Creating a Climate for Innovation on Internet2 - Eric Boyd Senior Director, Strategic Projects


Published on

Internet2 Innovative Application Awards

Published in: Education, Technology, Business
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide
  • Internet2 IS an unparalleled human network.The Internet2 human network includes over 350 member institutions—leading universities, corporations, government research agencies and not-for-profit networking organizations. Membership is structured in four levels across four institutional categories.[>>] The broader Internet2 community includes over 66,000 public institutions across the U.S. connected to the Internet2 Network through 39 state and regional education networks—from primary and secondary schools, to libraries and colleges, to healthcare organizations and museums.[>>] Initiatives, working groups and special interest groups are led by innovators from the domains of science and academic research, arts and humanities and health sciences—as well as advanced network researchers and developers.[>>] Internet2 partners with over 50 national research and education networks enabling connectivity to over 100 international networks.[>>] Internet2’s core staff joins forces with these communities, providing whatever support is needed to ensure that community ideas and innovations bring the expected results.images:portion of Muse connectivity mapFMM 2002portion of Int’l Partners mapAnn Doyle at TERENA Networking Conference, Catania (Sicily) Italy, 2006
  • Creating new innovation opportunities begins with understanding what enabled innovation in the past.The research and education community played a seminal role in the creation of the modern Internet and the applications that have made it the mosttransformative technology of the 20th and 21st centuries34 of these leading universities created Internet2 in 1996 [bit more on why/mission? - to make sure that innovation cd continue unfettered by commodity Internet constraints]
  • [This slide stays on the background without the two builds for opening comments.]Good Morning!This morning we have an extraordinary panel discussing the future of the Internet and key innovations that the research and education community may consider as we move to continuously position the R&E community in to the future.Before we get started, we wanted to take a few minutes and share with you an update on some key projects and programs that have come to fruition since we met just 12 months ago in Virginia.[Trigger first build] Our infrastructure programs continue to build on the nearly $1.5B federal investment in advanced networks through ARRA and the nearly $100M that the Internet2 community, NTIA and key partners like Juniper and Ciena invested in the Internet2 network. Now with 18 months of solid production experience, our 100G, nationwide 18,500 mile SDN-based network continues to provide the basis of our national innovation strategy for research and education[trigger second build] In the past 18 months, we have seen dozens of new high performance compute clusters connect to the network, nearly 100 campuses and regionals upgrade connections to the network and thousands of new SDN-enabled end points capabilities deployed throughout the community and our corporate partners.
  • Focus on virtualization in this talk, but integrating the network with compute and storage is equally important.
  • Creating a Climate for Innovation on Internet2 - Eric Boyd Senior Director, Strategic Projects

    2. 2. May 18, 2014 © 2013 Internet2 [ 2 ] 2 – 5/21/2014, © 2012 Internet2 2 – © 2012 Internet2 Unleashing new waves of global discovery, together.
    3. 3. • Internet2 brings together thought leaders from member organizations and the broader research and education community • Our community advances frontiers of network-enabled applications • Our community accelerates innovation and enables transformation The Internet2 Community: An unparalleled human network Nearly 400 member institutions
    4. 4. Internet2 Members and Partners 222 Higher Education members 72 Affiliate members 39 R&E Network members 64 Industry members 65+ Int’l partners reaching over 100 Nations 66,000+ Community anchor institutions Focused on member technology needs since 1996"The idea of being able to collaborate with anybody, anywhere, without constraint…" —Jim Bottum, CIO, Clemson University
    5. 5. May 18, 2014 © 2013 Internet2 • The research and education community played a seminal role in the creation of the modern Internet and the applications that have made it the most transformative technology of the last hundred years • 34 of these leading universities created Internet2 in 1996 Creating new innovation opportunities begins with understanding what enabled innovation in the past
    6. 6. R&E networking ROI has been staggering Total 30-year federal investment to enable the precursors of the Internet is very small compared to the massively successful businesses it sparked. Earth Venus ARPAnet, CSNET & NSFnet < $250 million total investment Contribution of Internet to U.S. economy: $684 billion annually, 4.7% of all economic activity in 2010 (CNNMoney, 2012) Value of Internet to U.S. economy — Employment: $300 billion Payments: $444 billion and Time: $680 billion Total: $1424 billion annually (Harvard Business School/ Hamilton Consultants, 2009) 3963 mi 10,842,768 mi 22,573,248 mi (not to scale) These seminal investments • Put the R&E community ―way out in front‖ of commercial markets • Created a new, bandwidth-rich playing field • Enabled innovations that led to a global transformation: our information-based economy
    7. 7. 7 – 5/21/2014, © 2013 Internet2 * Routers Stanford Computer Workstations Berkeley,Stanford Security Systems UnivofMichigan Security Systems GeorgiaTech Social Media Harvard Network Caching MIT Search Stanford
    8. 8. 8 – 5/21/2014, © 2013 Internet2 How do we create an at-scale Innovation Platform for the next era?
    9. 9. 9 – 5/21/2014, © 2013 Internet2 Innovation Platform vision: Abundant bandwidth Innovation route • Raw capacity now available on Internet2 Network a key imagination enabler • Incent disruptive use of new, advanced capabilities • Promote ―open‖ and creative freedom of use Innovation roadblocks • Limited capacity a major barrier—need more than incremental boosts • Too expensive and risky to try totally new approaches • Closed approaches limit applications or use cases
    10. 10. 10 – 5/21/2014, © 2013 Internet2 Innovation Platform vision: Software-defined networking (SDN) Innovation route • Open up network layer to innovation • Let innovators communicate with and program the network itself • Allow developers to optimize the network for specific apps Innovation roadblocks • Proprietary software in routers and switches • Communications with hardware limited by actual, physical, proprietary components • Application developers have to use the network as prescribed
    11. 11. 11 – 5/21/2014, © 2013 Internet2 Innovation Platform vision: Support for data-intensive science Innovation route • Architect a special solution to allow higher- performance data flows • Include end-to-end performance monitoring • Include SDN server to support programmability Innovation roadblocks • One-size-fits-all approach to network data flows • Lack of transparent performance monitoring solution • No way to customize and optimize the network via SDN
    12. 12. 12 – 5/21/2014, © 2013 Internet2 Innovation Platform Program Pilot Sites • 31 Campuses, 10 Regionals • 76 NSF CC-NIE Awardees • “Operating Innovative Networks” workshops offered in collaboration with ESnet, Indiana, Internet2
    13. 13. Title or Title • Event/Date • Presenter, PresenterTitle, Internet2 What will global innovators do with the next Innovation Platform?
    14. 14. This is what we have been able to say for over a year: The 100G testbed of innovation for tomorrow’s Internet is available nationwide, right now.
    15. 15. • Abundant Bandwidth: • 100G+ Nationwide Backbone • Programmable: • Native OpenFlow w/virtual slices • Support for Data Intensive Science • Interconnected with public Internet • Full 18 months of solid production experience • Open for your innovation! Provides production & innovation platform to: • Dozens of high performance compute clusters • Hundreds of campus data centers • Thousands of SDN ports • Hundreds of wireless access networks • Millions of potential collaborators
    16. 16. • Community is now moving nearly 50 PB/month over Internet2 • Backbone already growing beyond 100G
    17. 17. Does this create a platform for innovation? Abundant bandwidth to enable innovation?  Programmability to encourage application innovation?  Support data intensive science?  TODAY
    18. 18. Does this create a platform for innovation? Abundant bandwidth to enable innovation?  Software-defined networking substrate?  Support data intensive science?  Virtualization? ☐ Integrate network with compute and storage? ☐ TOMORROW
    19. 19. • Simple VLAN Tag based flowspace firewall / proxy • Policy definition and enforcement support range operations – < 1,000 policies to support 3 slices using the entire flowspace • Per slice total rule limits • Per slice per switch flow modification rate limits (planned) • Built upon FloodLight • Designed for production use. Developed by Internet2 with GlobalNOC Software Engineering FlowSpace Firewall
    20. 20. FlowSpace Firewall Config Example <flowspace_firewall> <switch name="foo" dpid="5" flush_rules_on_connect="false" /> <switch name="foo1" dpid="2" flush_rules_on_connect="false" /> <switch name="foo2" dpid="3" flush_rules_on_connect="false" /> <switch name="foo3" dpid="4" flush_rules_on_connect="false" /> <slice name="OESS1”> <switch name="foo" max_flows="10" flow_rate="1"> <port name="s5-eth1"> <range start="1" end="2000"/> </port> <port name="s5-eth2"> <range start="1" end="2000" /> </port> </switch> <controller ip_address="" ssl="false" port="6633" /> </slice> </flowspace_firewall> Symbolic names reduce policy churn limits protect network Range expression for sanity
    21. 21. • Researcher wanting to run on NDDI, iDREAM or AL2S opens ticket with Internet2 NOC • Submitters required to submit a compatibility document – Functionality – Resource requirements • All applications destined for AL2S will be tested on iDREAM lab first. • Testing will be performed in lab running the entire AL2S stack to ensure compatibility Application Vetting Process
    22. 22. • Same hardware as in the the AL2S + others – Brocade, Juniper, Cisco ASR, NEC • Verify Behavior – Does it behave as described in compatibility document – Can it function in a constrained FlowSpace • Verify Interoperability with entire system – When there is network churn or other event does the system as a whole function correctly – How does it handle network life cycle events like new ports or devices added? – Does this app exhibit behaviors that interfere with higher priority apps? • Verify FlowSpace Firewall is adequately protecting resources – Anticipated failure modes are covered – Testing if additional resources need protection – Vulnerabilities not covered in current environment • Oddly formed flowmod crashes switch scenario Lab Testing
    23. 23. • Loss of Forwarding – Forwarding rule triggers bug in Switch – Testable in lab with sufficient test coverage • Loss of Provisioning – A failure in the control plane caused by bug in FSF, OESS, etc – Not stop forwarding on switches, keeps proactive apps working – Reactive apps, start to exhibit loss of Forwarding • Application Flow Space Collisions – 2 Applications interfere with each other’s flowspace – Could be caused by bug in FSF, or possibly in policy defn – Would disrupt forwarding • Multiple Failure – Imagine a case where overlapping flowspace breaks backup paths but backup paths only activated when a trunk goes down Anticipated Failure Modes
    24. 24. • Monitor control channel resources to detect failures • Per Slice and per Slice/Switch – Control Channel Bandwidth – Packet IN/Out – FlowMods/Sec – Total Active Flow Mods – Controller Reconnections – FSF rejections • FlowSpace Firewall (FSF) – CPU / Mem / Running State – Controller Connection Status Monitoring
    25. 25. • Order of Operations 1. Service Restoration 2. Communication 3. Debugging • Each app has an identified owner and a defined priority • Higher priority wins (production will always be higher than research) • Today OESS has the highest priority (only production app) • If a bug in a lower priority service is causing an issue in a higher, we will shut down the lower • If a bug in virtualization is causing an issue in OESS we will revert to non-virtualized operations Incident Response
    26. 26. Does this create a platform for innovation? Abundant bandwidth to enable innovation?  Software-defined networking substrate?  Support data intensive science?  Virtualization?  In progress Integrate network with compute and storage? ☐ Next step
    27. 27. Title or Title • Event/Date • Presenter, PresenterTitle, Internet2 What will global innovators do with the next Innovation Platform?
    28. 28. 31 – © 2013 Internet2 GOLD SILVER BRONZE 2013 Internet2 Innovative Application Awards
    29. 29. Development of next generation, open source applications that are of general interest to the research and education community and take advantage of SDN-enabled capabilities that improve data movement across 100G OpenFlow-enabled networks such as the Internet2 Advanced Layer2 Service [ 32 ] Program Goal May 18, 2014 © 2013 Internet2
    30. 30. • May 2013 – Program Announcement • June 7, 2013 - Proposals Due • June 25, 2013 – Proposal Notification • July 1 - Oct 4, 2013 – Application Development • October 15, 2013 – Submission of Application • November 2013 – Application Demonstrations • December 2013 – Applications available online [ 33 ] Contest Timeline
    31. 31. • Aimed at undergraduate and graduate students at US based colleges and universities • $10,000 cash award – $2,000 for when the proposal is accepted – $8,000 for submission of completed, working project – Individual not institutional awards [ 34 ] Program Summary
    32. 32. • 8 proposals selected for completion • Project teams – Individuals as well as teams – Graduate, undergraduate and university staff – Most projects had a faculty or staff advisor • Project types – Data movement, network management software, security [ 35 ] Selected Projects
    33. 33. • Projects completed – Source code available from a public code repository – Documentation • Code documentation • Final report • Slide presentation • Application demonstrations – November 2013, April 2014 [ 36 ] Current Status
    34. 34. Application Developer: William J. Brockelsby
Organization: North Carolina State University (NCSU)
Location: Raleigh, NC
Project: The application aims to provide an efficient, customized, enriched "Bring Your Own Device" (BYOD) experience for any campus user or visitor. The user can be securely verified to access the enterprise network, and routed to the most pertinent digital resources based on the user's role, work, and/or discipline. For example, a visiting particle physics expert could have their identity and device recognized and immediately routed to the institution's dedicated physics network, cloud, and other digital resources. [ 37 ] Network Administration Control
    35. 35. Application Developer: Marc De Leenheer, Ali Al-Shabibi, William Snow, Guru Parulkar Organization: Open Network Laboratory
Location: Palo Alto, CA
Project: Researchers need networking environments that can be molded to meet their demanding requirements. OpenVirTex is a network virtualization platform that aims to allow each researcher to build discrete virtual networks on a shared physical infrastructure for research collaboration. Each virtual network in the multi-tenant cloud would look and behave just like a physical network, but provide complete control over the address space and topology. [ 38 ] OpenVirTex
    36. 36. • Application Developer: Mo Dong, Qingxi Li, and P. Brighten Godfrey • Organization: University of Illinois at Urbana-Champaign (UIUC)
Location: Urbana, IL
Project:– Big data senders are faced with the challenge of efficiently and effectively distributing large data sets among collaborators. BBCC aims to allow individual senders to observe and diagnose changing network conditions, optimize and control network characteristics, and meet specific data movement needs. [ 41 ] Black Box Congestion Control (BBCC)
    37. 37. Application Developer: Rasha El-Jaroudi, Aditi Ghag Organization: Georgia Institute of Technology (Georgia Tech)
Location: Atlanta, GA
Project:– All network users, especially those without networking expertise, need to reliably stream video content. The application aims to enable seamless video streaming by providing better control through a holistic view of network conditions, a programmable interface to an SDN controller, and leveraging content from optimal sources based on the global view of the network in order to reduce bandwidth costs. [ 42 ] Software Defined Networking (SDN) based Application for Efficient Video Streaming
    38. 38. Application Developer: Andy Li Xiaolin, Ze Yu Organization: University of Florida
Location: Gainesville, FL
Project: Scientists geographically distributed need to collaborate with massive amounts of data. MapReduce is a popular programming model for analyzing large data sets generated by experiments. This application aims to provide a solution that enhances the functionality of MapReduce by reserving network bandwidth and performing intelligent data transfers. [ 43 ] Elf: Network-Enhanced Data Prefetching Middleware for Geo-Distributed MapReduce
    39. 39. Application Developer: Pingping Lin, Jonathan Hart, Umesh Krishnaswamy Organization: Open Network Laboratory
Location: Palo Alto, CA
Project: Collaborators using different types of network architectures need them to reliably interoperate. This application aims to allow Software Defined Networks to interoperate with legacy networks, ensuring successful network collaboration. Utilizing BGP (Border Gateway Protocol), this solution has been implemented and evaluated for feasibility, and is now ready for deployment in production environments. [ 44 ] Seamless Interworking of Software Defined Networks and IP
    40. 40. Application Developer: Muhammad Shabaz, Arpit Gupta, Nick Feamster Organization: Georgia Institute of Technology (Georgia Tech)
Location: Atlanta, GA
Project: Network operators need controller software tailored for inter-domain routing due to the rise of SDN infrastructures at several commercial SDN Internet exchange (SDX) points. This application aims to create an architecture for SDX that can solve a variety of problems for network operators who must rely on BGP by enabling new functions and a richer set of policies than are possible with today's inter-domain routing systems. [ 45 ] Software Defined Networking (SDN) for Internet Exchange Points
    41. 41. Application Developer: Stephen Tredger, Patrick C. McGeer Organization: University of Victoria and PlanetWorks LLC
Location: Victoria, BC
Project:– Testing and developing new applications are crucial to advancing network infrastructure, innovating Internet technologies, and accelerating scientific discovery through enhanced research collaboration. The GEE aims to radically simplify researchers and educators workflow in using distributed systems to quickly deploy and develop distributed apps that work higher in the stack by using pre allocated dedicated VLANs creating an international network to isolate GEE experiments. [ 46 ] GENI Experiment Engine (GEE) Infrastructure
    42. 42. Title or Title • Event/Date • Presenter, PresenterTitle, Internet2 What will global innovators do with the next Innovation Platform?
    43. 43. CREATING A CLIMATE FOR INNOVATION ON INTERNET2 Thank you. For more information, visit or e-mail 48 – 5/21/2014, © 2012 Internet2