exploring networks of the futurenv/chip_elliott-sept_10_2011.pdf · 2012-07-30 · exploring...

9/7/2011

1

GENIExploring Networks of the Future

Status and outlook

Sponsored by the National Science Foundation

Chip ElliottGENI Project Director

September 2011www.geni.net

Outline

• GENI – Exploring future internets at scale• Introducing GENI: an example• GENI’s growing suite of infrastructure• Experiments going live across the US!• What’s next for GENI?• GENI and US Ignite

Accelerating the b ild o t

Sponsored by the National Science Foundation 2September 9-10, 2011 www.geni.net

• Accelerating the build-out

9/7/2011

2

Science IssuesWe cannot currently

understand or predict theInnovation Issues

Substantial barriers to

Global networks are creatingextremely important new challenges

understand or predict the behavior of complex,large-scale networks

Substantial barriers toat-scale experimentation with new architectures, services,

and technologies


Credit: MONET Group at UIUC

Society IssuesWe increasingly rely on

the Internet but are unsure we can trust its security,

privacy or resilience

What is GENI?

• GENI is a virtual laboratory for exploring future internets at scale, now rapidly taking shape in prototype form across the United States

• GENI opens up huge new opportunities– Leading-edge research in next-generation internets

– Rapid innovation in novel, large-scale applications


• Key GENI concept: slices & deep programmability– Internet: open innovation in application programs

– GENI: open innovation deep into the network

9/7/2011

3

Revolutionary GENI IdeaSlices and Deep Programmability

Install the software I want throughout my network slice(into firewalls, routers, clouds, …)

And keep my slice isolated from your slice,so we don’t interfere with each other


We can run many different “future internets” in parallel

GENI is now going live across the US GENI-enabling testbeds, campuses, and backbones


9/7/2011

4

Outline





A bright idea

I have a great idea! The original Internet architecture was designed to connect one computer to another – but a better parchitecture would be fundamentally based on PEOPLE and CONTENT!

That will never work! It won’t scale! What about security? It’s impossible to implement or operate! Show me!


9/7/2011

5

Trying it out

My new architecture worked great in the lab, so now I’m going to try a larger experiment for a few months.p

And so he poured his experimental


And so he poured his experimental software into clouds, distributed clusters, bulk data transfer devices (‘routers’), and wireless access devices throughout the GENI suite, and started taking measurements . . . He uses a modest slice of GENI, sharing its

infrastructure with many other concurrent experiments.

It turns into a really good idea

Boy did I learn a lot! I’ve published papers, the architecture has evolved in major ways, and I’m even attracting real users!g

Location-based social


His experiment grew larger and continued to evolve as more and more real users opted in . . .

Location based social networks are really cool!

His slice of GENI keeps growing, but GENI is still running many other concurrent experiments.

9/7/2011

6

The (opt-in) user’s view

Good old Internet

Interesting new services –I just use them through an app!

InternetSlice 0

Slice 1

Slice 1


Slice 2

Slice 3

Slice 4

Experiment turns into reality

My experiment was a real success, and my architecture turned out to be mostly compatible with today’s Internet after all –p yso I’m taking it off GENI and spinning it out as a real company.

I always said it was a good idea, but way too conservative.


9/7/2011

7

Meanwhile . . .

I have a great idea! If the Internet were augmented with a scalable control plane

d lti t t l it ldand realtime measurement tools, it could be 100x as robust as it is today . . . !

And I have a great concept for incorporating live sensor feeds into our daily lives !


If you have a great idea, check out theNSF CISE research programs for current opportunities.

Moral of this story

• GENI is meant to enable . . .– At-scale experiments, which may or may not beAt scale experiments, which may or may not be

compatible with today’s Internet– Both repeatable and “in the wild” experiments– ‘Opt in’ for real users into long-running experiments– Excellent instrumentation and measurement tools– Large-scale growth for successful experiments, so

d id b h k d t l


good ideas can be shaken down at scale

GENI creates a huge opportunity for ambitious research!

9/7/2011

8

Outline





Spiral DevelopmentGENI grows through a well-structured, adaptive process

• GENI Spiral 3Early experiments, meso-scale build, interoperable control frameworks, ongoing integration, system designs for security and instrumentation starting up operationsinstrumentation, starting up operations.

• Envisioned ultimate goalLarge-scale distributed computing resources, high-speed backbone nodes, nationwide optical networks, wireless & sensor nets, etc.

GENI scale & integration


Spirals: 1 2 3

Risk

4 5…GENI Prototyping Plan

9/7/2011

9

FederationGENI grows by “GENI-enabling” heterogeneous infrastructure

Campus#3 Commercial

Clouds

My experiment runs acrossthe evolving GENI federation.

Backbone #1

Backbone #2

Access#1

Clouds

CorporateGENI suites

Other-NationProjects

ResearchTestbed

Campus My GENI Slice

This approach looks remarkably familiar . . .


Goals: avoid technology “lock in,” add new technologies as they mature, and potentially grow quickly by incorporating existing infrastructure into the overall “GENI ecosystem”

NSF parts of GENICampus

#2

Projects

Enabling “at scale” experiments

• How can we afford / build GENI at sufficient scale?– Clearly infeasible to build research testbed “as big as the Internet”

– Therefore we are “GENI-enabling” testbeds, commercial equipment, campuses, regional and backbone networks

– Students are early adopters / participants in at-scale experiments– Key strategy for building an at-scale suite of infrastructure


“At scale” GENI prototype

Campus photo by Vonbloompasha

GENI-enabled campuses,students as early adopters

HP ProCurve 5400 Switch

NEC WiMAX Base Station

GENI-enabledequipment

9/7/2011

10

Georgia Tech: a great exampleOne of the first 14 GENI-enabled campuses

• OpenFlow in 4 GT lab buildings now

• OpenFlow/BGPMux coursework nowNick Feamster

PI

Russ Clark, GT-RNOC

Ellen Zegura

Ron Hutchins, OIT

coursework now

• Dormitory trial

• Students will “live in the future” – Internet in one slice, multiple future internets in additional slices

T i l f “GENI bl d” i l i t


Toroki LightSwitch 4810

Trials of “GENI-enabled” commercial equipment

Arista 7124S Switch

HP ProCurve 5400 Switch Juniper MX240 EthernetServices Router

NEC IP8800 Ethernet Switch

NEC WiMAX Base Station HTC Android smart phone

GENI racks

Building the GENI Meso-scale Prototype

WiMAXStanfordUCLAUC BoulderWisconsin

OpenFlowStanford

U WashingtonWisconsin

Indiana

ShadowNetSalt Lake City

Kansas CityDC

Atlanta

WisconsinRutgersPolytechUMassColumbia

OpenFlowBackbonesSeattleSalt Lake CitySunnyvaleDenverKansas CityHouston

IndianaRutgers

PrincetonClemson

Georgia Tech


Atlanta HoustonChicagoDCAtlanta

Toroki LightSwitch 4810

Arista 7124S Switch

HP ProCurve 5400 Switch Juniper MX240 EthernetServices Router

NEC IP8800 Ethernet Switch

NEC WiMAX Base Station HTC Android smart phone

GENI racks

9/7/2011

11

World-class GENI Partners National LambdaRail and Internet2

Internet2

P t GENI

National LambdaRail

ProtoGENI & SPP


Buildout for GENI prototyping within two national footprintsto provide end-to-end GENI slices (IP or non-IP)

National LambdaRailUp to 30 Gbps bandwidth

Photo by Chris Tracy

Meso-scale GENI, August 2011

• Layer 2 slices span campuses, Internet2, and NLR

• Each VLAN contains ~ 25 OpenFlow switchesand 40+ computers (PlanetLab & ProtoGENI)

• OpenFlow / FlowVisor manages slices within a VLAN

• Ongoing federated ops (8 campuses, 2 backbones, GMOC),


O go g ede ated ops (8 ca puses, bac bo es, G OC),each organization with its own operators, policies, etc.

• Now shaking down large-scale slices

9/7/2011

12

PlanetLab, ProtoGENI, OpenFlow resourcesspanning multiple campuses, Internet2, & NLR


Want to plug your campus in?Send email: [email protected]

Sponsored by the National Science Foundation

9/7/2011

13

Outline





Major research demos, Nov 2010


9/7/2011

14

9 major experiments demo’dat GEC 9 (Nov 2010)

• Some of the nation’s best young researchers . . .– Academic and industrial– Networking and distributed systems– Some helped build GENI, most have not

• Demonstrating their earliest research experiments– Many different ideas for “future internets”– Now being tried out experimentally for the first time


• On the nationwide, “meso-scale” GENI prototype

GENI supported 9 different future internet experiments,simultaneously, each in its own slice

Pathlet ArchitectureGEC 9 experiment demonstrationResilient Routing in thePathlet Architecture

Ashish Vulimiri and Brighten GodfreyUniversity of Illinois at Urbana-Champaign

Deploy innovative routing

• Lets users monitor and select their own

t k th t

path 1failed link

Deploy innovative routing architecture deep into

network switches across the US

Sponsored by the National Science Foundation 28September 9-10, 2011 www.geni.netSponsored by the National Science Foundation November 3, 2010

network paths to optimize their services

• Protects critical traffic even without waiting for adaptation time

28

path 2

9/7/2011

15

ActiveCDNGEC 9 experiment demonstration

Program content distribution services deep i t th t k d t di t ib ti i l

ActiveCDNActiveCDN

Utah

GPO

into the network, adapt distribution in real time as demand shifts


Kansas

Clemson

Benefits of ActiveCDN:• Dynamic deployment based on load• Localized services such as weather, ads and news Jae Woo Lee, Jan Janak, Roberto Francescangeli,

SumanSrinivasan, Eric Liu, Michael Kester, SalmanBaset, Wonsang Song, and Henning Schulzrinne

Internet Real-Time Lab, Columbia University

Generate “raw” live dataViSE/CASA radar nodesGenerate “raw” live dataViSE/CASA radar nodes

http://stb.ece.uprm.edu/current.jsphttp://stb.ece.uprm.edu/current.jsp

ViSE views steerable radars as shared, virtualized resourceshttp://geni.cs.umass.edu/vise

ViSE views steerable radars as shared, virtualized resourceshttp://geni.cs.umass.edu/vise

Weather NowCastingGEC 9 experiment demonstration

David Irwin et al

1 Spin up system in Amazon1 Spin up system in Amazon

“raw” live data

“raw” live data

Nowcast images for display

Nowcast images for display

Create and run realtime “weather service on demand”as storms turn life-threatening


Multi-radar NetCDF DataMulti-radar NetCDF Data

Nowcast ProcessingNowcast Processing

1. Spin up system in Amazon commercial EC2 and S3 services on demand

1. Spin up system in Amazon commercial EC2 and S3 services on demand

p yp y

9/7/2011

16

GEC 9 experiment demonstrationAster*x Load Balancing (via OpenFlow)

Nikhil Handigol et al, Stanford Univ.

Program realtime load-balancing functionality deep into the

network itself


Outline





9/7/2011

17

Looking forwardGrowing to the “at scale” GENI

• Suggest 100-200 US campuses as target for “at scale”– Both academia and national labs– GENI-enable the campusesGENI enable the campuses– Their students, faculty, staff can then “live in the future” using both

today’s Internet and many experiments– Build out backbones, regionals, and shared clouds to support the

campuses

• Grow via ongoing spiral development– Identify, understand, and drive down risks


– Learn what is useful and what is not– Early GENI campuses can help later ones

• Transition to community governance

Envisioned architecture

Metro

InternetISPU N I V E R S I T YU N I V E R S I T Y

MetroResearch

Backbones

U N I V E R S I T YU N I V E R S I T Y

Regional Networks Campus

g

g

gLegend

GENI-enabled hardware

Layer 3Control Plane

Layer 2Data Plane


• Flexible network / cloud research infrastructure

• Also suitable for physics, genomics, other domain science

• Support “hybrid circuit” model plus much more (OpenFlow)

• Distributed cloud (racks) for content caching, acceleration, etc.

Regional Networks Campus gg

9/7/2011

18

• Internet 2 NDDI slide?


Growing GENI to 100-200 campuses

GENI racks, OpenFlow, WiMAX, training, ops

Solicitation 3 efforts

14 30 100? 200?


Spiral development . . .

GENI racks, OpenFlow, WiMAX, training, ops

Campus expansions

9/7/2011

19

Solicitation 3’s role in campus buildouts

• GENI Solicitation 3– More WiMAX base stations

with Android handsets– GENI-enable 5-6

regional networks– Inject more

OpenFlow switchesGENI Racks serve as programmable routers, distributed clouds, content

distribution nodes, caching or transcoding nodes etc


into Internet2 and NLR– Add GENI Racks to 50-80 locations

within campuses, regionals, andbackbone networks

transcoding nodes, etc

A new solicitation comingGENI-enabling 20+ campuses

• Growing to 20+, then 100-200 campuses– Plan roughed out at 2nd CIO workshop, July 7, 2011

(thank you EDUCAUSE)“B dd t ” f h l t id 2 3– “Buddy system” for each meso-scale campus to guide 2-3 new campuses

– Increase GENI-enabled campuses from 14 to 40-50 in a staged manner, over several years

• GENI candidate campuses are now being lined up– Two-person teams will visit candidate campuses this fall,

helping campus CIOs draw up plans and proposal material


p g p p p p p– “Campus expansion” solicitation expected ~ Dec 2011

• Create a proposal to GENI-enable your campus

Larry Landweber is organizing this effort: <[email protected]>

9/7/2011

20

Outline





US Ignite

• US Ignite is an initiative to spark the development of gigabit applications and servicesin areas of national priorit ad anced• in areas of national priority: advanced manufacturing, health, education, energy, economic development, transportation, and public safety/emergency preparedness

• on an ultra high speed, deeply programmable, and sliceable network testbed


and sliceable network testbed.

9/7/2011

21

Research Infrastructurefor Computer Scientists

Public-Private Partnershipfor Next-Gen Applications

Future commercialofferings

federation

GENI US Ignite

US Ignite is now taking shapeBridging CS Experiments to Next-Gen Applications in Cities

CS Experiments Pre-commercial Applications

Campus and LabApplied Research

App creation teams

GENI members, policies, … US Ignite members, policies, …

GE

NI t

echn

olog

y

Service creatorsCS Research


US Ignite is a new organization that will promote advanced applications and infrastructure leveraging GENI research and technologies.

Experimental Usage and Demonstrations

Regional and backbone networks

Campus networks Municipal andcommercial networks

Commercial Applications

GENI-enabled citiesFirst concrete step in US Ignite activity

• Very strong interest from 6 US cities to date– Chattanooga, Cleveland, Lafayette LA, Philadelphia,

Salt Lake City region, Washington DC– Their citizens will be able to “live in the future”

• Cities can be GENI-enabled very rapidly– We have visited all 6 cities for surveys, discussions– GENI rack, OpenFlow, and Layer 2 connectivity appear quite

feasibleCan be federated into GENI very quickly


– Can be federated into GENI very quickly

• Can support experimental, gigabit applications in GENI slices through cities– Creates tremendous new research opportunities

9/7/2011

22

Existing ISP

Existing head-end

Draft of US Ignite City Technical Architecture

Early DRAFT CONCEPT –for discussion only!

connects

Layer 2

Layer 3 GENI control plane

Layer 2 connect to subscribers

OpenFlow switch(es)FlowvisorRemote management

Home

Most equipment not shown


Layer 2 Ignite Connect(1 GE or 10GE)

New GENI / Ignite rack pair

InstrumentationAggregate managerMeasurementProgrammable PCsStorageVideo switch (opt)

What would it look like?“Cities living in the future”

• Citizens’ view– Great new applications– New content services

Good old Internet

Interesting new services –I just use them through an app!

Slice 0

Slice 1 New content services . . .– New weather services . . .– New health services . . .– New energy services . . .– New government services . . .

Slice 2

Slice 3

Slice 4

Slice 1

• Entrepreneurs’ viewN t ti li ti b


– Next-generation applications can be cheaply & rapidly rolled out

– “My software goes HERE”– Create and try out new apps that

exploit deep programmability– Experiment with cities “living in the

future” to gain market edge

9/7/2011

23

US Ignite workshops

• NSF information: http://www.nsf.gov/cise/usignite/• Two workshops to date

May 16 at NSF– May 16 at NSF– June 9 at Case Western

• Basic goal: matchmaking between cities / researchers– Run research applications across one or more cities– Focus areas: health, energy, public safety, education

• NSF expects to solicit & fund proposals3 d k h b bl t GEC 12 (b i it ti l )


• 3rd workshop probable at GEC 12 (by invitation only)

• Very informative CCC blogs re workshops:• http://www.cccblog.org/2011/05/24/recapping-the-us-ignite-gigabit-applications-workshop/• http://www.cccblog.org/2011/06/11/us-ignite-gigu-workshops-living-the-future-today/

US Ignite in broadband citiesA huge opportunity for innovation & leap-ahead

• Slicing and deep programmability greatly expandthe revolutionary potential of broadband

Citizens of the fortunate cities can “live in the future”– Citizens of the fortunate cities can live in the future– Today’s Internet on Channel 1– Many new next-generation applications on Channels 2, 3, . . .– Opens up leading-edge, high impact research fields– Creates huge opportunities for innovation and leap-ahead

• Appears fairly simple / low-cost technicallyDepends on network equipment selected etc


– Depends on network equipment selected, etc.

• Social aspects are very important (city ≠ campus!)

9/7/2011

24

Outline





GENI moves to the next stage

• Ramp down of Solicitation 1 efforts (autumn) . . .– Some prototyping efforts will sunset as Sol 1 funds run out– Many will continue under Sol 2 / 3 funding– Others have funds remaining, will receive extensions

• . . . and major growth in GENI infrastructure– Rise in experimentation and continuous operations– Growth across 20+ campuses, regionals, and backbones– US Ignite cities & next-gen applications coming online soon

W ’ l ki f 20


• We’re looking for 20+ campusesto take GENI to the next stage– It’s really happening– About 30 campus CIOs are now actively engaged

9/7/2011

25

GENI Engineering ConferencesWe welcome your participation in creating GENI

• 12th meeting, open to all:November 2-4, 2011, Kansas City– 3 tracks: software, campuses, experimenters3 tracks: software, campuses, experimenters– Tutorials and workshops– Travel grants to US academics for participant diversity


exploring networks of the futurenv/chip_elliott-sept_10_2011.pdf · 2012-07-30 · exploring...

Documents