Transforming Networks with NFVI, HP Carrier-Grade Servers, and Intel ONP

Transcription

1 white paper Transforming Networks with NFVI, HP Carrier-Grade Servers, and Intel ONP Intel Network Platform Group Developing Solutions with Intel ONP Reference Architecture Network virtualization and cloud-enabled technologies are steadily reshaping the network architectures of telecommunication providers, creating strong incentives to demonstrate the viability of these technologies for commercial use. As commercial deployments accelerate, work still needs to be done. Through an open ecosystem and open standards approach, stakeholders are collaborating to improve interoperability at each layer of the Network Function Virtualization Infrastructure (NFVI). One of the primary objectives in this collaborative work is to measure performance in a consistent and predictable way. Another goal is to minimize the complexity in planning and implementing infrastructures based on software-defined networking (SDN) and Network Functions Virtualization (NFV). To these ends, Intel and HP are removing the guesswork from the integration and performance measurements through a commercial platform based on the Intel Open Network Platform (Intel ONP) reference architecture. Intel ONP offers a blueprint for redefining network architectures around the principles of virtualization. This reference architecture decouples network functions from the underlying hardware components, setting up an infrastructure for service chaining and providing a model open to centralized provisioning and management of network resources. HP has advanced the architecture further with the HP OpenNFV platform, adapting it for commercial deployments. The rapid transformation of network infrastructures from fixed, physically based constructs to virtualized, highly manageable frameworks demands a structured and open approach. Through the efforts of Intel and HP, telecommunication providers, enterprises, and cloud-service providers are able to more accurately assess the virtualized environments as part of an NFVI. This paper explains the concepts surrounding the HP NFVI evaluation platform and provides highlights of the initial test results. What is the Intel Open Network Platform Reference Architecture? To streamline the evaluation, design, and deployment of open SDN and NFV solutions, the Intel ONP reference architecture helps accelerate development of commercial hardware and software platforms (see Figure 1). With Intel ONP, key leaders in telecom carrier networks, enterprise environments, and cloud data centers can more easily build solutions using an opensource software stack running on industry-standard high-volume servers. The reference architecture gives solution providers a way to plan, evaluate, and benchmark components in advance of NFV deployments. This effort has the support of industry consortiums, telecommunications and cloud providers, and leading companies involved in open-source projects.

2 2 Table of Contents What is the Intel Open Network Platform Reference Architecture?...1 ETSI Established the Framework...2 Intel ONP Provides a Platform to Develop Commercial Solutions...3 Ensuring an Effective Framework for NFV Development...3 HP Introduces a Platform for NFV Development...4 HP NFV Options...4 NFV Reference Architecture...4 NFV System...4 HP OpenNFV Labs...5 Common Platform to Develop and Demonstrate End-Use Conditions...5 HP/Intel ONP Test Environment..6 Benchmark Results - Bare Metal L2 Forwarding...6 Benchmark Results - SR-IOV/VM L2 Forwarding...7 Benchmark Results OVS Forwarding...8 Summary of the Test Results...9 Future Study...9 Next Steps: Future Collaborations...11 Summary...11 Test Configuration Details...12 A Server Reference Architecture Optimized for SDN/NFV Software Stack Based on Open-Source Open Standards Industry-Standard Server Based on Intel Architecture VM VIRTUAL SWITCH HW OFFLOAD Intel ONP is based on the European Telecommunications Standards Institute (ETSI) framework, as described in the section that follows. ETSI Established the Framework ETSI through the Industry Specification Group (ISG) for NFV has provided much of the groundwork for implementing NFV technology, mapping out the architectural framework in a group specification, ETSI GS NFV 002 v As shown in Figure 2, this NFV architectural framework identifies functional blocks and the main interfaces between the blocks. A number of them are already available in current deployments; others can be added to further virtualize network operations. These ingredients provide the inspiration and technical basis for Intel ONP, which draws heavily on the framework that ETSI has developed. The primary functional blocks include: Virtualized Network Function (VNF) Element Management System (EMS) DPDK LINUX/ KVM Figure 1. The Intel Open Network Platform reference architecture enables efficient development of software-defined networking/network Functions Virtualization solutions. - Virtualized Infrastructure Manager(s) - Orchestrator - VNF Manager(s) Service, VNF, and Infrastructure Description Operations and Business Support Systems (OSS/BSS) The functional block diagram of the NFV architectural framework (shown in Figure 2) includes reference points (indicated by solid lines) that are within NFV scope. The functions shown in this architectural framework highlight only those elements that are essential for virtualizing network operations. The operators of the network can selectively decide which of the network functions should be virtualized. NFV components can also work effectively with legacy hardware in hybrid environments, allowing system architects to progressively phase in virtualized appliances over time. NFV Infrastructure, including: - Hardware and virtualized resources - Virtualization Layer

3 3 OSS/BSS Service, VNF and Infrastructure Description Os-Ma Ve- Vnfm NFV Orchestrator Or-Vnfm In short, the ETSI NFVI framework served as the basis for the Intel ONP reference architecture, which in turn underlies the HP NFV reference architecture (see Figure 3). The end result is a commercial, turnkey NFV system ready for deployments. EMS 1 EMS 2 EMS 3 VNF 1 VNF 2 VNF 3 NFVI Virtual Computing Virtual Storage Vn-Nf Virtualization Layer VI-Ha Virtual Network HARDWARE RESOURCES Ve- Vnfm Nf-Vi VNF Manager(s) Virtualized Infrastructure Manager(s) Vi-Vnfm Or-Vi Intel ONP Provides a Platform to Develop Commercial Solutions As SDN/NFV technology matures and participating companies define their open-source components and underlying standards, Intel ONP serves as a model for an interoperable hardware/software platform upon which building blocks can be added to deliver specific NFV capabilities. Computing Hardware Storage Hardware Network Hardware Execution reference points Other reference points Main reference points The open-source software stack runs on top of the hardware platform specified by Intel ONP as shown in Figure 4, which is based on standard, high-volume servers (SHVSs) powered by the Intel Xeon processor E5 family. Figure 2. An end-to-end functional block diagram of the Network Functions Virtualization architecture defined by the European Telecommunications Standards Institute. ETSI GS NFV 002 v1.1.1 Intel Open Network Platform reference architecture, release 1.4 Originally mapped out the architectural framework for implementing NFV technology. Developed by the European Telecommunications Standards Institute (ETSI) through the Industry Specification Group (ISG). Intel ONP provides a blueprint for building a hardware/ software NFV platform that supports NFV capabilities. Ensuring an Effective Framework for NFV Development A reliable, hardened NFVI platform is a prerequisite to commercial developments. This platform must be current integrating the latest versions of components from open-source projects and deliver assurance that the components specified are fully interoperable. HP OpenNFV, NFV reference architecture HP OpenNFV defines a reference architecture for delivering commercial NFV solutions to market. A complete, integrated NFV platform to accelerate virtual network function development. Turnkey NFV system Figure 3. The taxonomy of Network Functions Virtualization (NFV) architectures.

4 4 Open-Source Software Stack Based on ETSI-NFVI Reference Architecture Open vswitch An Open Virtual Switch Intel Xeon processor E5 v3 The value of Intel ONP lies in the demonstrated wide-scale interoperability of proven components. Through the tested reference architecture ingredients and the industry standardization work, NFV developers have a solid infrastructure and compatible building blocks to confidently work on solutions that further advance the development of SDN solutions. HP is the largest single contributor to OpenStack*, and Intel has also made substantial contributions to this and related open-source projects, such as OpenDaylight and Open vswitch*. To reduce the complexity for planners and developers, Intel and HP have shared knowledge and expertise, working jointly on proof-of-concept (PoC) projects. OpenStack* Cloud OS Open Daylight Controller DPDK Open vswitch* Linux* Fedora OS KVM Hypervisor Intel QuickAssist Technology Driver Intel Ethernet Drivers: 10 and 40 GbE Industry-Standard High Volume Server Intel Communications Chipset 89xx Series Figure 4. Software stack and hardware platform for the Intel Open Network Platform reference architecture, release 1.4. HP Introduces a Platform for NFV Development By integrating the latest SDN/ NFV technology advances into a standardized architectural framework, HP has engineered a flexible, proven NFVI to support NFV developers and their customers. HP OpenNFV details how a collection of NFV components can be rapidly assembled to construct a complete NFV infrastructure (NFVI). The distribution of NFV solutions takes place through: NFV reference architecture NFV System Intel Ethernet Controller XL710 HP NFV Options Communication service providers (CSPs) evaluating NFV solutions have two available options from HP. The NFV reference architecture supplements existing systems by identifying HP products that readily integrate with scalable, high-performing, and robust NFV solutions. NFV System, the second option, offers a fully integrated turnkey solution that can be ordered through single stockkeeping units (SKUs). NFV Reference Architecture The NFV reference architecture provides a set of hardware, software, and networking configuration rules and best practices for building NFV solutions. The architecture also outlines multiple hardware configurations to address typical use cases and different performance requirements. Addressing compute, storage, and networking with a broad set of products, the reference architecture is an effective tool for solution architects and accounts that are developing customized solutions for customers with specific needs and requirements. The architecture also covers software and virtualized solutions. NFV System HP NFV System is a complete integrated NFV platform designed to enable CSPs to accelerate VNF deployments. A fully integrated stack that comes with hardware and software installed and configured, this turnkey solution simplifies the move of CSPs VNFs into customer trials and production. The solution features components from HP s hardware portfolio that have been carefully chosen to satisfy industry requirements. Specific care has been taken to provide a platform that is

5 5 flexible, scalable, and able to meet any type of data plane performance, as well as meet typical telecom requirements. The HP NFV System solution includes HP Helion OpenStack Carrier Grade, which builds on HP Helion OpenStack to provide a highly available and highperformance cloud platform wellequipped to run demanding VNFs. Because physical infrastructure management (PIM) and virtual infrastructure management (VIM) are tightly integrated, managing this stack is easy. HP OneView is fully integrated with HP Helion OpenStack Carrier Grade. Hardware events are elevated throughout the stack and available for consumption. To simplify the process of provisioning hardware and software components, orchestration is fully integrated with HP Helion and HP OneView. HP also offers a full selection of management and orchestration components for the MANO layer of an NFVI environment, as part of the HP OpenNFV offering. HP OpenNFV Labs Developed around principles of openness, OpenNFV gives CSPs the means to select their own VNF partners. HP then helps partners validate the interoperability of components within complex NFVI environments through HP OpenNFV Labs. These labs are located worldwide: in Houston, Texas, USA; Grenoble, France; Tel Aviv, Israel; and Seoul, South Korea. CSPs can also conduct NFV PoC and feasibility tests for new NFV applications on the HP NFV reference architecture. Through the availability of HP s NFVI, software vendors (ISVs), technology partners, system integrators, and network equipment providers (NEPs) can more efficiently establish NFV environments and take advantage of OpenNFV labs located worldwide for performing partner integration and developing PoCs. The labs serve as incubators for NFV innovation and are open to all companies that want to validate their applications (subject to resource limitations and funding of necessary resources). HP can provide testing and assurance to CSPs that their applications and their preferred partners can be validated on the reference architecture platform. Why Network Functions Virtualization? Today s mobile and fixed networks are populated with many types and instances of proprietary hardware devices designed to run communication service provider (CSP) applications. With NFV, CSPs are migrating these applications to a common operating environment on a common platform, taking advantage of modern open and standard hardware and software advances (in particular, cloud, converged IT infrastructure, and virtualization). With these, CSPs can realize significant gains in agility and a reduction in operational expenses, as well as potential reductions in capital expenses, through the use of virtualization for network applications and services. 2 With more than 20 global NFV proof-of-concept projects in motion, we are well positioned to understand and help guide carriers on their journey to cloud-based delivery models. Our open architectural approach, in collaboration with Intel and our OpenNFV technology partners, is designed to ensure that carriers have a flexible, multivendor platform from which to quickly test and then launch new and innovative services. 3 - Werner Schaefer, VP Network Functions Virtualization Business, Hewlett Packard Common Platform to Develop and Demonstrate End-Use Conditions HP is actively involved in many standards organizations as a board member, as a committee chair, or making significant contributions to the following: Alliance for Telecommunications Industry Solutions (ATIS) CloudEthernet Forum, ETSI Open Networking Foundation (ONF) TM Forum OASIS Open Data Center Alliance (ODCA) Internet Engineering Task Force (IETF) OPNFV Top contributor and user of NFV/ SDN open-source initiatives, such as OpenStack and Open Daylight

6 6 Using the ingredients of Intel ONP, HP identified a set of key building blocks for an NFV solution. This set of building blocks resulted in the specification for the HP OpenNFV reference architecture, the basis of tested commercial offerings by HP. The reference architecture brings a number of components together as a complete deployable package. The package includes HP s cloud, OpenStack, networking, storage, server, operations support system (OSS), IT management, and professional services expertise, along with key partners and HP virtualized network functions. HP plans to continue aligning its commercial NFVI platform development in concert with upcoming releases of Intel ONP. Through this cooperation, HP and Intel are helping serve the needs of ecosystem partners that require reliable and proven solutions to support commercial deployments. HP/Intel ONP Test Environment The HP NFV Infrastructure Lab partnered with Intel to measure packet-processing performance on HP Servers within the Intel ONP reference architecture. The key objective of the engagement was to demonstrate that Intel ONP enables data plane performance on standard servers in support of real-time, low-latency NFV application environments. Hardware and software components of the test environment included: HP ProLiant DL380 Gen9 server - Includes the Intel Xeon processor E v3 Intel Ethernet Controller XL710 10/40 GbE Spirent network test generator Red Hat Enterprise Linux*, RHEL 7 Data Plane Development Kit (DPDK) 2.0 Open vswitch with dpdk-netdev More details of the test configuration appear on the last page of this paper. Testing focused on determining baseline packet-processing performance to enable NFV architects to develop optimal configurations for their applications. RFC 2544 methodology was applied to determine the maximum load at which zero packet loss is observed and then to measure throughput, latency, and jitter at that load for a range of packet sizes. A hardware network packet generator created network traffic for DPDKenabled L2 and L3 forwarding tests, used in combination with a hardware network test generator (see Figure 5). A series of software configurations was used to measure different models of DPDK application usage: Host-based. DPDK application running directly on the host environment, determining the maximum possible packet processing performance on the given hardware configuration. SR-IOV virtual functions. The DPDK application is running in a virtual machine (VM) bound directly to a NIC virtual function. PCI passthrough is employed to pass data directly to the NIC with no intervening vswitch. This configuration represents the maximum possible performance for a VM-based DPDK app. Accelerated vswitch. The DPDK application is running in a virtual machine using Open vswitch with DPDK-netdev to route data to the NIC. This configuration models typical VNF usage and measures DPDK performance that can be expected in a fully virtualized environment. Benchmark Results - Bare Metal L2 Forwarding Computer Node 2 L2/L3 Forwarding Application DPDK 10GbE NIC 1 3 Traffic Generator/Tester Figure 5. Test configuration for bare metal L2 forwarding. This test provided a baseline measurement of raw packet forwarding performance with a DPDK-enabled L2 forwarding test connected directly to NIC ports (see Figure 6). The throughput approaches line rate for packet sizes of 128 bytes and greater.

7 7 Throughput by Frame Size versus Theoretical Max Throughput Rate (bps) 30,000,000 20,000,000 10,000,000 Throughput Theoretical Throughput TOTAL TRIALS FAMILY SIZE (BYTES) INTENDED LOAD (%) Figure 6. Bare metal L2 forwarding results. OFFERED LOAD (%) (%) (fps) THEORETICAL MAX (fps) (Mbps) Benchmark Results - SR-IOV/VM L2 Forwarding Computer Node VM The next step after bare metal testing was to get the test running on a VM. This test utilized single-root I/O virtualization (SR-IOV) to present virtual functions on the NIC and iommu pass through to bind VMs directly to those virtual functions, with no intervening vswitch (see Figure 7). The results show that throughput levels in this configuration exhibited little or no line degradation in the network when using the VM (see Figure 8). 2 3 Pass-through or SR-IOV DPDK 10GbE NIC 1 4 Traffic Generator/Tester Figure 7. Test configuration for SR-IOV/VM L2 forwarding.

8 8 Throughput by Frame Size versus Theoretical Max Throughput Rate (bps) 30,000,000 20,000,000 10,000,000 Throughput Theoretical Throughput TOTAL TRIALS FAMILY SIZE (BYTES) INTENDED LOAD (%) OFFERED LOAD (%) Figure 8. Benchmark results for SR-IOV/VM L2 forwarding. (%) (fps) THEORETICAL MAX (fps) (Mbps) Benchmark Results OVS Forwarding Computer Node 2 This test utilized Open vswitch with DPDK-netdev in place of the L2 forwarding application to measure raw packet processing performance through the vswitch (see Figure 9). While throughput levels were lower for the smallest packet size, performance was comparable to the previous configurations for 128-byte packets and above (see Figure 10). Open vswitch DPDK 10GbE NIC 1 3 Traffic Generator/Tester Figure 9. Test configuration for OVS forwarding.

9 9 Throughput by Frame Size versus Theoretical Max Throughput Rate (bps) 30,000,000 20,000,000 10,000,000 Throughput Theoretical Throughput TOTAL TRIALS FAMILY SIZE (BYTES) INTENDED LOAD (%) OFFERED LOAD (%) Figure 10. Benchmark results for Open vswitch* forwarding. (%) (fps) THEORETICAL MAX (fps) (Mbps) Summary of the Test Results The test results outlined in this white paper validate that for baseline configurations that measure the maximum possible throughput, line rate speeds approaching 10GbE can be maintained for all but the smallest packet size (64 bytes) in our measurements. This confirms that DPDK-enabled configurations have the theoretical bandwidth capability to support the most demanding NFV application environments. With applications running in the VM, using SR-IOV, one can get almost line-rate performance as can be accomplished with a host-based configuration. Future Study Key areas for future study include: Additional Open vswitch testing configurations, including routing of packets between multiple VMs through the vswitch. This will represent another more realistic NFV application configuration and will indicate how much packet-processing bandwidth can be expected in such a deployment scenario. Execution of DPDK packet processing throughput tests with Intel Ethernet Controller XL710 10GbE and 40GbE NICs. Initial results for Intel Ethernet Controller X710 10GbE NICs are promising. Measuring throughput with 40GbE NICs should result in much higher packet rates, constrained only by PCI bus capacity on the card. Investigation of latency results. Currently, average latency looks reasonable (in the 10 microsecond range), but there is some variability in maximum latency numbers that warrant further study. Different Intel Xeon processors. Testing is planned to determine the impact of different Intel Xeon processor types, to include four processor configurations and Intel Xeon processor D product family System-on-a-Chip (SOC). Allocation of VMs to separate NUMA zones on the processor is a promising area of investigation.

10 10 Testing conclusively demonstrated the viability of Intel ONP NFV Infrastructure technologies running on HP ProLiant server platforms for supporting NFV application deployments on standard servers. New hardware configurations and advanced test configurations are planned as part of the continuing partnership between HP and Intel. Engineering and technical support that we received from Intel enabled us to make tremendous progress on the testing, especially as we got deeper into points involving the Intel ONP architecture, DPDK, and Open vswitch. Having the Intel team on board proved very useful in dealing with the complexities of the infrastructure and determining the best ways to tune and optimize at a very low level to ensure top performance. Intel absolutely helped us with our ability to get results. - Al Sanders, R&D Project Manager, Hewlett-Packard NFV Infrastructure Lab First Commercial Deployment of the Intel Open Network Platform Through a joint Intel and HP project, HP has released the first commercially available NFV environment: HP Helion OpenStack*. Intel provided optimizations for Open vswitch* and OpenStack that were incorporated into HP Helion. The turnkey commercial release, an HP market-ready NFV system, helps minimize the difficulty and complexity that many companies experience when adopting cloud-based IT solutions. HP Helion OpenStack* HP Helion Orchestrator HP Helion Linux* Debian Linux, KVM Hypervisor 1 HP and Intel collaborating to deliver commercial NFV solutions. OpenStack Optimizations for Networking: SR-IOV Open vswitch with DPDK Open vswitch An Open Virtual Switch 2 HP Helion uses key Intel optimizations across Open vswitch*, OpenStack*. Intel Xeon processor E5 v3 Intel Communications Chipset 89xx Series Intel Ethernet Controller XL710 3 HP Helion is a commercial product built with ONP hardware elements. HP ProLiant Server blade and rack mounted Figure 11. HP Helion OpenStack* released for commercial use.

11 11 Next Steps: Future Collaborations HP and Intel plan to continue work in this area, with the ongoing evaluation and use of Intel ONP as a means for HP to build commercial NFVI solutions. Keeping up with the progression of major open-source projects including OpenStack, Open Daylight, Open vswitch is a key deliverable. As new versions of Intel ONP are released, HP plans to perform benchmarking similar to the tests detailed in this paper to track and improve performance and efficiency across the network. A new industry project, the OPNFV Project, shares the goal of accelerating the release of NFV products and services, with continuing development of a carrier-grade, integrated, opensource platform for testing components from relevant upstream projects. Ongoing work by Intel, HP, and other industry leaders in OPNFV continues the ecosystem development based on open-source NFV standards. Along with other industry organizations that develop ingredients for constructing a full-featured NFVI, Intel and HP helped the community in the first release of OPNFV (Arno) in June OPNFV focuses on supporting developer efforts to build, install, and explore platform capabilities for NFVI components. We re already seeing a positive impact from NFV on the telecom market segment through a wide variety of successful proofs of concept and active involvement in solutions and standards development from every facet of the telecommunication industry. But Arno, and future OPNFV releases, will help to speed the transition from PoC to industry adoption by providing a standardized, proven, opensource NFV infrastructure that is suitable for all NFV applications. Arno is the first instantiation of an OPNFV platform and comprises the NFV infrastructure and VIM components of the NFV architecture specified by the European Telecommunication Standards Institute (ETSI). Intel and HP are both active proponents and ongoing contributors to this effort. 4 - John Healy, General Manager SDN Division, Intel Corporation Summary Benchmarking results produced on the HP ProLiant DL380 Gen9 server platform demonstrated conclusively that standard, high-volume servers can successfully deliver the level of dataplane performance required to support demanding NFV architectures and carrier-grade installations. Through ongoing technical and architectural collaboration between HP and Intel, as well as other partners working in the NFV space, Intel ONP continues to enhance the performance for enterprise and carrier-grade telecommunications deployments. Mutual contributions to industry consortia and standards bodies also move network transformation forward toward faster adoption. The knowledge and expertise accumulated from all of the collaborative projects reaches end users through the Intel Network Builders program, making it possible to add VNF on top of an established, proven infrastructure. Adoption of NFV technology adds agility and manageability to networking environments and the ability to create new services for customers more efficiently at lower costs. By establishing the framework and commercializing the infrastructure that makes large-scale NFV deployments possible, Intel and HP have given telecom providers and enterprises the confidence to move ahead with the virtualization of network services. Only eight months after its formation, OPNFV has met one of its major goals by creating an integrated build, deployment, and testing environment that accelerates NFV implementation and interoperability. With Arno, we now have a solid foundation for testing some of the key resource orchestration and network control components for NFV. This is a great testament to the power of an open-source collaborative model and the strength of the NFV ecosystem, which is supported by key IT partners Intel and Hewlett Packard. - Prodip Sen, Chairman of the OPNFV Board of Directors

12 Test Configuration Details HP ProLiant DL360 Gen9 Server. Processor type: Intel Xeon processor E GHz. BIOS settings: Hyperthreading off. Intel Turbo Boost Technology: enabled. DCU data prefetcher: enabled. DCU instruction prefetcher: enabled. Intel Virtualization Technology for Directed I/O: enabled. HP power regulator: HP Static High Performance Mode. Memory pre-failure notification mode: disabled. Operating system: Red Hat Enterprise Linux* 7. Linux configuration - Firewall: disabled. Network manager: initially disabled, but re-enabled to support networking in VM; enabled in most recent tests. Irqbalance: disabled. ssh: enabled. gdm: disabled in both base system and VM. IPV4 forwarding: disabled. Host kernel boot parameters: hugepagesz=1g, hugepages=32, isololate cores of CPU where card is with isolcpus (but not core 0). Original DPDK used: Final DPDK used: Test equipment is Spirent for traffic generation and analysis. Test methodology Is RFC2544. Throughput requires lossless packet flow. Latency measurement is LIFO. Throughput measurements were 1-minute runs typically, with some 5-minute runs. Latency measurements were 1-hour runs. CALL TO ACTION Organizations evaluating the viability of SDN/NFV solutions for enterprise deployments can choose from numerous resources, including the following: Learn more about HP s NFV solutions and HP OpenNFV: hp.com/go/nfv Learn more about the Intel Open Network Platform: Download the Intel Open Network Platform Server Reference Architecture for NFV and SDN: 01.org/packet-processing/ Learn more about Intel Network Builders: networkbuilders.intel.com/ 1 ETSI, Network Functions Virtualization (NFV): Architectural Framework (2013). 2 HP CSP NFV Cloud, (2015) INFORMATION IN THIS DOCUMENT IS PROVIDED IN CONNECTION WITH INTEL PRODUCTS. NO LICENSE, EXPRESS OR IMPLIED, BY ESTOPPEL OR OTHERWISE, TO ANY INTELLECTUAL PROPERTY RIGHTS IS GRANTED BY THIS DOCUMENT. EXCEPT AS PROVIDED IN INTEL S TERMS AND CONDITIONS OF SALE FOR SUCH PRODUCTS, INTEL ASSUMES NO LIABILITY WHATSOEVER, AND INTEL DISCLAIMS ANY EXPRESS OR IMPLIED WAR- RANTY, RELATING TO SALE AND/OR USE OF INTEL PRODUCTS INCLUDING LIABILITY OR WARRANTIES RELATING TO FITNESS FOR A PARTICULAR PURPOSE, MERCHANTABILITY, OR INFRINGEMENT OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT. UNLESS OTHERWISE AGREED IN WRITING BY INTEL, THE INTEL PRODUCTS ARE NOT DESIGNED NOR INTENDED FOR ANY APPLICATION IN WHICH THE FAILURE OF THE INTEL PRODUCT COULD CREATE A SITUATION WHERE PERSONAL INJURY OR DEATH MAY OCCUR. Intel may make changes to specifications and product descriptions at any time, without notice. Designers must not rely on the absence or characteristics of any features or instructions marked reserved or undefined. Intel reserves these for future definition and shall have no responsibility whatsoever for conflicts or incompatibilities arising from future changes to them. The information here is subject to change without notice. Do not finalize a design with this information. Intel technologies features and benefits depend on system configuration and may require enabled hardware, software or service activation. Performance varies depending on system configuration. Check with your system manufacturer or retailer or learn more at [intel.com]. The products described in this document may contain design defects or errors known as errata, which may cause the product to deviate from published specifications. Current characterized errata are available on request. Contact your local Intel sales office or your distributor to obtain the latest specifications and before placing your product order. Copies of documents, which have an order number and are referenced in this document, or other Intel literature, may be obtained by calling , or by visiting Intel s Web site at Intel processor numbers are not a measure of performance. Processor numbers differentiate features within each processor family, not across different processor families. Go to: Learn About Intel Processor Numbers *Other names and brands may be claimed as the property of others. Copyright 2015 Intel Corporation. All rights reserved. Intel, the Intel logo, and Xeon are trademarks of Intel Corporation in the U.S. and other countries. Printed in USA 0815/LB/MESH/PDF Please Recycle US