1. University of Kentucky Joint Techs 2011 Monitoring GENI Networks Jim Griffioen and Zongming Fei Laboratory for Advanced Networking University of Kentucky Lexington, KY January 31, 2011

2. University of Kentucky Project Participants m University of Kentcuky q James Griffioen (PI) q Zongming Fei q Hussamuddin Nasir (Lead Programmer) q Xiongqi Wu (Research Assistant) q Jeremy Reed (Research Assistant) q Charles Carpenter (Programmer) q Lowell Pike (Network Administrator) q Woody Marvel (Technical Support) m AT&T Labs Research q Jacobus van der Merwe q Emmanuil Mavrogiorgis m Internet2 q Eric Boyd q Brian Cashman Joint Techs 2011 January 31, 2011

3. University of Kentucky Talk Outline m Motivations for Monitoring GENI Networks m Architecture for a ProtoGENI Monitoring and Instrumentation Tool m Using the Instrumentation Tool m Future Work -- Monitoring Virtual Routers in a ProtoGENI network Joint Techs 2011 January 31, 2011

4. University of Kentucky Introduction and Motivations Joint Techs 2011 January 31, 2011

5. University of Kentucky January 31, 2011 Joint Techs 2011 GENI A national-scale suite of infrastructure for long-running, realistic experiments in Network Science and Engineering. Deeply programmable Virtualized Mobile Wireless Network Edge Site Sensor Network Federated International Infrastructure Programmable, federated, with end-to-end virtualized slices

6. University of Kentucky January 31, 2011 Joint Techs 2011 Source: http://groups.geni.net/geni/attachment/wiki/presentations/protogeni_Ricci_gec3.pdf ProtoGENI m ProtoGENI components m A nationwide high-speed backbone on Internet2’s wave infrastructure m Slicable PCs with NetFPGA cards embedded within the backbone m A set of subnets, including a variety of wireless networks, residential broadband, and programmable edge clusters

7. University of Kentucky Motivations for Monitoring GENI Networks o Much of the work on testbeds has focused on creating, setting up, and running an experiment o Create application-specific virtual networks o Time scale: minutes o The real challenge can often be monitoring and analyzing the behavior of an experiment; it can be a very involved, time consuming, manual process that is repeated many times. o It involves setting up and launching a monitoring environment. Joint Techs 2011 January 31, 2011

8. University of Kentucky Project Goals/Objectives m The ultimate objective is to q Automate the task of deploying an instrumentation and measurement infrastructure within a GENI slice. q Provide users with a convenient and simple-to-use interface to the measurement infrastructure. m Preliminary steps toward this objective q Implement an instrumentation and measurement tools to work in the ProtoGENI environment (INSTOOLS) Joint Techs 2011 January 31, 2011

9. University of Kentucky January 31, 2011 Joint Techs 2011 Monitoring Challenges o Multiple uses o Debug o Verify the configuration o Verify the experiment is behaving correctly o Learn/Discover o Understand what happens and why o Measure performance o Requires o deciding what information should be monitored about the network, OS, and applications o selecting the level of detail for monitoring o monitoring desired information o filtering information to be reported o collecting/transporting recorded data o making information available/viewable to users.

10. University of Kentucky January 31, 2011 Joint Techs 2011 Monitoring Challenges o Scaling to large experiments.  Monitoring load on components  Collect/transport instrumentation data  Without interfering with the experiment  Collecting data from many components o Logging for history view and archival purposes. o Security and authorization to access monitoring capabilities. o Sandboxing the monitoring system. o Accounting for monitoring activity.

11. University of Kentucky Architectrure Joint Techs 2011 January 31, 2011

12. University of Kentucky INSTOOLS Architecture (Functional Components) 1. Setup: deploy and initialize topology-specific software and services 2. Capture: capture measurement data 3. Collection: move data to processing/storage environments 4. Storage: store data on a temporary, short term, long term, and archival basis 5. Processing: filter, convert, aggregate, summarize, etc., data 6. Presentation: present data to users in meaningful ways 7. Access Protection: protect resources and data 8. Measurement Control: Dynamically control the above components Joint Techs 2011 January 31, 2011

13. University of Kentucky INSTOOLS Architecture (Functional Components) 1. Setup: deploy and initialize topology-specific software and services 2. Capture: capture measurement data 3. Collection: move data to processing/storage environments 4. Storage: store data on a temporary, short term, long term, and archival basis 5. Processing: filter, convert, aggregate, summarize, etc., data 6. Presentation: present data to users in meaningful ways 7. Access Protection: protect resources and data 8. Measurement Control: Dynamically control the above components Joint Techs 2011 Conventional network management solutions exist January 31, 2011

14. University of Kentucky INSTOOLS Architecture: (Automated Setup) Joint Techs 2011 Slice 1 Topology Slice 2 Topology Physical Links, VLANs, or GRE Tunnels MC Physical Links, VLANs, GRE Tunnels, or normal IP path MCs are automatically added to collect measurement data January 31, 2011

15. University of Kentucky Implemenation Approach Joint Techs 2011 January 31, 2011

16. University of Kentucky Implemenation Approach Joint Techs 2011 January 31, 2011

17. University of Kentucky Using the Instrumentation Tool Joint Techs 2011 January 31, 2011

18. University of Kentucky Joint Techs 2011 Maps InterfaceScripts Interface ProtoGENI Aggregate AAggregate BAggregate C ProtoGENI System January 31, 2011

19. University of Kentucky Joint Techs 2011 Maps InterfaceScripts Interface ProtoGENI Aggregate AAggregate BAggregate C User’s Slice User creates a “slice” using one of the ProtoGENI APIs. January 31, 2011

20. University of Kentucky Joint Techs 2011 Maps InterfaceScripts Interface ProtoGENI Aggregate AAggregate BAggregate C User’s Slice INSTOOLS Software MC Software SNMP, Netflow, OS Utilities, etc User instruments the slice January 31, 2011

21. University of Kentucky Joint Techs 2011 Maps InterfaceScripts Interface ProtoGENI Aggregate AAggregate BAggregate C User’s Slice INSTOOLS Software MC Software SNMP, Netflow, OS Utilities, etc MC INSTOOLS adds MC to slice January 31, 2011

22. University of Kentucky Joint Techs 2011 Maps InterfaceScripts Interface ProtoGENI Aggregate AAggregate BAggregate C User’s Slice INSTOOLS Software MC Software SNMP, Netflow, OS Utilities, etc MC INSTOOLS adds monitoring software to slice resources January 31, 2011

23. University of Kentucky Joint Techs 2011 Maps InterfaceScripts Interface ProtoGENI Aggregate AAggregate BAggregate C User’s Slice INSTOOLS Software MC Software SNMP, Netflow, OS Utilities, etc MC Web Server Access to MC User views measurement data on the MC. January 31, 2011

24. University of Kentucky Portal for Monitoring January 31, 2011 Joint Techs 2011

25. University of Kentucky Portal for Monitoring January 31, 2011 Joint Techs 2011

26. University of Kentucky Traffic Graphs of a Experimental Node Joint Techs 2011 January 31, 2011

27. University of Kentucky Future Work Joint Techs 2011 January 31, 2011

28. University of Kentucky Future Work Monitoring Virtual Routers m Deploy “virtualizable” commercial routers (Juniper m7i) in the ProtoGENI backbone that support commercial OS/software. m Add software support to these virtual routers that will enable per-slice monitoring and measurement. m Develop tools and interfaces that will allow slice users to use the measurement infrastructure in simple and easy ways. January 31, 2011 Joint Techs 2011

29. University of Kentucky January 31, 2011 Joint Techs 2011 Source: http://groups.geni.net/geni/attachment/wiki/presentations/protogeni_Ricci_gec3.pdf ProtoGENI ShadowNet Sites

30. University of Kentucky ProtoGENI Backbone Node Architecture January 31, 2011 Joint Techs 2011 Sliver 1 NetFPGA Sliver n General Purpose Slivers Gigabit Ethernet Switch Internet 2 Non-sliced PC Sliced PC

31. University of Kentucky ProtoGENI Backbone Node Architecture January 31, 2011 Joint Techs 2011 Logical Router 1 Logical Router 2 Logical Router n Juniper M7i Router Sliver 1 NetFPGA Sliver n ShadowBox Controller Juniper Component Manager Virtual Server perfSONAR 1perfSONAR n General Purpose Slivers Measurement Slivers Gigabit Ethernet Switch Internet 2 Non-sliced PC Sliced PC GENI ShadowNet Router

32. University of Kentucky Availability of Operational/Utilization Data m Infrastructure data: We have the ability to collect physical router data for network operations: q Router/Link up/down status q Link utilization (#bits, #frames, #packets / second) q CPU utilization q Interface (link) error counters q Interface packet loss/drop counters m Slice-specific data q Slice owner controls access to this data q May be made available to network operations January 31, 2011 Joint Techs 2011

33. University of Kentucky Joint Techs 2011 Thank You! Questions? This material is based upon work supported in part by the National Science Foundation under grant number CNS-0834243. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of GPO Technologies, Corp, the GENI Project Office, or the National Science Foundation. January 31, 2011