未来互联网体系结构研究综述（PPT讲稿）Future Internet Architecture：From Network Virtualization to Clean- slate Post-IP

EPSRC University of Essex Ploneering ressarch SEVENTII FRAMCWORK and skills PROGRAMMC Future Internet Architecture:From Network Virtualization to Clean- slate Post-IP Kun Yang(杨鲲) University of Essex,Colchester,UK Thursday 21 March 2013 Xidian U.,Xi'an,China 1

F t I t t A hit t F F u ture I n terne t Architec ture: From Network Virtualization to Clean Network Virtualization to Clean￾slate Post slate Post-IP Kun Yang (杨鲲 ) U i i f E C l h UK U n ivers i t y o f Essex, C o l c hester, UK Thursday 21 March 2013 d ’h 1 @ Xi dian U., Xi ’an, C hina

Agenda ■Motivation:why FIA? Network Virtualization Post-IP Future Internet Architecture Converged Network Service Test- bed ■Q&A 2

Agenda M ti ti h FIA? Motivation: why FIA? Network Virtualization Post-IP Future Internet Architecture Converged Network & Service Test Converged Network & Service Test￾bed Q&A 2

Challenges faced by Current Internet IoM(Internet of Media),IoT(Internet of Things),IoS (Internet of Services)have pushed Internet to its limit in terms of: Performance: 0.5Million backbone routers,9M BGP routes,1T webpage items on Google,15hours video content added into video websites per minute, ..60%increase in Internet traffic flow p.a. Furthermore,IoT makes billions and trillions of"things"connected into the Internet! The Internet becomes huge and more complex. ■ Security: e.g.,90M visa card users hacked in Europe and North America at 2007; DNS server of Storm (Video)Co.being attacked at 2009,leading to 15 provinces'network malfunction in China. 3

Challenges faced by Current Internet Challenges faced by Current Internet IoM ( ), ( g ), ( Internet of Media), IoT (Internet of Thin g s), IoS (Internet of Services) have pushed Internet to its limit in terms of: Performance:  0.5Million backbone routers, 9M BGP routes, 1T webpage items on Google, 15hours video content added into video websites per minute, … 60% increase in Internet traffic flow p a 60% increase in Internet traffic flow p.a.  Furthermore, IoT makes billions and trillions of “things” connected into the Internet!  The Internet becomes huge and more complex. Security:  e.g., 90M visa card users hacked in Europe and North America at 2007;  DNS server of Storm (Video) Co. being attacked at 2009, leading to 15 3 DNS server of Storm (Video) Co. being attacked at 2009, leading to 15 provinces’ network malfunction in China

Challenges faced by Current Internet Energy Consumption: ICT energy consumption in 2007 in China equals to the energy generated by Three-Gorge Power Station 1 Google search 11W bulb for 1 hour ■Green！ Network Management: Becoming more difficult and almost unmanageable. including content management(e.g.,piracy) 4

Challenges faced by Current Internet Challenges faced by Current Internet Energy p Consum ption:  ICT energy consumption in 2007 in China equals to the energy generated by Three-Gorge Power Station  1 Googl h b lb f h le searc h = 11W b ulb for 1 hour  Green! Network Management: Network Management:  Becoming more difficult and almost unmanageable.  including including content content management management (e.g., piracy) piracy) 4

Path to Future Internet:Dirty Slate Solutions:back-compatible,evolutionary,incremental Quality of Service problem?Then IntServ/DiffServ More efficient routing?->MPLS(Multi-Protocol Label Switching) Security problem?Then IP VPN:Virtual Private Network Not enough IP addresses?Then IPv6. Mobility problem?Then Mobile IP or MIPv6 ■Initiatives: NGI,Internet2 at US,Ambient Network in EU, APAN in Japan,CNGI in China. ■Problems: Adding patches and making the Internet more complex and thus more difficult to achieve global optimization. Limiting the further development of the Internet. 5

Path to Future Internet: Dirty Slate Path to Future Internet: Dirty Slate Solutions: back-compatible, evolutionary, incremental  Quality of Service problem? Then IntServ/DiffServ  More efficient routing? -> MPLS (Multi-Protocol Label Switching)  Security problem? Then IP VPN: Virtual Private Network  Not enou gh IP addresses? Then IPv6.  Mobility problem? Then Mobile IP or MIPv6 Initiatives:  NGI, Internet2 at US, Ambient Network in EU,  APAN in Japan, CNGI in China. P bl ro ems:  Adding patches and making the Internet more complex and thus more difficult to achieve global optimization 5 thus more difficult to achieve global optimization.  Limiting the further development of the Internet

Path to Future Internet:Clean-slate Revolutionary (bold!) ■ Do endpoints really matter?Or is it the information that is more important? IP may be "the"problem! Endpoint-centric services move towards information retrieval through,e.g.,CDNs(Content Distribution Networks),youTube,Twitter But how about at the network layer? You BitTorrent WORLD Tube f 6

Path to Future Internet: Clean Path to Future Internet: Clean -slate Revolutionar y( ) bold! ) Do endpoints really matter? Or is it the information that is more im portant?  IP may be “the” problem! Endpoint-centric services move towards information retrieval through, e.g., CDNs (Content Distribution N k) T b T i Networ k s ), you T u be, T w itter  But how about at the network layer? 6

Ossification of IP-based Infrastructure Huge innovation- email WWW in applications phone... SMTP HTTP Ossification BIP of the core protocols IP Relentless evolution ethernet PPP. of the underlying technology CSMA async Courtesy to M.Handley of UCL conner fiber

Ossification of IP Ossification of IP -based Infrastructure based Infrastructure email WWW phone... Huge innovation in applications SMTP HTTP Ossification RTP... of the core TCP UDP… protocols IP Relentless evolution ethernet PPP… CSMA as ync ion g Relentless evolution of the underlying technology 7 y sonet... copper fiber Courtesy to M. Handley of UCL

Observations The current Internet is end point centric which has contributed to a number of its problems,such as SPAM,virus,denial-of-service attacks, "poor support for mobility and multimedia content distribution. The current Internet is in favour of senders and receivers passively receive. Economically,receivers are forced to carry the cost of unwanted traffic Google youTube's content distribution networks get bigger and more expensive(also for ISPs). There is a consensus that a fundamental reform of the Internet is inevitable in the near future. 8

Observations The current Internet is end point centric which has The current Internet is end point centric which has contributed to a number of its problems, such as  SPAM, virus, denial SPAM, virus, denial-of-service attacks, service attacks,  poor support for mobility and multimedia content distribution. The current Internet is in favour of senders and receivers passively receive.  Economically, receivers are forced to carry the cost of unwanted t ffi ra c Google youTube’s content distribution networks get bigger and more expensive (also for ISPs) and more expensive (also for ISPs). There is a consensus that a fundamental reform of the Internet is inevitable in the near future. 8 Internet is inevitable in the near future

What is the Way Out? g ■UV Cost Cost Clean Dirty rrequent scalable Slate VS Slate Update 里n Easy to Deploy Support Deploy Arch Reduce Innovation Deployment More Feasible? Complexity Net work Virtualization 9

What is the Way Out? What is the Way Out? High C t Low Cos t C t os t Not scalable Frequent Update Clean Slate Dirty Slate Hard to Eas y to Deploy Deploy M F ibl ? Net work More Feasibl e ? 9 Net work Virtualization

Agenda ■Motivation:why FIA? Network Virtualization Post-IP Future Internet Architecture Converged Network Service Test- bed ■Q&A 10

Agenda M ti ti h FIA? Motivation: why FIA? Network Virtualization Post-IP Future Internet Architecture Converged Network & Service Test Converged Network & Service Test￾bed Q&A 10