GENI: Catalyzing Network Research May 31, 2007 Larry Peterson Princeton University

Historical Context Overcoming Barriers to Disruptive Innovation in Networking, NSF Workshop Report, January Recommendations 1.Immediately initiate a research program on experimental architectural research in networking. 2.Foster experimental validation of new architectural research in networking. 3.Fund the development and deployment of suitable testbeds.

From the Report “… in the thirty-odd years since its invention, new uses and abuses… are pushing the Internet into realms that its original design neither anticipated nor easily accommodates.” “Freezing forevermore the current architecture would be bad enough, but in fact the situation is deteriorating. These architectural barnacles—unsightly outcroppings that have affixed themselves to an unmoving architecture—may serve a valuable short-term purpose, but significantly impair the long-term flexibility, reliability, security, and manageability of the Internet. ”

GENI Initiative Research Program –NeTS: Future InterNet Design (FIND) –CyberTrust: Clean slate secure network architectures –Theory: Scientific foundations (SING) –… Experimental Facility –proposal to the MREFC program –funds (cutting-edge) facility construction, not research –requires NSB and Congressional approval

Community Planning Group - grass roots individuals GENI Science Council (GSC) - under auspices of the CRA - represents researchers (users) GENI Project Office (GPO) - contract to BBN - responsible for “on time” and “on budget”

Research - Moon Shot To reinvent the Internet; to create a global communication infrastructure worthy of society’s trust.

Research - Problems Security –weak notions of identity that are easy to spoof –protocols that rely on good behavior Mobility –hierarchical addressing closely tied with routing –presumption that communicating hosts are connected Availability –poor visibility into underlying shared risks –multiple interconnected protocols and systems Managability –many coupled, decentralized control loops

Research - Approaches Revisiting definition & placement of function –naming, addressing, and location –routing, forwarding, and addressing –management, control, and data planes –end hosts, routers, and operators Designing with new constraints in mind –selfish and adversarial participants –mobile hosts and disconnected operation –large number of small, low-power devices –ease of network management

Validation Gap AnalysisSimulation / EmulationExperiment At Scale With Real Users Deployment (models)(code) (results) (measurements)

Facility Requirements Generality –minimal constraints on packet formats –diverse collection of technologies Slicability –support many experiments in parallel Fidelity –realistic devices –arranged in realistic topologies Real Users –allow real users to access real content using real applications Sustainability –extend with new technologies –keep operational costs manageable

Facility Architecture GMC User Services Physical Substrate - name space for users, slices, & components - set of interfaces (“plug in” new components) - support for federation (“plug in” new partners)

Greater Landscape The NSF Initiative –we’re writing a proposal –there are rules… GENI Ecosystem –other technology & capacity partners –new user communities Computing Research Community –guinea pig for “big science” projects

Two Sides of the Same Coin Research Infrastructure Two views of PlanetLab –useful research instrument –prototype of a new network architecture Programmability & virtualization deep in the network

PlanetLab 784 machines spanning 382 sites and 40 countries Supports distributed virtualization each of 600+ network services running in their own slice

Slices

User Opt-in Server NAT Client

Usage Stats Users: Slices: 600+ Long-running services: ~20 –CDN, file download, multicast, pub-sub, routing overlays… Bytes-per-day: 4 TB IP-flows-per-day: 190M Unique IP-addrs-per-day: 1M

Lessons Learned Trust relationships –owners, service providers, operators Decentralized control –owner autonomy –delegation –federation Virtualization –scalability is critical –control plane and node OS are orthogonal –least privilege in support of management functionality Resource allocation –decouple slice creation and resource allocation –best effort + overload protection

Lesson (cont) Evolve based on experience –Support users quickly –Let experience dictate what problems to solve Operations –PlanetLab: We debug your network –From universal connectivity to gated communities –If you don’t talk to your university’s general counsel, you aren’t doing network research

VINI - “PlanetLab on Layer 2” Deployed in NLR & I2’s NewNet

VINI - Flexible Topology Supports arbitrary virtual topologies

VINI - Network Events Exposes (can inject) network failures

VINI - External Connectivity s c Experiments can carry traffic for real end-users

VINI - Routing Peers s c BGP Experiments can participate in Internet routing

Challenges Virtualization –Protocol stack –BGP Topology Management Performance –NetFPGA Stanford) –Blade Server + NPs WashU) Internet-in-a-Slice –Click data plane –XORP control plane

Other Activity Prototyping the GENI interfaces Interconnecting PlanetLab and ORBIT (Rutgers) Interconnecting PlanetLab and Emulab (Utah) Federating PlanetLab and OneLab (EU) Replicating VINI on other backbones

Success Scenarios Narrow the validation gap –Sound foundation for future network architectures –Experimental evaluation, rather than paper designs Create new services –Demonstrate new services at scale –Attract real users Aid the evolution of the Internet –Demonstrate ideas that ultimately see real deployment –Provide architectural clarity for evolutionary path Lead to a future global network –Purist: converge on a single new architecture –Pluralist: virtualization supporting many architectures

More Information PlanetLab – VINI – GENI –