Slide 1: Easy To Build: All Programmable Carrier Grade VNF Platform
Easy To Build: All Programmable Carrier Grade VNF Platform David Levi, CEO Ken Dennen, Senior Director of Business Development for the U.S. . Intel® Network Builders Partner
Slide 2: Agenda
Agenda • Introduction • Overview: Carrier-grade Virtual Services at the Edge • Carrier Ethernet: The Foundation for VNF Services • Hardware Acceleration: Efficiency at the Edge • An FPGA Platform for Programmable Hardware Acceleration • Summary 1
Slide 3: Company Snapshot
Company Snapshot • Public company traded on LSE, founded in 2004 • $25m investment in technology development • Holds secure, enforceable and renewable Patents on its technology • Patents are around extreme lower die size implementation of network processing architecture • Result in price competitive
Slide 4: CSP Network Virtualization
CSP Network Virtualization Cloud-based Central Offices Transformation Business drivers: • Service agility – improve service experience of on-demand users • Capture new revenue generating services – many are latency sensitive • CAPEX reduction – COTS servers based on high volume merchant CPUs • OPEX
Slide 5: CSP Service Virtualization
CSP Service Virtualization Flexible NFV Workloads vBRAS vFW vEPC Virtual Machines & Containers Control Plane Disaggregation SDN Efficiency/Performance Optimizations Data Plane Programmable Vertically Integrated Intel® DPDK Hardware Proprietary Equipment Acceleration COTS Servers 3/18/2018 4
Slide 6: Delivering Services at the Edge
Delivering Services at the Edge Open source platform Open source NFV platform Standardized NFV orchestration (MANO) modeled on cloud DC for virtual network functions (VNFs) Multi-access (or Mobile) Edge Computing (MEC) Central Offices (COs): • House the edge of carrier networks • Global footprint an
Slide 7: Virtualized Central Office Architecture
Virtualized Central Office Architecture CORD CO Architecture COTS SmartNIC Programmable Hardware Acceleration Benefits: • Extend 100% COTS data plane solution • Power efficiency versus software implementation • Reduction in latency • Conserves CPU cycles better spent on higher layer services • Provi
Slide 8: Carrier-Grade Edge Service Challenges
Carrier-Grade Edge Service Challenges Enterprise Must deliver low latency vs. OTT cloud services Metro and Core OTT Mobile Access Central Core Data Data Networks Office Networks Centers Centers Residential Leverage infrastructure assets CO power, cooling and space at a premium vs. core/OTT DCs Deliv
Slide 9: CE Packet Processing
CE Packet Processing Metro and Access NID Access NID Network PE Core PE Network Networks Customer Customer Site #2 Site #1 CSP Tunnel E-Line Service • Service classification • Tunnel processing - include: • Identify service/flow based on multiple fields • Provider Bridging - VLAN stacking from Ether
Slide 10: Traffic Management – Hierarchical QoS
Traffic Management – Hierarchical QoS Access Networks (low cost, small buffers) Metro and Central Core Office Networks Egress bandwidth profile- Maintain per service SLAs Prioritized scheduling for bandwidth and latency Shaping (i.e rate limit) at multiple layers Prevent packet loss from access netw
Slide 11: Today's Carrier Ethernet at the Edge
Today’s Carrier Ethernet at the Edge: Carrier Ethernet Switch Router (CESR) Vertically Integrated Proprietary CESRs Typical CESR general attributes: • Proprietary systems can lead to vendor lock-in • Control plane is integrated, proprietary software • Data plane mostly proprietary ASICs and NPUs • F
Slide 12: HW Acceleration: Efficiency at the Edge
HW Acceleration: Efficiency at the Edge Hardware Performance = Efficiency Edge Programmability = Agility to Match Virtual Software Hardware Acceleration Scalability = Address Various Deployment Sizes COTS = High Volume Semiconductor Economics 3/18/2018 11
Slide 13: Multi-core CPU/NPU Acceleration Option
Multi-core CPU/NPU Acceleration Option Core Core Core Core Core Core Core Core Proprietary NPU cores Core Core Core Core Configurable, but accelerate Fixed Hardware specific functions - Limited flexibility Acceleration Proprietary, relatively low volume COTS 3/18/2018 12
Slide 14: FPGAs – True Programmable Hardware
FPGAs – True Programmable Hardware vBRAS or vEPC IPsec SD-WAN OVS & VxLAN Offload 10G CE Carrier Ethernet Scalable Across Price/Performance FPGA = Programmable Logic Array Flexible SoC Resource Allocation (e.g. uCPE NIC w/CE NID) (e.g. SmartNIC w/CE+VNF) COTS Platform with Broad Applications Beyond
Slide 15: FPGA Adoption in Cloud Data Centers
FPGA Adoption in Cloud Data Centers Azure SmartNIC with FPGA acceleration Network infrastructure uses: ▪ Packet processing ▪ Traffic management ▪ Crypto “…allows hosts to use FPGAs for acceleration with low latency… improving the economics of the accelerator deployment, as hosts running services tha
Slide 16: Ethernity's Carrier Ethernet and VNF offload
Ethernity’s Carrier Ethernet and VNF offload Telecom based SmartNIC – 40G/100G • Packet Processing Features • MEF 2.0 Carrier Ethernet services • L3 including IPv4, IPv6 & MPLS • Tunnels, including VxLAN, GRE • Virtual routers • NAT & NAPT • TCP/UDP stateless offloads • Dynamic load balancing • 1M f
Slide 17: Integrated Carrier Ethernet Switch and Aggregation as NFVi
Integrated Carrier Ethernet Switch and Aggregation as NFVi • Carrier Ethernet features including: • H-QoS • Counters vBRAS vRouter vFW VNF3 • Classification • External buffering • L2/L3/MPLS/Overlay OVS-DPDK forwarding • Forwarding options • SW fast data path send data to relevant card Virtual Outpu
Slide 18: Building VNFs on Carrier Ethernet & HW Acceleration
Building VNFs on Carrier Ethernet & HW Acceleration • CE provides connectivity guarantees & SLAs • Without CE guarantees, QoS of higher level VNF services is indeterminate • MEF is currently extending CE-like service definitions to higher level VNF services • Reuse instead of reinvent • Once Carrier
Slide 19: Opportunities for Hardware Acceleration
Opportunities for Hardware Acceleration • L3-L4 packet processing & traffic management • Including: vBRAS, vFW, NAT, vBNG • Deep packet inspection for SD-WAN & security monitoring • Load balancing across scale out infrastructure • Service monitoring – higher level OAM • Cryptography such as IPSec •
Slide 20: VNF Acceleration Example: vEPC
VNF Acceleration Example: vEPC Ideal Solution for MEC virtual Evolved Packet Core (vEPC) S6a MME HSS IMS S11 S1-U S5 SGi S-GW P-GW Intranet LTE/5G eNB Carrier Ethernet Transport Internet Offload CE transport 3GPP protocol from CPU to Ethernity ACE-NIC 1 millions classified flows and 10 million count
Slide 21: Turbocharge Edge DC with ACE-NIC
Turbocharge Edge DC with ACE-NIC CORD CO Architecture Edge Server vBRAS/vBNG VM CG-NAT L3 Forwarding Carrier Ethernet Transport ACE-NIC ACE-NIC ACE-NIC ACE-NIC ACE-NIC ACE-NIC • Accelerated VNF Platform • Up to 6 NICs per server – 240Gbps • Scale out with multiple servers for multi-terabit performan
Slide 22: Summary: VNFs need FPGA acceleration
Summary: VNFs need FPGA acceleration • Edge amplifies need for efficiency & drives need for HW acceleration • Carrier Ethernet is a logical foundation for NFV infrastructure (NFVi) for Carrier Grade NFV platforms • FPGAs - the industry choice for programmable, COTS hardware • Ethernity - FPGA & CE e
Slide 23: Thank you for attending!
Thank you for attending! • Q&A • Related whitepaper: http://www.ethernitynet.com/wp-content/uploads/2018/01/WP-Virtual-Carrier- Ethernet-on-COTS.pdf • Live Demo at MWC 2018 • Contacting Ethernity: sales@ethernitynet.com 22