Kansei Genie Genification to Orcafication Anish Arora Hongwei

  • Slides: 8
Download presentation
Kansei. Genie: Geni-fication to Orca-fication Anish Arora, Hongwei Zhang, Rajiv Ramnath, Mukundan Sridharan, Wenjie

Kansei. Genie: Geni-fication to Orca-fication Anish Arora, Hongwei Zhang, Rajiv Ramnath, Mukundan Sridharan, Wenjie Zeng, Xi Ju July 2, 2009

Kansei Architecture © 2004 - mid 2008 Researcher Web Interface Web Server Experiment configuration

Kansei Architecture © 2004 - mid 2008 Researcher Web Interface Web Server Experiment configuration / Output logs Experiment DB Policy Verification/ Deployer Kansei Scheduler Measurement Manager (Chowkidar) Health Status Hardware Array File upload, Execution and Clean up

GENI-fication: Kansei. Genie Architecture © Q 1 -Q 2 Researcher Portal Web Service Layer

GENI-fication: Kansei. Genie Architecture © Q 1 -Q 2 Researcher Portal Web Service Layer RSpec adv/ Slice creation Kansei. Genie Policy Authority Kansei. Genie Aggregate of Aggregate Manager Ticket honoring/ Slice instantiation/ Management calls Slice Manager Measurement Manager (Chowkidar) Researcher Portal calls Experiment Manager O&M Manager File upload, Execution and Clean up Hardware Array Legend: to be implemented already implemented

Kansei. Genie Orca Integration Architecture/Status ORCA Broker Policy Plugin RSpec Repository ORCA Site Authority

Kansei. Genie Orca Integration Architecture/Status ORCA Broker Policy Plugin RSpec Repository ORCA Site Authority ORCA Slice Controller Handlers for slice setup/teardown Event Handlers Ticket request/grant Researcher Portal Resource request (needs modification) Web Services Slice setup/ tear down Kansei. Genie Aggregate of Aggregate Manager File upload, Execution, Viz Kansei. Genie Slice Manager Kansei. Genie Experiment Manager Kansei. Genie Measurement Manager Kansei. Genie O&M Manager Hardware Arrays Secure SOAP XML-RPC Legend: to be implemented already implemented

Kansei. Genie Experiments by Outside Researchers • Kansei Neteye § UCLA Michigan State §

Kansei. Genie Experiments by Outside Researchers • Kansei Neteye § UCLA Michigan State § Northwestern USC § UT, Dallas City University of Hong Kong § ICT, Australia University of Sci. & Tech. , China § Michigan State Southeastern University, China § Wayne State § SUNY, Buffalo • Several groups at OSU use Kansei for research & education • We are motivated by our own federation scenarios: § Multi-Fabric Sensing (seamless tracking, tripwire) § Seamless Regression Testing § Portable Array and Stationary Array § Fabric via Cloud 5

Challenges in Spiral 1 • Motivation for engaging external partners in GENI § as

Challenges in Spiral 1 • Motivation for engaging external partners in GENI § as opposed to using/replicating our testbeds § likewise in cross-GENI partner experiments • Budget overly constrained § science education projects are of interest to Spiral 2/3 § key WSN programmability, network virtualization, and I&M needs dependent on Solicitation 2 proposal § Java Sun. SPOT and. NET Imote 2 substrates still of interest § (university operations costs high: 60 K @ warehouse alone!) • Some effort in GENI-fication and ORCA-fication § Jboss learning curve § ORCA-fication benefited from help we’ve received 6

Internet 2 Connection • OSU (via OARnet) § OIT at OSU has fibre running

Internet 2 Connection • OSU (via OARnet) § OIT at OSU has fibre running to our testbed buildings § VLAN connections involve tap into POP at Chicago (Omni. POP) § OIT still checking on VLAN costs § L 3 node costs $235 per month - not easy for OSU to pick up cost • WSU (via MERIT) § L 3 connection available: no cost § L 2 costs unaffordably high: ~30 K/year • End-to-end GENI experiments eventually L 2, but L 3 should suffice for us for Year 2 7

Spiral 2 Plans 1. Port v 1. 0 Kansei Genie Installer on Net. Eye

Spiral 2 Plans 1. Port v 1. 0 Kansei Genie Installer on Net. Eye § accommodate new substrate § port hierarchical AM/CM to new substrate 2. Experiment Interaction User Service v 1. 0 § GENI-fication of Kansei researcher portal 3. Basic federated resource discovery, embedding, and scheduling § jointly with ORCA-Ben and Cluster D § authentication across brokers for federated slices (ORCA chaining mechanism/policy extensions? ) § federated slice discovery, embedding & scheduling across sites 4. Realize v 2. 0 of software installer package, incl. basic federation 5. Support open use of federation use cases 8