Rethinking Access Networks with High Performance Virtual Software BRASes Roberto Bifulco, Thomas Dietz, Felipe Huici, Mohamed Ahmed, Joao Martins, Saverio.

Slides:



Advertisements
Similar presentations
Building Fast, Flexible Virtual Networks on Commodity Hardware Nick Feamster Georgia Tech Trellis: A Platform for Building Flexible, Fast Virtual Networks.
Advertisements

1 Building a Fast, Virtualized Data Plane with Programmable Hardware Bilal Anwer Nick Feamster.
Enabling Fast, Dynamic Network Processing with ClickOS
Chapter 1: Introduction to Scaling Networks
All Rights Reserved © Alcatel-Lucent 2009 Enhancing Dynamic Cloud-based Services using Network Virtualization F. Hao, T.V. Lakshman, Sarit Mukherjee, H.
Virtual Switching Without a Hypervisor for a More Secure Cloud Xin Jin Princeton University Joint work with Eric Keller(UPenn) and Jennifer Rexford(Princeton)
Layer 3 Switching. Routers vs Layer 3 Switches Both forward on the basis of IP addresses But Layer 3 switches are faster and cheaper However, Layer 3.
Muse confidential Service Rich Access Networks: The Service Plane Solution Edith Gilon – de Lumley Bell Labs R&I, Alcatel-Lucent BroadBand Europe Antwerp,
L. Alchaal & al. Page Offering a Multicast Delivery Service in a Programmable Secure IP VPN Environment Lina ALCHAAL Netcelo S.A., Echirolles INRIA.
Network Virtualization Overlay Control Protocol Requirements draft-kreeger-nvo3-overlay-cp-00 Lawrence Kreeger, Dinesh Dutt, Thomas Narten, David Black,
Implementing Inter-VLAN Routing
IPv6 over xDSL: The DIODOS Proposal Athanassios Liakopoulos Greek Research & Technology Network International IPv6 Workshop, Kopaonik,
Performance Evaluation of Open Virtual Routers M.Siraj Rathore
Multimedia over DSL By Phil Moy. May 14, Agenda n DSL Forum Working Text 80 - Multiservice Architecture & Framework Requirements n DSL Forum Working.
DSL Access Architectures and Protocols. xDSL Architecture.
Novell Server Linux vs. windows server 2008 By: Gabe Miller.
1 In VINI Veritas: Realistic and Controlled Network Experimentation Jennifer Rexford with Andy Bavier, Nick Feamster, Mark Huang, and Larry Peterson
Efficient IP-Address Lookup with a Shared Forwarding Table for Multiple Virtual Routers Author: Jing Fu, Jennifer Rexford Publisher: ACM CoNEXT 2008 Presenter:
Introduction An introduction to the software and organization of the Internet Lab.
CISCO NETWORKING ACADEMY Chabot College ELEC Router Introduction.
Hash, Don’t Cache: Fast Packet Forwarding for Enterprise Edge Routers Minlan Yu Princeton University Joint work with Jennifer.
Data Center Virtualization: VirtualWire Hakim Weatherspoon Assistant Professor, Dept of Computer Science CS 5413: High Performance Systems and Networking.
1 MASTERING (VIRTUAL) NETWORKS A Case Study of Virtualizing Internet Lab Avin Chen Borokhovich Michael Goldfeld Arik.
Jennifer Rexford Princeton University MW 11:00am-12:20pm Programmable Data Planes COS 597E: Software Defined Networking.
Enabling Innovation Inside the Network Jennifer Rexford Princeton University
Networking Virtualization Using FPGAs Russell Tessier, Deepak Unnikrishnan, Dong Yin, and Lixin Gao Reconfigurable Computing Group Department of Electrical.
PRESENTATION ON BROADBAND NETWORKING SUBMITTED BY:- DEEPALI SHUKLA MASHARRAT ASHFAQUE YUVRAJ BHUWAL ASHISH KUMAR PATEL ARSHAD ALI BSNL VOCATIONAL TRAINING.
Hosting Virtual Networks on Commodity Hardware VINI Summer Camp.
(1) Univ. of Rome Tor Vergata, (2) Consortium GARR, (3) CREATE-NET
PREPARED BY :-  HIMANSHU MINZ  VIKAS UPADHYAY VOCATIONAL TRAINING AT BSNL,DURG BATCH 4 SESSION 2014.
1 IP Forwarding Relates to Lab 3. Covers the principles of end-to-end datagram delivery in IP networks.
Improving Network I/O Virtualization for Cloud Computing.
Corporate Firewalls and DMZs By Matt Bertram ISQS 6342 (Spring 2003) Professor John Durrett.
Politecnico di Torino Dipartimento di Automatica ed Informatica TORSEC Group Performance of Xen’s Secured Virtual Networks Emanuele Cesena Paolo Carlo.
A study of introduction of the virtualization technology into operator consoles T.Ohata, M.Ishii / SPring-8 ICALEPCS 2005, October 10-14, 2005 Geneva,
Processes and OS basics. RHS – SOC 2 OS Basics An Operating System (OS) is essentially an abstraction of a computer As a user or programmer, I do not.
A Measurement Based Memory Performance Evaluation of High Throughput Servers Garba Isa Yau Department of Computer Engineering King Fahd University of Petroleum.
Shivkumar Kalyanaraman Rensselaer Polytechnic Institute 1 Based upon slides from Jay Lepreau, Utah Emulab Introduction Shiv Kalyanaraman
Network Virtualization Sandip Chakraborty. In routing table we keep both the next hop IP (gateway) as well as the default interface. Why do we require.
Reactive Logic in Software-Defined Networking: Measuring Flow-Table Requirements Maurizio Dusi*, Roberto Bifulco*, Francesco Gringoli”, Fabian Schneider*
: MobileIP. : r Goal: Allow machines to roam around and maintain IP connectivity r Problem: IP addresses => location m This is important for efficient.
Coping with Link Failures in Centralized Control Plane Architecture Maulik Desai, Thyagarajan Nandagopal.
Integrating base stations with a software- defined core network Bruno Hernández Zamora.
ClickOS and the Art of Network Function Virtualization Joao Martins*, Mohamed Ahmed*, Costin Raiciu§, Roberto Bifulco*, Vladimir Olteanu§, Michio Honda*,
Slide 1/12 Network Function Virtualization and its Dependability Challenges Relevant papers: 1.Gember-Jacobson, Aaron, Raajay Viswanathan, Chaithan Prakash,
1 © 2004, Cisco Systems, Inc. All rights reserved. CCNA 2 v3.1 Module 2 Introduction to Routers.
On-the-Fly TCP Acceleration with Miniproxy Giuseppe Siracusano 12, Roberto Bifulco 1, Simon Kuenzer 1, Stefano Salsano 2, Nicola Blefari Melazzi 2, Felipe.
NFP: Enabling Network Function Parallelism in NFV
INTERNET PROTOCOL TELEVISION (IP-TV)
NFV Group Report --Network Functions Virtualization LIU XU →
IEEE ICC ‘16 Dynamic M2M Device Attachment and Redirection in Virtual Home Gateway Environments Apostolos Papageorgiou, NEC Labs Europe Roberto Bifulco,
Ready-to-Deploy Service Function Chaining for Mobile Networks
NFV Compute Acceleration APIs and Evaluation
Instructor Materials Chapter 1: LAN Design
Introduction An introduction to the software and organization of the Internet Lab.
Heitor Moraes, Marcos Vieira, Italo Cunha, Dorgival Guedes
6WIND MWC IPsec Demo Scalable Virtual IPsec Aggregation with DPDK for Road Warriors and Branch Offices Changed original subtitle. Original subtitle:
INTERNET PROTOCOL TELEVISION (IP-TV)
NFP: Enabling Network Function Parallelism in NFV
Introduction An introduction to the software and organization of the Internet Lab.
Broadband Components Customer Premises Equipment(CPE)
Network Function Virtualization: Challenges and
PPPoE Internet Point to Point Protocol over Ethernet
NFP: Enabling Network Function Parallelism in NFV
YANG data model of Control-Plane and User-Plane Separation BNG
Lecture9: Embedded Network Operating System: cisco IOS
Introduction An introduction to the organization of the Networks Course and the Internet Lab.
Lecture9: Embedded Network Operating System: cisco IOS
A Closer Look at NFV Execution Models
Presentation transcript:

Rethinking Access Networks with High Performance Virtual Software BRASes Roberto Bifulco, Thomas Dietz, Felipe Huici, Mohamed Ahmed, Joao Martins, Saverio Niccolini, Hans-Joerg Kolbe NEC Europe Ltd., NEC Laboratories Europe Heidelberg, Germany

Acknowledgements ▐NEC Laboratories Europe, Heidelberg, Germany ▐The work has been partially funded by the European FP-7 projects

HGW subscribers core network (IP) aggregation network DSLAM Internet EDGE ROUTER DC services IP PPP PPPoE Ethernet+VLAN SDH/PDH (DSL) IP Ethernet SDH, WDM, …. BRAS Broadband Access Network

 PPPoE/PPP sessions management  Header encapsulation/decapsulation  Interface to AAA  Traffic shaping  Access control  ARP proxy  IP Forwarding  IP addresses assignment  IGMP processing

Broadband Access Network  PPPoE/PPP sessions management  Header encapsulation/decapsulation  Interface to AAA  Traffic shaping  Access control  ARP proxy  IP Forwarding  IP addresses assignment  IGMP processing  Expensive  Monolithic architecture  Single point of failure  Hard to change

Network Function Virtualization ▐Decouple Network Functions from the hardware that provides them! VM

▐From Hardware to Software Do we need to maintain the same monolithic approach? ▐Server virtualization technologies are IT-oriented Network functions require optimized I/O Network Function Virtualization

▐From Hardware to Software Do we need to maintain the same monolithic approach? ▐Server virtualization technologies are IT-oriented Network functions require optimized I/O Network Function Virtualization Fast SOFTWARE VSWITCH Fast I/O Tiny, Fast VMs

Modular Network Functions ▐Exploit successful approaches: The Click Modular Router Element

Putting all together: ClickOS © NEC Corporation 2013 Page 10 [1] Netmap:

WTF? What’s The Function?

BRAS on ClickOS

EVALUATION An early performance

ClickOS: VNF instantiation ▐Cut of VM boot time down to less than 30 msec Up to 200 msec when running 400 VMs on a single PC ▐Instantaneous processing of middlebox functions ~20 msec when several VMs are already running © NEC Corporation 2013 Page 14 all measurements made on a single PC, Intel i7, 3,2GHz, 6 cores, 4GB RAM

Networking performance with 128 VMs ▐Line rate (10 Gbps) for all packet sizes (256 bytes and greater) 4.8 Mpps for 64 bytes packets © NEC Corporation 2013 Page 15 all measurements made on a single PC, Intel i7, 3,2GHz, 6 cores, 4GB RAM

▐Session setup rates Reached >1000 sessions/s Current BRAS technology reaches about 300s/s Same result with multiple (up to 25) VMs on the same server Not able to measure further because of limitations of test equipment PPP Control Plane © NEC Corporation 2013 Page 16 all measurements made on a single PC, Intel i7, 3,2GHz, 6 cores, 4GB RAM * ACK: Thanks to Xantaro Gmbh and Carsten Michel for the test-bed setup

Evaluation (summary) ▐Linux VM performance (without any network function) ~1% of 10Gbps for 256B packets ~1-10 sec boot time several MB memory footprint (hundreds..) ▐ClickOS performance 100% of 10Gbps for all packet sizes bigger than 256B (100x faster) ~30 msec boot time (30x faster) 5 MB memory footprint (at least 10x smaller) Able to run hundreds VMs on a single PC ▐PPP control plane Session establishment rate: >1000 sessions/s Memory requirement: 1MB to support 1000 sessions ▐PPP data plane 100% of 10Gbps for all packet sizes bigger than 512B

Conclusion ▐We introduced ClickOS, an innovative technology to build VNF ▐We built a BRAS VNF based on ClickOS ▐We provided a early performance evaluation of our prototype 100% of 10Gbps for all packet sizes bigger than 256B for plain ClickOS 100% of 10Gbps for all packet sizes bigger than 512B for PPP data plane ~30 msec boot time (30x faster) 5 MB memory footprint (at least 10x smaller) Able to run hundreds VMs on a single PC Session establishment rate: >1000 sessions/s Memory requirement: 1MB to support 1000 sessions ▐…▐…

Future work ▐We shown how to decompose network functions, in order to enable a new way of composing them again!! HGW Subscribers DSLAM PPPoE IPoE Controller PPP termination PPP decapsulation ▐An enabler to build Service Chains ▐There is still room for performance optimizations ▐Standard activities ETSI NFV, …