Update on the FI Testbed Activities in Korea in Korea
Sunhee Yang@ETRI Sunhee Yang@ETRI, Sangkyun Kim@NIA, Dongkyun Kjm@KISTI
APAN FI Testbed Workshop
- Aug. 24, 2011
Update on the FI Testbed Activities in Korea in Korea Sunhee - - PowerPoint PPT Presentation
Update on the FI Testbed Activities in Korea in Korea Sunhee Yang@ETRI Sunhee Yang@ETRI, Sangkyun Kim@NIA, Dongkyun Kjm@KISTI APAN FI Testbed Workshop Aug. 24, 2011 Contents R&D activities for K-FIT R&D activities for K FIT
APAN FI Testbed Workshop
3
(미래를 대비한 인터넷 발전계획, ‘11.6.29., KCC)
Smartphone
4 Wired/Wireless Traffic Investment shrinking of Network Provider
(Evolutionary tech.)
Greater Mobility
(Innovative architecture)
Quality of Service Seamless Service Smart Service Platform y Contents Centric Network Security for a National SoC Cloud Computing Archi. Polymorphic Network
y p
Innovation of Industry & Technology on the Testbed
5
dea &D
lobal Collaboration
raining roducts esults
Programmable
Virtualized FI Node
L1/L2 unified switch
ex2) Contents Centric Network
User Defined Node Clearing House
ex1) Future Mobile Network Virtual
Aggregated Manager
Fed ICT Control Layer Infrastructure eration Control Layer Physical Infrastructure
Virtualized L2 Transport (Carrier Ethernet) Optical 6 Virtualized Radio Access(CR) Optical Transport
FIRST Platform (’09 ’13) – FIRST Platform ( 09- 13)
– FiRSTProNET FiRSTProNET
– PC-based high-performance software router platform – Accelerates the core packet processing with Graphics Processing Unit(GPUs)
– Commodity HW-based high-performance virtual router platform – Virtualization based on Xen and QoS guaranteed I/O Virtualization
– Unified virtual network control @ Yonsei Univ. – OpenFlow related activies @ GIST, CNU….
7
p @
8
8
– COTS (Commercial off-the-shelf) blades and ATCA standard IF Octeon NP (10G) based data plane – Octeon NP (10G) based data plane
9
9
– Hardware(Oecteon NP , x86), VMM(PacketvisorVMM, Xen), VM – Packetvisor: PacketvisorVMM + PacketvisorWorks
10
10
– Data path handling and control software – Stablized performance depending on # of cores, size of flow table and size of packet
11
11
– Network control mechanisms and sub systems for secure and independently shared virtual network – ICT Resource Management
y, , g,
– Virtual Network(slice) and virtual node(sliver) control – User/Project Management User/Project Management – Authentication, Authorization. etc. – Virtual Resource Federation
12
12
– ‘10: ProtoGENI adapter based on Java/Perl – SSL, XMLRPC, MySQL, Mobile User UI on Andriod Phone – ‘11: Expanding for Global Testbed Federation
13
13
– Accelerates the core packet processing with Graphics Processing Unit(GPUs) – Offloads computation and memory-intensive router applications to GPUs – scalable with # of multi-core CPUs, GPUs, and high-speed NIC – 40Gbps IP packet processing on 8-core Nehalem server – 40Gbps IP packet processing on 8-core Nehalem server – Supports IPv4, IPv6, OpenFlow, and Ipsec
– Control Plane Integration – Opportunistic offloading w/ CPU and GPU – Multi-fuctional, modular programming environment Multi fuctional, modular programming environment
14
Ref: “PacketShader 2.0: Design Considerations”, FIF Testbed WG 3rd meeting
Virtual Router for Future Internet Architecture PC HW Platform and virtualization based on Xen PC HW Platform and virtualization based on Xen Single Root I/O Virtualization – QoS guaranteed I/O Virtualization Rigid Network isolation and BW control g – VLAN based network isolation and Backward Congestion Notification
Guest Domain 2 Guest Domain 1 V-Channel V-Channel
VNIC1 VNIC2 VNIC3
Hardware (SR-IOV Support)
VNIC1 VNIC2 VNIC3
Virtual Machine Monitor (Xen) NIC1 NIC2 NIC3 Hardware (SR IOV Support)
Vi t l R t (X b )
15
Virtual Router (Xebra)
Ref: “Xebra: Commodity HW based Virtual Router”, FIF Testbed WG 3rd meeting
16
17
100G 18
ROADM L2 (Ethernet) S/W
estbed Deployment and Experiments
19
Facility #4 Multiple Virtual Network (DCN OpenFlow UCS FIRST) Facility #4 Multiple Virtual Network (DCN, OpenFlow, UCS, FIRST) Multiple Virtual Network Control Platform Facility #3 Cloud research and development environment Facility #2 CCN research and development environment Facility #3 Cloud research and development environment Facility #1 POTS Control Platform MPLS-PT & ROADM Integration – POTS Experiment KOREN Ethernet & Lamda Busan Seoul Daejeon
20
Ad d A li ti d C ll b ti N t k ith 14 M b C t i i A i N th
America, Europe, and Africa
21
22
23
EU JP APII KOREN EU, Asia, etc. US GENI K-GENI KOREN & KREONET Backbone (~100G) KOREN GLORIAD & GLIF NetFPGA Productio n Internet Korea Cleainghouses + dvNOC (Meta Operations) KRLight GOLE Asia, South A i Af i GLORIAD & GLIF FIRST@ETRI & NetFPGA OpenFlow OF@KREONET NetFPGA OpenFlow America, Africa, etc. @ Openflow Core Nodes OF@KREONET, Further Testbeds NetFPGA OpenFlow FIRST@ETRI Access Node NetOpen/ MediaX ORBIT Control Framework First@PC Accses Node C A (CNU GIST
NetFPGA FIRST@ETRI Control Framework
FIRST@ETRI/FI Access Nodes FIRST@ETRI/FI NetFPGA OpenFlow Campus-A (CNU, GIST KHU, Postech, … )
24
Site-A (ETRI) Campus-B (KAIST, SNU, … ) Site-B (Wired/Wireless Labs) FIRST@ETRI/FI Control Framework(s) PlanetLab/FI Control Framework(s) Originated by Myung-Ki Shin@ETRI, Reorganized by Dongkyun Kim@KISTI
25
26