. The cloud bursting proofs of concept demonstrated by HP, Intel, and Verizon at ONS is one such example.
  37. 37. 42 Communications Edition Secondly, SDN-enabled virtual devices provide dynamic programmability for NFV. SDN-enabled virtual devices open up many possibilities. For example, SDN protocols allow virtual firewalls and load balancers to be modified dynamically and cooperate with centralized controllers to simplify operations. Finally, SDN and NFV deployed cleverly have the effect of 1 + 1 > 2. With NFV, you can easily decompose network functions and implement them using the transport layer itself. This includes any kind of policy enforcement implemented at layer 3 or 4, and distributed into SDN-enabled generic fabric. The result is the kind of efficiency not previously possible with physical network devices—without SDN capability. Sentinel is a good example of using the network to implement basic IPS services across the entire fabric, rather than redirecting traffic to a physical or virtual IPS. We are running this with a large, multinational service provider. Industry Edge Looking at SDN, what do you see as its strongest drivers and use cases­—reducing network costs, creation of new service types, or perhaps something else? Dave Necessity is the mother of invention. Our customers are usually in one of the two situations or both. They see the cost of bandwidth rising while revenues stay relatively flat; they see OTT providers taking more and more wallet share of the end users. Therefore, the drivers and use cases will be in both cost reduction and new revenue streams. If I were to pick one, I would focus on top-line growth, create new revenue streams, and compete with OTT head-on. This cannot be possible with the legacy network model as it does not allow for business agility whereas the SDN- enabled network does. Industry Edge Industry pundits have raised potential obstacles to NFV and SDN adoption, such as the need to meet the stringent performance requirements of network functions infrastructure. Do you see this as a key issue, and is it solvable? Dave Both NFV and SDN have been adopted in various use cases. It is causing a big stir for the incumbent siloed solution providers. NFV is based on industry standard servers, storage, and networking hardware. With the help of technologies like SR-IOV and DPDK, virtualized devices provide no less performance than the specialized hardware. In fact, the specialized hardware often lags behind the Moore’s Law. Virtualization allows you to have the flexibility to distribute and scale. Industry Edge It has been suggested that NFV will coexist with traditional network functions and architectures for CSPs, and that this will require solid integration between NFV and OSS. Do you see this as a key consideration? Dave We see this as a key requirement for the CSP. Virtualized network devices and architecture
  38. 38. 43Communications Edition should be interoperable and manageable just like physical networks, with the added benefits of being more flexible, agile, and scalable. Industry Edge It has also been suggested that as the number of NFV network functions grow, coherent orchestration across the multiple NFV functions into the underlying infrastructure will become increasingly important. Is this a view you agree with? If so, should scalable orchestration be built in to NFV planning from the outset? Dave Orchestration is important to fully benefit from the agility and simplicity that NFV promises. In our view, orchestration strategy isn’t just about the ability to create, manage, and automate individual NFV functions. It is also an opportunity to cleverly construct the NFV so that it requires level orchestration or management. HP Networking is well known for simplified scaling by virtualizing across physical switches or routers, using the Intelligent Resilient Framework (IRF). We are applying our IP and expertise to bring simplified NFV architecture to large-scale NFV deployment for CSPs. Industry Edge Anything else you want to leave us with? Dave HP began the development of SDN components for NFV at a very early stage. They include the HP Virtual Services Router (VSR) designed for multitenant, hosted public clouds, and virtualized branch CPE deployments. We offer CSPs a ready-to-be- deployed architecture that can enable the transformation from legacy networks to NFV- and SDN-enabled infrastructure. For more information, visit
  39. 39. Learn more For more information, visit and come see us in our booth in Hall 3 Stand 3A20 at Mobile World Congress. Follow us on HP CME Twitter Read HP Telecom IQ Blog Editor Julia Ochinero ( © Copyright 2014 Hewlett-Packard Development Company, L.P. The information contained herein is subject to change without notice. The only warranties for HP products and services are set forth in the express warranty statements accompanying such products and services. Nothing herein should be construed as constituting an additional warranty. HP shall not be liable for technical or editorial errors or omissions contained herein. ARM is a registered trademark of ARM Limited. Google is a registered trademark of Google Inc. Intel and Xeon are registered trademarks of Intel Corporation in the U.S. and/or other countries 4AA5-1106ENW, February 2014 Rate this documentShare with colleagues Sign up for updates