Changes between Version 13 and Version 14 of GEC20Agenda/EveningDemoSession
- Timestamp:
- 06/24/14 13:02:53 (9 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
GEC20Agenda/EveningDemoSession
v13 v14 44 44 ==== Measurements of Netmap-Based Data Plane Acceleration on xDPd ==== 45 45 46 The poster will illustrate and report our current measurement results in characterizing the performance improvements achieved through netmap integration with xDPd. xDPd is an open source, modular software switch project: https://www.codebasin.net/redmine/projects/xdpd/wiki. We recently started to collaborate with the FIRE team in Berlin, BISDN, on data plane acceleration for xDPd through netmap integration. Our measurements are in progress as xDPd also evolves. We also would like to receive GENI community feedback on such measurement challenges. 46 The poster will illustrate and report our current measurement results in characterizing the performance improvements achieved through netmap integration with xDPd. xDPd is an open source, modular software switch project: https://www.codebasin.net/redmine/projects/xdpd/wiki. We recently started to collaborate with the FIRE team in Berlin, BISDN, on data plane acceleration for xDPd through netmap integration. Our measurements are in progress as xDPd also evolves. We also would like to receive GENI community feedback on such measurement challenges. Poster is attached [http://groups.geni.net/geni/attachment/wiki/GEC20Agenda/EveningDemoSession/GEC20NETMAPmeasurementchallenges.pdf here]. 47 47 48 48 Participants: … … 52 52 ==== Integrating Dell SDP on GENI ==== 53 53 54 The poster will present the state of the integration on GENI, our ideas and considerations on an aggregate manager for the Split Data Plane (SDP) switches on GENI. Dell has decided to donate ~10 SDP switches on GENI. UH is placing them on GENI rack locations, both ExoGENI and InstaGENI. The SDP switches should be considered GENI resources. However, there are challenges in sharing such a resource for simultaneous experimentation. We would like to present such considerations and receive discussions and feedback from GENI community. 54 The poster will present the state of the integration on GENI, our ideas and considerations on an aggregate manager for the Split Data Plane (SDP) switches on GENI. Dell has decided to donate ~10 SDP switches on GENI. UH is placing them on GENI rack locations, both ExoGENI and InstaGENI. The SDP switches should be considered GENI resources. However, there are challenges in sharing such a resource for simultaneous experimentation. We would like to present such considerations and receive discussions and feedback from GENI community. Poster is attached [http://groups.geni.net/geni/attachment/wiki/GEC20Agenda/EveningDemoSession/GEC20SDPonGENI.pdf here]. 55 55 56 56 Participants: … … 133 133 ==== Simulation, Detection, and Denial of ping Attack ==== 134 134 135 The demo topology consist of 4 nodes which includes two hosts, Host1 and Host2, one !OpenvSwitch (OVS) node and one Floodlight Controller. The hosts are connected to the OVS and OVS is connected to the Floodlight controller. Host1 will perform ping attack to Host2. Host2 will continuously measure the number of pings received from Host1 and detect the ping attack when the frequency and number of pings exceed a threshold. The detection will trigger re-programming of the network through OVS to deny the pinging connection. All ping attack detection should be monitored using !LabWiki and the re-programming of the network is through the GIMI framework. 135 The demo topology consist of 4 nodes which includes two hosts, Host1 and Host2, one !OpenvSwitch (OVS) node and one Floodlight Controller. The hosts are connected to the OVS and OVS is connected to the Floodlight controller. Host1 will perform ping attack to Host2. Host2 will continuously measure the number of pings received from Host1 and detect the ping attack when the frequency and number of pings exceed a threshold. The detection will trigger re-programming of the network through OVS to deny the pinging connection. All ping attack detection should be monitored using !LabWiki and the re-programming of the network is through the GIMI framework. Poster is attached [http://groups.geni.net/geni/attachment/wiki/GEC20Agenda/EveningDemoSession/GEC20SimulationofpingattackonGIMI.pdf here]. 136 136 137 137 Participants: … … 280 280 ==== End-to-End Flow Manipulations ==== 281 281 282 Demonstrate how end-to-end flow manipulations can be performed on programmable OpenFlow-based switches. The purpose is to create a framework for network engineers to utilize as a network troubleshooting/debugging tool. Network debugger uses flow separation at aggregation points to increase granularity in monitoring by using end point analysis tools as debugging points. We will create an environment on GENI to show the framework and its proof-of-concept viability as a network troubleshooting tool. 282 Demonstrate how end-to-end flow manipulations can be performed on programmable OpenFlow-based switches. The purpose is to create a framework for network engineers to utilize as a network troubleshooting/debugging tool. Network debugger uses flow separation at aggregation points to increase granularity in monitoring by using end point analysis tools as debugging points. We will create an environment on GENI to show the framework and its proof-of-concept viability as a network troubleshooting tool. Poster is attached [http://groups.geni.net/geni/attachment/wiki/GEC20Agenda/EveningDemoSession/GEC20networkdebugging.pdf here]. 283 283 284 284 Participants: … … 288 288 ==== Measurements of Forwarding Performance of the LINCx Switch ==== 289 289 290 This project mainly focuses on the performance measurement of the LINCx software switch ( https://github.com/FlowForwarding/lincx) as well as investigations for performance improvements in the forwarding systems. LINCx switch has its flow table hardcoded instead of using ETS table as seen in regular LINC switch. Its flow table code is automatically generated based on the flow instruction message from controller, and recompiled as a module to keep the switch operating without interrupting. This demo will measure the forwarding performance of LINCx after recompilation with different sets of flows. 290 This project mainly focuses on the performance measurement of the LINCx software switch ( https://github.com/FlowForwarding/lincx) as well as investigations for performance improvements in the forwarding systems. LINCx switch has its flow table hardcoded instead of using ETS table as seen in regular LINC switch. Its flow table code is automatically generated based on the flow instruction message from controller, and recompiled as a module to keep the switch operating without interrupting. This demo will measure the forwarding performance of LINCx after recompilation with different sets of flows. Poster is attached [http://groups.geni.net/geni/attachment/wiki/GEC20Agenda/EveningDemoSession/GEC20_LINCx_Measurement.pdf here]. 291 291 292 292 Participants: