Aiding Open Flow Controller by Enhancing Open Flows
Aiding Open. Flow Controller by Enhancing Open. Flow's Control Model, and Behaviour of Flows Othman M. M. , Koji Okamura Kyushu University Proceedings of the TNC 2013 Conference 6 June 2013, Maastricht, Netherlands 1
Outline: 1. Motivation and Goal. 2. An attempt to solve the problem (3 enhancements). 1. First: Network Equipment to Equipment flow installation. 2. Second: Proactive Flows. 3. Evaluation. 4. Conclusion. 2
1 -Motivation and Goal. 1. Controller might be bottleneck. �Tight coupling between Open. Flow switch and controller. �Every thing is up to the controller. �[1] “When using Open. Flow in high speed networks with 10 Gbps links, today’s controller implementations are not able to handle the huge number of new flows. ” 2. Equipment can also be a bottleneck. �Limited resources �Flow Table size. �Processing �etc. Michael Jarschel, Simon Oechsner, Daniel Schlosser, Rastin Pries, Sebastian Goll, and Phuoc Tran-Gia. 2011. “Modeling and performance evaluation of an Open. Flow architecture”. In Proceedings of the 23 rd International Teletraffic Congress (ITC’ 11). ITCP 1 -7. 3
1 -Motivation and Goal. �Considering Future Internet with many applications supported by Open. Flow. � Specific routing, � Video streaming, � Security, � Qo. S. � etc. �Controller will have many tasks. Application/service provider. Security policy maker/manager Figure 1: 4
1 -Motivation and Goal. �Improve Open. Flow. �Support self-reactive behavior. �Reduce load on controller. �Step towards having wider adoption of Open. Flow. �If Open. Flow is thought of as one of the Future Internet technologies, However some debate that Open. Flow have some limitations. �So our aim is to aid Open. Flow, to make it more suitable for wider adoption and implementation within networks or in whole Open. Flow networks. 5
2 -An attempt to solve the problem. 1. Network equipment to Network equipment Flow Programming: �To create self-reactive network. �Can be used to delegate some flows to less loaded network equipment. �To easily program whole network without loading controller. 2. New type of Flows (Proactive Flows): � To provide the controller with a more relaxed way to handle precisely timed tasks. � Programed as inactive flows and later activated by the flow. � Can cooperate with Device to Device programming. 6
�Current Open. Flow’s control model: �Controller to Equipment only: Equipment exchange information only with the controller. �Equipment to Equipment only: equipment exchange information with each other. �Controller to Equipment, AND Equipment to Equipment: to give Open. Flow the ability to exchange information between equipment in addition to controller. Figure 3: Enhanced Open. Flow Control Mode �Target: Fig 2. Regular Network Information exchange. �Current Internet: Figure 1: Open. Flow Control Mode 2 -1 - Network Equipment to Equipment flow installation. 7
2 -1 - Network Equipment to Equipment flow installation. �To reduce load off the controller. Flows to manipulate headers in packets Packet P P Packet PE �Give the equipment ability to act by their own to reduce load off loaded equipment. Packet P PE Flows to manipulate headers in packets P Packet PE Packet P P Fig 1. Equipment overloaded, due to many flows to carry out. Fig 2. Overloaded equipment delegates some flows to other equipment. Fig 3. Reduced load off the overloaded equipment. Packet 8
2 -1 - Network Equipment to Equipment flow installation. �Alternative way to install flows to whole network (e-e propagation). �To reduce load off the controller. Controller Fig 1. Regular way of installing flows. Controller installs to equipment one by one. Controller Fig 2. Network equipment install flows to each other. 9
2 -2 - Proactive Flows. �Original Open. Flow: flows activated by default, �controller keeps track of time and install flows on time. Fig 1. Migration and Redirection using Open. Flow. Fig 2. Delay due to controller overload in Migration ? 10
2 -2 - Proactive Flows. �Proactive Flows: �Initially installed as inactive. (not usable). �Activated on right time, by: Fig 1. Migration and Redirection using Open. Flow and Inactive Flows. � Explicit activation packet. � Activation Flow. � Preset time. �Controller can install the flow ahead of time. Proactive Flows Migration Flows activation Migration 11
3 - Evaluation: �Run simulation on OMNet++ using : �Regular Open. Flow. �Modified Open. Flow. 12
3 -Evaluation: 1 -Ne-Ne FI for distributing flows on behalf of the controller. �Scenario. �Shown in figure 1. �To measure: 1. Totoal_Ne. FI_install_time: Figure 1. (a) Regular flow installation. (b) The flow installation using Ne-Ne FI. � the total time needed for the flows to be installed on the whole set of required equipment. � (1) � T(e) represents the time at which the flow installation reaches equipment e where eϵE , and T(0) is the time at which the controller imitated the Ne-Ne FI method. 13
3 -Evaluation: 1 -Ne-Ne FI for distributing flows on behalf of the controller. �Scenario. �Shown in figure 1. �To measure: 2. Totoal_number_of_Ne. FI_messages: Figure 1. (a) Regular flow installation. (b) The flow installation using Ne-Ne FI. � total number of packets exchanged in order to enable the Ne-Ne FI installation on behalf of the controller. � � MNe-Ne FI(e) is the number of all messages belonging to the Ne-Ne FI sent by equipment e where eϵE. 14
3 -Evaluation: 1 -Ne-Ne FI for distributing flows on behalf of the controller. �Scenario. �Shown in figure 1. �To measure: 3. Totoal_size_of_Ne. FI_messages : Figure 1. (a) Regular flow installation. (b) The flow installation using Ne-Ne FI. � total size of packets exchanged in order to enable the Ne- Ne FI installation on behalf of the controller. � � SNe-Ne FI (e) is the size of all messages belonging to the Ne-Ne FI sent by equipment e, where eϵE. 15
3 -Evaluation: 1 -Ne-Ne FI for distributing flows on behalf of the controller. �Scenario. �Shown in figure 1. �Compare all to regular case. Figure 1. (a) Regular flow installation. (b) The flow installation using Ne-Ne FI. 16
3 -Evaluation: 2 -Delegating Flows off the Overloaded Equipment using Ne-Ne FI �Scenario: �The concept of overloading: � � Can be TCAM memory is running out. Or other resources. �To Measure: 1. Ratio_of_overloading_pe: Figure 1. (a) Equipment overloaded, due �the ratio of load of the edge equipment to many flows to carry out. (b) overloaded over the average load of the other network equipment delegates some flows to other equipment. (c) Reduced load off the overloaded equipment. � � Lavg(e) represents the average load of equipment e , where eϵN. 17
3 -Evaluation: 2 -Delegating Flows off the Overloaded Equipment using Ne-Ne FI �Scenario: �The concept of overloading � � Can be TCAM memory is running out. Or other resources. �To Measure: 2. Time_to_reduce_pe_load: � the time needed to reduce load off the overloaded equipment. Figure 1. (a) Equipment overloaded, due to many flows to carry out. (b) overloaded equipment delegates some flows to other equipment. (c) Reduced load off the overloaded equipment. � � Te(load value) represents the instance of time at which equipment e, where eϵN, has reached the specified load value. 18
3 -Evaluation: 3 - Proactive Flows. �Scenario: Figure 1. (a) Migration and Redirection using regular Open. Flow. (b) Migration and Redirection using Open. Flow and Proactive Flows. 19
3 -Evaluation: 3 - Proactive Flows. �Scenario: �To Measure: �Tredir which represents the time at which the controller installs the redirection flows, and compare it with that Treq, which is the time at which the first request will arrive to the server after it completes the migration. Figure 1. (a) Migration and Redirection using regular Open. Flow. (b) Migration and Redirection using Open. Flow and Proactive Flows. 20
3 -Evaluation: 3 - Proactive Flows. �Scenario: �To Measure: Figure 1. (a) Migration and Redirection using regular Open. Flow. (b) Migration and Redirection using Open. Flow and Proactive Flows. �Treq, and Tproactive while also counting the number of unanswered request. �And finally we will compare the number of unanswered requests in the cases of regular Open. Flow and the proactive flows. 21
4 - Conclusion: �Aim to aid Open. Flow by reducing load off the controller, make Open. Flow’s equipment self-aware and self-reactive. �Achieving goals by proposing a new enhancements to Open. Flow: �Network equipment to equipment Flow Installation. �Proactive Flows. 22
Q&A: �Thanks for listening 23
- Slides: 23