geni meso scale prototype for mobilityfirst evaluation
play

GENI Meso scale Prototype for MobilityFirst Evaluation May 14 th , - PowerPoint PPT Presentation

GENI Meso scale Prototype for MobilityFirst Evaluation May 14 th , 2012 WINLAB Ivan Seskar 1 Objectives Software prototypes of main arch. pieces with end to end protocol implementations Consider OpenFlow/SDN, FPGA platforms,


  1. GENI Meso ‐ scale Prototype for MobilityFirst Evaluation May 14 th , 2012 WINLAB Ivan Seskar 1

  2. Objectives  Software prototypes of main arch. pieces with end ‐ to ‐ end protocol implementations  Consider OpenFlow/SDN, FPGA platforms, and optical components  Mobility, first  Mobile host platforms and multiple wireless access technologies  Ad hoc, disconnection, network mobility as common cases  Meaningful at ‐ scale evaluation  Realistic network conditions  Access ‐ Core ‐ Access evaluations  Target long running deployment and opt ‐ in users WINLAB 2

  3. Prototyping and Evaluation: Execution Summary Phase 2 Phase 3 Phase 1 Content Host/Device Addressi Context Addressing ng Stack Addressi Stack ng Stack Encoding/Certifying Layer Global Name Resolution Service (GNRS) Storage Aware Locator ‐ X Routing Routing (e.g., GUID ‐ based) Context ‐ Aware / Late ‐ bind Routing Prototype Integrated MF Protocol Standalone Modules Deployable s/w pkg., box Stack and Services Evaluation Distributed Testbed Simulation and Emulation Smaller Scale Testbed E.g. ‘Live’ on GENI WINLAB 3

  4. Prototype: Click ‐ based Router Early Dev. Inter ‐ Domain Locality ‐ Aware R3 DNS Integrate PacketCloud Framework DMap – DiHT GSTAR User ‐ level Name Compute Content Routing Mgmt. Resolution Services Cache Service Processes Host Rx Q Host Tx Q Click Rsrc To Next ‐ hop To/From Host Forwarding Engine Lookup Control Wired and wireless i/f Wired and wireless i/f Forwarding Table Packet Block Service Block Classifier Aggregator Classifier Segmentor Rx Q Tx Q Next ‐ hop Look up Hold buffer x86 hardware and runtime WINLAB 4

  5. Prototype: Host Protocol Stack and Network API Linux PC/laptop with WiMAX & WiFi ‘Socket’ API App ‐ 1 App ‐ 2 App ‐ 3 open send Context API send_to recv Network API recv_from Context Services close E2E Transport GUID Services Sensors Android device with WiMAX & WiFi Network Layer Security Routing User policies Interface Manager ‘Hop’ Link Transport WiFi WiMAX Device: HTC Evo 4G, Android v2.3 ( rooted), NDK (C++ dev) Early Dev. WINLAB 5

  6. GENI Deployment & Evaluation Legend Internet 2 National Lambda Rail OpenFlow Backbones OpenFlow WiMAX ShadowNet MobilityFirst Router & GNRS Servers Mobile Hosts Static Hosts Deployment Goals Mapping onto GENI Infrastructure • Large scale, multi ‐ site (ProtoGENI nodes, OpenFlow switches, • Mobility centric GENI Racks, DieselNET buses, • Realistic, live WiMAX/outdoor ORBIT nodes) WINLAB 6

  7. GENI Deployment: GEC ‐ 12 (Nov. ’11) Demo Topology • Storage aware routing • Name resolution • Multihoming BBN Cambridge, MA ProtoGENI Backbone WiFi AP WiMAX BSS Mesoscale WiMAX MF Router + Name Resolution Server and OF Android Client w/ WiMAX + WiFi Linux PC/laptop w/ WiMAX + WiFi Vehicular node w/ WiMAX WINLAB N. Brunswick, NJ WINLAB 7

  8. GENI Deployment: Physical Topology W ashington W isconsin NLR BBN Seattle I ndiana NLR I 2 Denver NLR New York NLR SUNW Rutgers Chicago Clem son I 2 W ashington I 2 Georgia Tech. Los Angeles I 2 Stanford Atlanta I 2 Houston I 2 OF Sw itch VLAN 3 7 1 5 NLR OF Sw itch VLAN 3 7 1 6 Edge OF Sw itch WINLAB

  9. GENI Deployment: Mapping to Logical Topology W ashington W isconsin NLR BBN Seattle pg49@Stanford pg33@GTech pc8@BBN pg51@Rutgers I ndiana WiFi AP WiFi AP NLR I 2 Denver GUID=3 GUID=5 GUID=1 NLR New York NLR SUNW Rutgers GUID=6 Chicago pg47@ Bridge Clem son I 2 Clemson W ashington GUID=2 GUID=4 GUID=7 I 2 Georgia Tech. Los Angeles I 2 Stanford Atlanta PG1@Clemson WiMAX BTS pg50@Rutgers pc4@BBN WiMAX BTS I 2 Houston I 2 OF Sw itch VLAN 3 7 1 5 NLR OF Sw itch VLAN 3 7 1 6 Edge OF Sw itch WINLAB

  10. Application: Content Delivery to Mobile Hosts DATA NA Content Subscriber DATA DATA GUID=3 GUID=5 WiFi AP WiFi AP R7 Bridge DATA GUID=6 GUID=1 DATA # 9 DATA GUID=7 GUID & SID GUID=2 GUID=201 DATA GUID=4 GUID=101 WiMAX BTS Content WiMAX BTS Publisher BBN Wireless Edge Rutgers Wireless Edge NLR path using VLANs 3716, 3799 (Clemson) I2 path using VLANs 3715, 3745(BBN), 3798 (Clemson) ProtoGENI host running MF Router WINLAB 10

  11. Visualization Data collection framework with API, monitors, filters and data warehouse E.g., Orbit Measurement Library (OML) MF Network element e.g. Router Network State Repository NRS Web Server Monitor and filter Click HTTP , XML, JSON Runtime/ OS Browser: AJAX/ JS/ Flash W hat’s on? 1. Network statistics 2. Packet and flow tracing 3. Routing events 4. Application events Network map credits: ProtoGENI’s Flack WINLAB 11 tool. http: / / protogeni.net/ trac/ protogeni

  12. Next Steps • Applications, applications, applications • Deploy and evaluate GNRS services across full GENI backbone and Spiral 4 GENI racks (~80 ‐ 100 nodes) • Deploy MobilityFirst services in overlay mode over the commodity internet • Even larger scale with emulation co ‐ deployment • Multi ‐ domain deployment using world ‐ wide GENI collaborators WINLAB 12

Recommend


More recommend