OpenFlow/SDN: New Approach to Networking. Guru Parulkar

Transcription

1 OpenFlow/SDN: New Approach to Networking Guru Parulkar 1

2 Problem with Internet Infrastructure? Routing, management, mobility management, access control, VPNs, Feature Operating System Feature Million of lines of source code RFCs Specialized Packet Forwarding Hardware Billions of gates Bloated Power Hungry Vertically integrated, complex, closed, proprietary Not suitable for experimental ideas Not good for network owners & users; Not good for researchers.

3 Problem: No Abstractions for Control Plane Addition of a new function to the network Highly complex distributed system problem Networks too difficult to program and to reason about no good abstractions and interfaces Distributed Network Functions Router/Switch/Appliance Router/Switch/Appliance State Distribution Mechanism Router/Switch/Appliance Not good for even network vendors 3

4 Software-Defined Network with Key Abstractions in the Control Plane Well-defined open API Routing TE Network OS Mobility Network Map Abstraction Packet Forwarding Open interface (OpenFlow) to Forwarding Abstraction: L1/L2/L3 Separation of Data and Control Plane Packet Forwarding Packet Forwarding Programmable Basestation

5 Software Defined Network with Virtualization Abstract Network Model Network Control Virtualization Program Global Network View Network OS

6 Nation-wide OPEN Infrastructure Part of NSF s GENI

7 Example Research Enabled Data center: energy conservation, routing, and management Seamless use of diverse wireless networks Network based load balancing Packet/circuit convergence, traffic engineering Simpler control plane for converged packet/circuit MPLS nets Slicing and remote control/management of home networks Distributed snap shot of VMs (by DOCOMO researchers) Inter-domain routing with pathlets (by UIUC) Redundant traffic elimination [for CDNs] (by Univ of Wisconsin) And many more

8 Example Load-balancing as a network primitive Nikhil Handigol, Mario Flajslik, Srini Seetharaman

9 LOAD-BALANCER

10

11

12 Load Balancing is just Smart Routing

13

14 Feature Feature Nikhil s Experiment: <500 lines of code Network OS (NOX)

15 Other Examples and Demonstrations 15

16 Major Change in Paradigm No longer designing distributed control protocols Now just defining a centralized control function Configuration = Function(view) This spells the end for distributed protocols Easier to write, reason about, maintain,. Beginning of the software era of networking Rate of change, nature of standards, different culture, etc.

17 Industry Embracing SDN Largest Network Providers/Operators Vendors and start-ups More... More... Note: Level of interest varies 17

18 Pmargin Fmargin Why Does Verizon Care? Software defined network enables inexpensive feature insertion for new services and revenue uplift. Stu Networking Summit ARPU(PMO) $ Unacceptably high cost escalation results in a non-sustainable business case. The cost of PMO needs to be matched to the characteristics of revenue growth to support consistent margin. Target FMO costs Cost of PMO The cost of transition must not be prohibitive t Using COTS hardware following mass market cost curves lowers equipment expenditures Confidential and proprietary materials for authorized Verizon personnel and outside agencies only. Use, disclosure or distribution of this material is not permitted to any unauthorized persons or third parties except by written agreement Verizon. All Rights Reserved.

19 They want to offer VPN Services Traffic Engineering And other value added services Distributed Network Functions Router Operating System Carrier Networks Today Go to vendors and ask them for solutions: State Distribution Mechanisms PE Label Distrib ution Distributed Network Functions each with their own State Distribution Mechanisms Vendors find it hard to develop solutions LDP E-BGP learned Route Advert I-BGP + RR IGP- Route Advert, Link- State VPN- IPv4 Route Advert MP- BGP TE Label Distrib ution Providers find it too slow and too expensive to deploy and operate Providers depend too much on vendors RSVP- TE OSPF v2 19

20 Carrier Networks with SDN Services / Network Applications (3) TE 2.0 VPN Simpler Control Plane (2) Routing Discovery Label Distribution NETWORK OPERATING SYSTEM Recovery Vendor-agnostic Open Interface Simpler Cheaper Multi-Vendor Data Plane (1) PUSH SWAP POP 20

21 Cloud Provider Today Assign VMs to any server without considering L2/L3 scalability Support multiple tenants with lots of VMs interconnected Support each tenant with its own customized network for VMs: topology, bandwidth, security, load-balancing And everything is dynamic Providers find current Tenant-B solutions too complex and too expensive Tenant-C? Vendors cannot create solutions fast enough Load balancers Firewalls Tenant-A VM VM VM VM VM VM VM VM VM IDS s VM VM VM VM VM VM VM VM VM VM VM Network is the long pole VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM 21

22 Cloud Provider with SDN Tenant-A Tenant-B Tenant-C Load Balancing Firewall Load Balancing IDS Load Balancing Firewall IDS NETWORK OS NETWORK OS NETWORK OS VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM 22

23 Opportunities for Future Work Big Connector Abstractions Virtual Network Network Map Flow Space Scalability Reliability Security OF Switch OF Switch Feature Mininet & Tools Network OS OF Switch Feature Programming System/Tools Network HyperVisor Open Radio Feature Network Network OS OF OS-wireless OpenRadio Switch Big Connector Virtual Network Logical Map of Network FlowVisor Prog Systems Systems Xlities OF Switch

24 Opportunities for Future Work Domains of Use Abstractions Enterprise Networks Big Connector Virtual Network Network Map OF Switch Datacenter Networks Flow Space Scalability Reliability Security OF Switch Mininet & Tools Open Radio Network OS Network FlowVisor OS-wireless Prog Systems Systems Service Provider Networks Cellular Networks Xlities Home Networks Research Prototype Platforms Demonstrate Deploy

25 How You can Participate Researcher Deploy OpenFlow network & demonstrate interesting network capabilities Design and build better building blocks: FlowVisor, Network OS, For different domains of use: home, enterprise, backbone, data center, and different network technologies: wired, wireless, optical Vendor Add OpenFlow to a key product and have researchers build on it Build and demonstrate interesting network capabilities for your customers Join Open Networking Foundation (ONF) Design and build better building blocks: FlowVisor, Controller, For different domains of use and network technologies as above Provider Ask your vendors to support OpenFlow Deploy OpenFlow network & demonstrate interesting network capabilities potentially with partner vendors and/or researchers Join Open Networking Foundation (ONF)

26 Open Networking Summit A Premier Event for OpenFlow/SDN Next Event: April 2012, Bay Area CA Put your name on

27 Final Takeaways Network owners/operators/vendors face real challenges scale and management => capex and opex Networking has to and will change Give more control to network owners/operators Enable and accelerate innovation and customization SDN is the solution of choice Owners/operators want to deploy it Vendors stepping up to provide products/solutions SDN has to mature to deliver on the promise OpenFlow protocol, switches, and SDN software stack And it will with all this community interest & participation

28 OpenFlow and Software Defined Networking (SDN) are not only here to stay, but they will define the future of networking. Network World, 10/18/2011

29 Team at Stanford Nick McKeown, Guido Appenzeller, Nick Bastin, David Erickson, Glen Gibb, Nikhil Handigol, Brandon Heller, TY Huang, Peyman Kazemian, Masayoshi Kobayashi, Jad Naous, Johan van Reijendam, Srini Seetharaman, Rob Sherwood, Dan Talayco, Paul Weissman, Tatsuya Yabe, KK Yap, Yiannis Yiakoumis and many more. With Scott Shenker and team at Berkeley and Martin Casado at Nicira

30 Time to engage and help shape this important revolution in networking

31 Important Questions Is the networking problem real? How does OpenFlow/SDN help? Are there any proof points? Is SDN just a cool idea or is it more fundamental? How can I experiment with OpenFlow/SDN What is next?

32 Enterprise Networking Enterprise network operators want.. Firewall and access control Delegate management to departments Share data network with VoIP network Lots of VLANs that stretch across buildings By-passing bottlenecks/check points for specific applications Host web services with load balancing Easy guest wireless access with security And more How do they do it today? 32

33 Enterprise Network: Today s solution Load balancer Firewall Firewall Load balancer ACL IDS ACL IDS ACL ACL ACL Proliferation of appliances Increased management complexity - Device oriented management - Each device type has its own management High Capex, high Opex Too much reliance on vendors 33

34 Router Operating System Carrier Network Providers They want to offer VPN Services Traffic Engineering And other value added services Go to vendors and ask them for solutions: Distributed Network Functions State Distribution Mechanisms PE Label Distrib ution Distributed Network Functions each with their own State Distribution Mechanisms Vendors find it hard to develop solutions LDP E-BGP learned Route Advert I-BGP + RR IGP- Route Advert, Link- State VPN- IPv4 Route Advert MP- BGP TE Label Distrib ution Providers find it too slow and too expensive to deploy and operate Providers depend too much on vendors RSVP- TE OSPF v2 34

35 Cloud Provider Assign VMs to any server without considering L2/L3 scalability Support multiple tenants with lots of VMs with interconnected Support each tenant with its own customized network for VMs: topology, bandwidth, security, load-balancing And everything is dynamic Providers find current Tenant-B solutions too complex and too expensive Tenant-C? Vendors cannot create solutions fast enough Load balancers Firewalls Tenant-A VM VM VM VM VM VM VM VM VM IDS s VM VM VM VM VM VM VM VM VM VM VM Network is the long pole VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM VM 35

36 NETWORKING AND SERVER VIRTUALIZATION! The hidden headaches of moving to private cloud Guido ~3-6x increase in networking related trouble tickets per month after moving to private cloud computing model! Source: team estimates based on interviews across 8 enterprise and university network professionals in planning, pilots or production with private cloud architectures proprietary and confidential! 36

37 Imagine You re a Hot-Shot Programmer/Researcher And want to create a new service with real users Using Amazon EC2/S3, for example, you can allocate Computing Resources Storages But you cannot create a customized network: topology, BW, security, routing,. You cannot ask a cloud provider for a (virtual) network of your own and program it? 37

38 Confused User Poorly managed network due to lack of knowledge More devices, more complexity Home Network More players rely on Home Net Want to reach users with new, better and faster services Want to use home network for reducing costs (e.g., smartmeter, femtocell BS) Disconnection WAN 38

39 Home Network Wouldn t it be nice if you could.. Outsource Management WiFi Guest Access Control Plane 10% Share Electric Utility Control Plane Customize Video Streaming Control Plane 39

40 Summary: Network Operators/Vendors Face Real Challenges Networks are built using vertically integrated complex boxes Addition of a new function to the network Highly complex distributed system problem Proliferation of appliances Net Result Increased management complexity Lack of innovation and customization by providers/third parties High CapEx & OpEx Distributed Network Functions Router/Switch/Appliance Router/Switch/Appliance State Distribution Mechanism Router/Switch/Appliance

41 Important Questions Is the networking problem real? How does OpenFlow/SDN help? Are there any proof points? Is SDN just a cool idea or is it more fundamental? How can I experiment with OpenFlow/SDN What is next?

42 Restructured Network Feature Feature Network OS Feature Feature Operating System Specialized Packet Forwarding Hardware Feature Operating System Feature Feature Operating System Feature Specialized Packet Forwarding Hardware Feature Operating System Feature Specialized Packet Forwarding Hardware Feature Feature Specialized Packet Forwarding Hardware Operating System Specialized Packet Forwarding Hardware 42

43 Software-Defined Network with Key Abstractions in the Control Plane Well-defined open API Routing TE Network OS Mobility Network Map Abstraction Packet Forwarding Open interface (OpenFlow) to Forwarding Abstraction: L1/L2/L3 Separation of Data and Control Plane Packet Forwarding Packet Forwarding Programmable Basestation

44 Isolated slices SDN with Virtualization versioning, healthy competition, protection from security threat/bugs Many operating systems, or Many versions App App App App App App App App Network Operating System 1 Network Operating System 2 Network Operating System 3 Network Operating System 4 Open interface to hardware Virtualization or Slicing Layer (FlowVisor) Open interface to hardware Packet Forwarding Hardware Packet Forwarding Hardware Packet Forwarding Hardware Packet Forwarding Hardware Packet Forwarding Hardware 44

45 Enterprise Network And you can even delegate control to someone else Centralized Control Plane Research Labs Load Access IDS Balancing Control Policy NETWORK Routing OPERATING SYSTEM NETWORK OS Vender-agnostic Open Interface Financial Department Policy Routing IDS Access Control NETWORK OS Load balancer Firewall Firewall Load balancer Simple, Cheaper Multi-vendor Data Plane ACL IDS ACL ACL ACL IDS ACL 45

46 Cloud Provider (2) Tenant B Tenant C L F I L F I L F I L F I L F I L F I NETWORK OS NETWORK OS NETWORK OS NETWORK OS NETWORK OS NETWORK OS L F L F NETWORK OS NETWORK OS 46

47 Home Network Manage 10% Share Guest Wifi Access Control Plane Access Control Bandwidth Control NETWORK OS WiFi Guest Access Control Plane Electric Utility Video Streaming Control Plane Control Plane Access Control NETWORK OS QoS Control Electric Utility Control Plane Network OS QoS Routing Customize Virtualization or Slicing Layer Video Streaming Control Plane 47

48 Important Questions Is the networking problem real? How does OpenFlow/SDN help? Are there any proof points? Is SDN just a cool idea or is it more fundamental? How can I experiment with OpenFlow/SDN What is next?

49 Example 2 Using all the wireless capacity around us KK Yap, Masayoshi Kobayashi, Yiannis Yiakoumis, TY Huang

50

51 Feature Network OS (NOX) KK s Experiment: <250 lines of code WiMax

52 Industry Embracing SDN Largest Network Providers/Operators Vendors and start-ups More... More... Note: Level of interest varies 52

53 Open Networking Foundation to continue standardization of OpenFlow and other SDN interfaces/apis Board of Directors Deutsche Telekom, Facebook, Google, Microsoft, Verizon, Yahoo! 50+ Members Broadcom, Brocade Ciena, Cisco, Citrix, Dell, Ericsson, Extreme, Force10, HP, IBM, Juniper Networks, Marvell, NEC, Netgear, NTT, Riverbed, Vmware, [List growing]

54 More Proof Points May 2011 Interop Las Vegas 14 companies demonstrated OpenFlow Interop 2011 could have been called The OpenFlow Show IT world October st Open Networking Summit Sold out, 400+ registrations, 20+ demos from different industry segments April nd Open Networking Summit 450+ people on pre-registration list!!

55

56 Important Questions Is the networking problem real? How does OpenFlow/SDN help? Are there any proof points? Is SDN just a cool idea or is it more fundamental? How can I experiment with OpenFlow/SDN What is next?

57 Important Questions Is the networking problem real? How does OpenFlow/SDN help? Are there any proof points? Is SDN just a cool idea or is it more fundamental? How can I experiment with OpenFlow/SDN What is next?

58 The Future of Networking, The Past of Protocols Scott Shenker, UC Berkeley Open Networking Summit, October

59 Software-Defined Networking SDN clearly has advantages over status quo But is SDN the right solution? Scott s talk: Not what SDN is, but why SDN is

60 Key to Internet Success: Layers Applications built on Reliable (or unreliable) transport built on Best-effort global packet delivery built on Best-effort local packet delivery built on Physical transfer of bits

61 Layers are Great Abstractions Layers only deal with the data plane We have no powerful control plane abstractions! How do we find those abstractions? Define our problem, and then decompose it.

62 The Network Control Problem Compute the configuration of each physical device E.g., Forwarding tables, ACLs, Operate without communication guarantees Operate within given network-level protocol Only people who love complexity would find this a reasonable request

63 From Requirements to Abstractions 1. Operate without communication guarantees Need an abstraction for distributed state 2. Compute the configuration of each physical device Need an abstraction that simplifies configuration 3. Operate within given network-level protocol Need an abstraction for general forwarding model

64 1. Distributed State Abstraction Shield mechanisms from vagaries of distributed state While allowing access to this state Natural abstraction: global network view Annotated network graph provided through an API Control mechanism is now program using API No longer a distributed protocol, now just a graph algorithm E.g. Use Dijkstra rather than Bellman-Ford

65 Network Software Traditional Defined Switches Control Network and/or Mechanisms Routers (SDN) e.g. routing, access control Control Program Global Network View Distributed algorithm running between neighbors Network OS

66 2. Specification Abstraction Control program should express desired behavior It should not be responsible for implementing that behavior on physical network infrastructure Natural abstraction: simplified model of network Simple model only enough detail to specify goals This is network virtualization

67 Simple Example: Access Control Abstract Network Model Global Network View

68 Software Defined Network Abstract Network Model Network Control Virtualization Program Global Network View Network OS

69 3. Forwarding Abstraction Control plane needs flexible forwarding model Abstraction should not constrain control program Should support whatever forwarding behaviors needed It should hide details of underlying hardware Crucial for evolving beyond vendor-specific solutions

70 OpenFlow Interface to Forwarding Abstraction Routing Network OS If header = x, send to port 4 If header = y, overwrite header with z, send to ports 5,6 If header =?, send to me Flow Table

71 Forwarding Primitives <Match, Action> Match arbitrary bits in headers: Header Data Match on any header, or new header Allows any flow granularity Action Match: 1000x01xx x Forward to port(s), drop, send to controller Overwrite header with mask, push or pop Forward at specific bit-rate

72 Important Questions Is the networking problem real? How does OpenFlow/SDN help? Are there any proof points? Is SDN just a cool idea or is it more fundamental? How can I experiment with OpenFlow/SDN? What is next?

73 OpenFlow Building Blocks oftrace oflops openseer Monitoring/ debugging tools Stanford Provided ENVI (GUI) LAVI n-casting Expedient Applications NOX Beacon Trema Maestro ONIX Controller FlowVisor Console Commercial Switches HP, NEC, Pronto, IBM, Juniper.. and many more FlowVisor Stanford Provided Software NetFPGA Ref. Switch OpenWRT PCEngine WiFi AP Broadcom Ref. Switch Open vswitch Slicing Software OpenFlow Switches 73

74 Current SDN hardware Juniper MX-series NEC PF5240 HP Procurve 5400 IBM G8264 Netgear 7324 WiMax (NEC) Pronto 3240/3290 Ciena Coredirector PC Engines Ask your vendors 74

75 Commercial Switch Vendors Model Virtualize Notes HP Procurve 5400zl or 6600 NEC PF5240 Pronto 3240 or 3290 with Pica8 or Indigo firmware IBM G OF instance per VLAN 1 OF instance per VLAN 1 OF instance per switch 1 OF instance per switch -LACP, VLAN and STP processing before OpenFlow -Wildcard rules or non-ip pkts processed in s/w -Header rewriting in s/w -CPU protects mgmt during loop -OpenFlow takes precedence -Most actions processed in hardware -MAC header rewriting in h/w -No legacy protocols (like VLAN and STP) -Most actions processed in hardware -MAC header rewriting in h/w - Run in either Legacy or OpenFlow mode - 10G x 48 port & 40G x 4 ports 75

76 Open Controllers Name Lang Platform(s) License Original Author OpenFlow Reference C Linux OpenFlow License Stanford/Nici ra Notes not designed for extensibility NOX Python, C++ Linux GPL Nicira actively developed Beacon Java Win, Mac, Linux, Android GPL (core), FOSS Licenses for your code David Erickson (Stanford) runtime modular, web UI framework, regression test framework Maestro Java Win, Mac, Linux LGPL Zheng Cai (Rice) Trema Ruby, C Linux GPL NEC includes emulator, regression test framework RouteFlow? Linux Apache CPqD (Brazil) virtual IP routing as a service 76

77 Closed-source Controller Vendors Ask these vendors for more info BigSwitch Nicira NEC 77

78 OpenFlow Deployment Recipes Three scenarios: Scale Budget Work/Time 1 Within a single PC ~ $0 (use your laptop) 1 Ph.D student, 1 day 2 3 switches in a Lab ~ $15K 2-3 PhD students, 1 week 3 Within a building (3 switches/floor) ~$15K/floor 1 staff engineer + 1 Ph.D student, 3 month

79 Mininet Deploying OpenFlow within a single PC (1) Creating scalable software-defined (e.g. OpenFlow) networks on a single PC by using Linux processes in network namespaces Scales better than running VMs Best place to start: OpenFlow Hands-on Tutorial Page Step-by-step instruction: how to run mininet, controllers and debugging tools You can download all-in-one VM image (contains Mininet, open-source OpenFlow controllers and OpenFlow debugging tools) OpenFlow Tutorial VM Image OpenVswitch (OpenFlow switch) OpenVswitch (OpenFlow switch) Mininet Host Host Host OpenFlow Controller Debugging Tools (Wireshark) Linux (Guest OS) Your PC s host OS (Windows, Linux, MacOS X, etc).

80 Example: Deploying OpenFlow within a single PC (2) (Step1) Create a OpenFlow network (1 switch, 3 hosts) $ sudo mn --topo single,3 --mac --switch ovsk --controller remote mininet> (Step2) Run reference controller (mimicking L2 switch network) $./controller ptcp: (Step3) Ping between hosts mininet> h2 ping h3 OpenFlow Controller s1: OpenVswitch h2: Host ( ) h3: Host ( ) h4: Host ( )

81 OpenFlow Deployment with 3 Switches in the Lab (1) Case1: Switch doesn t support multiple OpenFlow switch instances (e.g., Pronto) Step1: Create the physical topology - Datapath connections - Control path connections Step2: Configure switch pointing to the controller Step3: Run Controller OpenFlow Switch Host Host Controller Port OpenFlow Switch Controller NOX Host OpenFlow Switch Controller Port Host Controller Port

82 OpenFlow Deployment with 3 switches in the Lab (2) Case2: Switch supports multiple OpenFlow switch instances (e.g., HP, NEC) Step 1: Create VLANs Step 2: Create Physical Topology - Datapath connections - Control path connections Step 4: Assign VLANs to switch instances Step 5: Configure each switch instance pointing to the controller Step 6: Run Controllers VLAN A VLAN B OpenFlow Switch Host Host Host Controller Port OpenFlow Switch Controller NOX A Host Host OpenFlow Switch Controller Port Controller NOX B Host Host Controller Port

83 OpenFlow Deployment in a Building Goal: Run the following non-openflow and OpenFlow VLANs non-openflow VLANs : For legacy production network OpenFlow VLANs : For experiments & OpenFlow production network (Assumption: You have hybrid OpenFlow switches) Production Controller Ctrl A Ctrl B Ctrl C Flow Visor OpenFlow VLAN (for production) OpenFlow VLAN (for experiments) Legacy VLANs

84 OpenFlow Deployment in a Building Step1: - Configure all VLANs as non-openflow to check if it works (few days ~ 1 week) Non-OpenFlow VLAN (for production) Non-OpenFlow VLAN (for experiments): Legacy VLANs

85 OpenFlow Deployment in a Building Step2: - Change 1 non-production VLAN into OpenFlow and run stable production controller - See performance (at least 1 week) Production Controller Non- OpenFlow VLAN (for production) Non-OpenFlow VLAN VLAN (for experiments) Legacy VLANs

86 OpenFlow Deployment in a Building Step3: - Change production VLAN into OpenFlow and run stable production controller - See performance (at least 1 week) Production Controller Production Controller Non-OpenFlow VLAN VLAN (for production) OpenFlow VLAN (for experiments) Legacy VLANs

87 OpenFlow Deployment in a Building Step4: - Change non-openflow Controller to an experimental controller - See performance (at least 1 week) Production Controller Production Ctrl A Controller OpenFlow VLAN (for production) OpenFlow VLAN (for experiments) Legacy VLANs

88 OpenFlow Deployment in a Building Step 5: - Insert FlowVisor to OpenFlow VLAN - See performance (at least 1 week) Production Controller Ctrl A Flow Visor OpenFlow VLAN (for production) OpenFlow VLAN (for experiments) Legacy VLANs

89 OpenFlow Deployment in a Building Step 6 (final): - Run multiple experimental controllers - Keep watching the performance Production Controller Ctrl A Ctrl B Ctrl C Flow Visor OpenFlow VLAN (for production) OpenFlow VLAN (for experiments) Legacy VLANs

90 OpenFlow Deployment at Stanford Gates (CS) 12 HP/NEC/Pronto 30 WiFi APs 1 NEC WiMAX BS CIS (EE) 7 HP Procurve 14 WiFi APs Packard (EE) 8 HP Procurve 15 WiFi APs 1 NEC WiMAX BS Research Use High-profile demos SIGCOMM 08 & 09 best demos Numerous GEC Plenary Research activities OpenFlow Wireless, FlowVisor, Aster*x, OpenPipes, Flow Aggregation Production Use Two WiFi subnets (5300 unique MAC in 2 days) Uses commercial OpenFlow controller

91 OpenFlow Deployment in NSF s GENI Two national backbones 8 pioneer universities +10 universities received OpenFlow switches Regional networks to be OpenFlow enabled in more campuses in to 300 over next few years

92 NDDI Project NDDI (Network Development and Deployment Initiative) Internet2, Indiana University and Stanford University 30+ backbone OpenFlow nodes connected with 10G

93 What is next?

94 Platform Development Deployment Demo Stanford/Berkeley SDN Activities With Partners SIGCOMM08 GEC3 SIGCOMM09 GEC6 GEC9 Interop 2011 VM Migration (Best Demo) Other countries US R&E Community Stanford University GENI software suite Trans-Pacific VM Migration SDN Concept (Best Demo) ~45 switch/aps ~25user In McKeown Group Baby GENI Nation Wide GENI The OpenFlow Show IT World Expedient/Opt-in Manager/FOAM Over 68 countries (Europe, Japan, China, Korea, Brazil, etc.) GENI: 8 Universities + Internet2 + NLR Many other campuses CIS/EE Building Production Network Virtualization FlowVisor FlowVisor (Java) Network OS NOX SNAC Beacon Tools Test Suite oftrace Measurement tools Mininet Ethane Reference Switch NetFPGA Software +Broadcom OpenFlow Spec v0.8.9 v1.0 v

95 How Scott Shenker Put It We have much to learn about the design/use of SDN Just beginning to grapple with the fundamental issues Now this is not the end. It is not even the beginning of the end. But it is, perhaps, the end of the beginning.

96 Stanford/Berkeley Plan for Next Stage of SDN Open Networking Research Center (ONRC) Open Networking Laboratory (ON.Lab) Develop SDN scientific foundation Develop, distribute, support SDN platforms and tools Facilitate deployments in various domains of use Demonstrate new capabilities enabled by SDN for various domains of use Build partnerships to accomplish goals We cannot do it all Enable research community

97 Scope of Activities Abstractions Big Connector Virtual Network Network Map Flow Space Scalability Reliability Security OF Switch OF Switch Feature Mininet & Tools Network OS OF Switch Feature Programming System/Tools Network HyperVisor Open Radio Feature Network Network OS OF OS-wireless OpenRadio Switch Big Connector Virtual Network Logical Map of Network FlowVisor Prog Systems Systems Xlities OF Switch

98 Scope of Activities Cont. Domains of Use Abstractions Enterprise Networks Big Connector Virtual Network Network Map OF Switch Datacenter Networks Flow Space Scalability Reliability Security OF Switch Mininet & Tools Open Radio Network OS Network FlowVisor OS-wireless Prog Systems Systems Service Provider Networks Cellular Networks Xlities Home Networks Research Prototype Platforms Demonstrate Deploy

99 Example Integrative Projects Datacenter Networking Research Cluster OpenFlow networking substrate for GENI Internet2 NDDI: SDN based nation-wide backbone for R&E Home networking with SDN and slicing CloudNet: a nation-wide programmable cloud for R&E - in planning Wireless Cloud: sliceable programmable basestation with intelligence in the cloud

100 ONRC Structure Scott Shenker Berkeley PhD/Postdocs Research Nick McKeown, Guru Parulkar, Scott Shenker Open Network Lab (ON.Lab) Engineers/Tech Leads Technology development e.g. NOX, FV, Beacon, Mininet Explore new use cases Facilitate R&E deployments Nick McKeown Guru Parulkar Stanford PhD/Postdocs Research

101 ONF and ONRC Ideas and technologies To Standards ONF Standards Organization ONRC Research and Advanced Development Relevant Problems & Insights SDN Protocol & Interface Standards SDN Scientific Foundation Technologies Deployments & Interesting Applications