Constructing Energy-Aware Software- Defined Network Virtualization

Transcription

1 Software- Defined Network Virtualization Kuala Lumpur, 08/2015 Tran Manh Nam, Nguyen Huu Thanh, Nguyen Hong Van, Kim Bao Long, Nguyen Duc Lam, Nguyen Van Huynh, Nguyen Van Ca 1

2 Content 1. Introduction 2. Background 1. Software-Defined Networking 2. Network Virtualization 3. Virtual Network Embedding VNE 4. Network HyperVisor - FlowVisor 3. Energy-Aware Software-Defined Network Virtualization System 1. Architecture 2. VNE algorithm 4. Conclusion 2

3 1. Introduction Internet services are booming and continue to play an increasingly vital role in supporting key fields of research, business, news, education Significant energy consumption [*] Big carbon emission (2000kWh ~ 1 ton of CO2) the annual growth of electricity consumption of communication network is 10,4% [1] In 2007, the total electricity consumption in communication network was 200 TWH, reaching 330 TWH by 2012 ICT equipment, its use and disposal account for 2% of global CO2 emissions, which is equivalent to the contributions from the aviation industry [2]. * - Future Internet: A Survey of Existing Approaches and Trends in Energy-Aware Fixed Network Infrastructures - IEEE (Volume:13, Issue: 2), 2nd Quarter

4 1. Introduction The recent trend of energy consumption over the years has presented some profound negative impacts, such as: (1) Increased production cost of enterprises, putting upward pressures on product prices; (2) Growing carbon emission that severely hurt the surrounding environment. The ICT should be Green and energy-aware: Software-Defined Networking (SDN) and Network Virtualization is therefore to be presented as a technology with great potential in terms of green networking 4

5 2. Background For Green ICT, we build the Energy-Aware Software-Defined Network Virtualization system that allows virtualize the SDN networks. In the other hand, we contribute the Energy-Aware VNE Alogrithm. In this work, the following technologis that we used: 1. Software-Defined Networking 2. Network Virtualization 3. Virtual Network Embedding VNE 4. Network HyperVisor - FlowVisor 5

6 2.1. Software-Defined Networking (SDN) Software-Defined Networking SDN Architecture Decouples the control plane and data plane of networking devices Enables the control plane to become directly programmable Centralized monitoring and controlling Network protocols Control Programs Global Network View Network Operating System Control system over the network Network protocols 6

7 2.1. Software-Defined Networking (SDN) OpenFlow Controller OpenFlow Protocol (SSL/TCP) Control Path Path (Software) OpenFlow OpenFlow Protocol OpenFlow Controller: POX with python, Floodlinght, OpenDayLight OpenFlow Protocol: packet-received send-packet-out modify-forwarding-table get-stats. Data Path (Hardware) 7

8 2.2. Network Virtulization Network Virtualization - coexistence of multiple virtual networks on the same physical substrate network. Greatest Advantage: sharing of resources among heterogeneous logical virtual networks VNoM: a virtual node may map to several substrate nodes VLiM: virtual link spanning to several substrate links 8

9 2.3. Virtual Network Embedding - VNE VNE mapping virtual network to substrate network VNE problem - reach the maximum exploitation of system resources. It includes Virtual Node Mapping (VNoM) and Virtula Link Mapping (VLiM): VNoM maps virtual nodes onto suitable substrate nodes. VLiM maps virtual links onto substrate links. Two virtual networks mapped onto one substrate network [18] 9

10 2.4. FlowVisor Network Hypervisor acts as transparent proxy between forwarding elements and Controllers FlowVisor [8] Transparency ability to prototype and debug protocols on realistic topologies decouple network virtualization technology from controller Isolation Bandwidth Topology CPU FlowSpace Flow Entries OpenFlow Control Isolation 10

11 2.4. FlowVisor Message Handling Alice Controller Bob Controller Cathy Controller Rule Policy Check: Is this rule allowed? Full Line Rate Forwarding OpenFlow FlowVisor OpenFlow OpenFlow Firmware Exception Policy Check: Who controls this packet? Packet Data Path 11

12 Software-Defined Virtual Network Requests (VNRs) Openflow Controllers (POX, FloodLight) CTL 1 CTL 2 Slicer Slicing, port mapping, Switches, Controllers coordinating and configuring Controller management Control and configure Controllers Management GUI Java-based Application SDVNR 1 (CTL 1:POX) SDVNR 2 (CTL 2:FloodLight) Extended FlowVisor Power VNE Algorithms - CG VNE - HEE VNE Power monitoring: monitor NW power consumption Power controlling: Control NW devices state -Using NetFGPA energy model- 3. Energy-Aware Software- Defined Network Virtualization System Four Elements: Management; OpenFlow Controllers; Extended FlowVisor Slicer Power Substrate Network. SDVNR 1 SDVNR 2 Substrate network (Openflow-enable Switches) 12

13 3. EA SDNV System - WorkFlow 1) Step 1: Identifying inputs: Management identifies the set of SDVNR and substrate network which contain following detail information 2) Step 2: Virtual Network Embedding: After receiving a set of SDVNRs, the VNE calculates and gives results that contain both successful virtual node mapping (VNoM) and virtual link mapping (VLiM) results 3) Step 3: Slicer and Power control: The FlowVisor configures virtual networks by creating flowspaces and allocates resources to them. 4) Step 4: Configuring OpenFlow Controllers: Based on the successful SDVNR results, the OpenFlow Controllers block configures corresponding controllers and connects them to FlowVisor for their communicating to SDVN. 5) Step 5: Finish and monitoring Software-Defined Virtual Network Requests (VNRs) Openflow Controllers (POX, FloodLight) CTL 1 CTL 2 Slicer Slicing, port mapping, Switches, Controllers coordinating and configuring SDVNR 1 Controller management Control and configure Controllers Management GUI Java-based Application SDVNR 1 (CTL 1:POX) SDVNR 2 (CTL 2:FloodLight) Extended FlowVisor Power VNE Algorithms - CG VNE - HEE VNE Power monitoring: monitor NW power consumption Power controlling: Control NW devices state -Using NetFGPA energy model- SDVNR 2 Substrate network (Openflow-enable Switches) 13

14 3. EA SDNV System Energy model Net-FPGA configuration Power (mw) Static (baseline) 4496 FPGA Core At least one 1Gbps 2694 One port The 4-Gigabit-port NetFPGA [4] 48-Ports Pronto 3240 [6] Ethernet port Not all ports running at 1Gbps (10/ 100Mbps) 1394 Idle 81 Idle 23 10Mbps Mbps Mbps 1080 Configuration Power (mw) P static P 10Mpbgs per port 63 P 100Mpbgs per port 260 P 1000Mpbgs per port 913 Switch s profile: Psw = Pstatic + Nport*Pport + [P FPGACore ] Because of a lack of OpenFlow s state-control mechanism, we extend the OpenFlow v1.0 protocol by changing a Vendor message 14

15 3. EA SDNV System - GUI 15

16 3. EA SDNV System VNE Algorithms Algorithm 1: Greedy Node Mapping 1. input: T S (E S, N S ) and SDVNR(E R, N R ) 2. begin 3. //Sort N S by CPU in non-decreasing order 4. N S sort CPU (N S ) 5. for all n R in N R do 6. for all n S in N S do 7. if V n n S N nei n R then 8. continue 9. else if C a n S R C n R then 10. M N = M N + {n S n R } 11. SUCCESS 12. break 13. end if 14. end for 15. end for 16. end 17. output: T S (E S, N S ), M N Greedy VNoM Algorithm 2: HEE Node Mapping 1. input: T S (E S, N S ) and SDVNR(E R, N R ) 2. begin 3. //Sort N S by N nei (N S ) in nondecreasing order 4. N S sort nei (N S ) 5. for all n R in N R do 6. for all n S in N S do 7. if V n n S N nei n R then 8. continue 9. else if C a n S R C n R then 10. M N = M N + {n S n R } 11. SUCCESS 12. end if 13. if n S is off then 14. turn on n S 15. N n. push(n S ) 16. N S. remove(n S ) 17. N n. push(n nei (n S )) 18. N S. remove(n nei (n S )) 19. N S N n. push(n S ) 20. end if 21. break 22. end for 23. end for 24. end 25. output: T S (E S, N S ), M N Heuristic Energy-efficient VNoM (HEE) algorithm that realizes node mapping in order of priority as follows: - Maps a virtual node on the already active substrate node - Maps onto the neighboring substrate nodes that have the biggest number of connected neighbors N nei (N S ) 16

17 3. EA SDNV System - Evaluation Acceptance ratio s comparison Energy consumption ratio s comparison 17

18 Conclusion For the Green ICT, using SDN with NV is good approach. And in this paper we successfully deploy the energy-aware Software-Defined Network Virtualization system that allows: monitoring energy consumption controlling energy consumption of the substrate network by changing the states of physical devices We also define the HEE mapping algorithm that reduce energy consumption of the network while maintains the acceptance ratio In the near future, we are going to develop a system for QoS and fault- tolerance evaluation 18

19 References 1. Ward Van Heddeghem, Sofie Lambert, Bart Lannoo, Didier Colle, Mario Pickavet, Piet Demeester, Trends in worldwide ICT electricity consumption from 2007 to 2012, Computer Communications, Volume 50, 1 September 2014, Pages Global Action Plan, An Inefficient Truth. Global Action Plan Report, Dec Diego Kreutz, Fernando M. V. Ramos, Paulo Verissimo, Christian Esteve Rothenberg, Siamak Azodolmolky, and Steve Uhlig, Software-Defined Networking: A Comprehensive Survey, 4. N.M. Mosharaf Kabir Chowdhury, Raouf Boutaba, A survey of network virtualization, Computer Networks 54 (2010) OpenFlow Switch Specifiation v1.5, Open Network Foundation, September 27, Available: 6. POX Floodlight R. Sherwood, G. Gibb, K. Yap, G. Appenzeller, N. McKeown, and G. Parulkar, FlowVisor: A network virtualization layer OpenFlow Switch Consortium, Tech. Rep. OPENFLOW-TR , [Online]. Available: 9. Rob Sherwood, Michael Chan,Glen Gibb, Nikhil Handigol,Te-Yuan Huang,Peyman Kazemian, Masayoshi Kobayashi, David Underhill, Kok-Kiong Yap, Guido Appenzeller, and Nick McKeown, Carving Research Slices Out of Your Production Networks with OpenFlow Newsletter ACM SIGCOMM Computer Communication Review archive Volume 40 Issue 1, January 2010 Pages Exploring networks of the future. [Online]. Available: Ofelia, OpenFlow in Europe. [Online]. Available: OF@TEIN - OpenFlow@Trans-Eurasian Information Network, [Online] T. Anderson, L. Peterson, S. Shenker, and J. Turner, Overcoming the internet impasse through virtualization, Computer, vol. 38, pp , April N. Feamster, L. Gao, and J. Rexford, How to lease the internet in your spare time, ACM SIGCOMM Computer Communication Review, vol. 37, pp , N. M. K. Chowdhury and R. Boutaba, A survey of network virtualization, Computer Networks, vol. 54, no. 5, pp , A. Bavier, N. Feamster, M. Huang, L. Peterson, and J. Rexford, In VINI Veritas: realistic and controlled network experimentation, SIGCOMM Comput. Commun. Rev., vol. 36, pp. 3 14, August D. Schwerdel, D. Gunther, R. Henjes, B. Reuther, and P. M uller, German-lab experimental facility, in Future Internet - FIS 2010, ser.lecture Notes in Computer Science, A. Berre, A. Gomez-Perez, K. Tutschku, and D. Fensel, Eds., vol Springer Berlin /Heidelberg, 2010, pp. 1 10, / _ Andreas Fischer, Juan Felipe Botero, Michael Till Beck, Hermann de Meer, and Xavier Hesselbach, Virtual Network Embedding: A Survey, IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 15, NO. 4, FOURTH QUARTER 2013, DOI: /SURV

20 Tran Manh Nam - Future Internet Laboratory School of Electronics and Telecommunications Hanoi University of Science and Technology (HUST) - Hanoi, Vietnam [email protected] Skype: namtm.vn Energy-aware routing - namtm 20