1、the Internet as emerging critical infrastructure: what needs to be measured?cooperative association for internet data analysis30 november 2005presented to lsn jet nsf (http:/www.nitrd.gov/subcommittee/lsn/jet/)kccaida.orgoutline of talkwhat is critical infrastructuretop problems of Internethistorica
2、l context (incongruity)what have we learned and how can we apply it? case study: scalability (separate talk) what we (all) can do to helpcritical infrastructurewhat is it? how does it get that way?what are common characteristics?is the Internet one? or will it be soon?what are the implications for p
3、ublic and private sectors? underlying goals: innovation, economic strength, democracy, freedom, health, science, arts, society. it really is about living in a better world.16 operational internet problemssecurityauthenticationspamscalable configuration management robust scalability of routing system
4、compromise of e2e principledumb networkmeasurementpatch management“normal accidents”growth trends in traffic and user expectationstime management and prioritization of tasks stewardship vs governanceintellectual property and digital rightsinterdomain qos/emergency services inter-provider vendor/busi
5、ness coordinationtop Internet problemspersistently unsolved problems for 10+ years(see presentations at www.caida.org )why were not making progresstop unsolved problems in internet operations and engineering are rooted in economics, ownership, and trust (EOT).even the most theoretical computer scien
6、tists are convinced.does not mean there arent useful technical problems to study. but there will be no technical solutions to these problems that dont solve the EOT issues.top Internet problemshistorical context1966: Larry Roberts, “Towards a Cooperative Network of Time-Shared Computers” (first ARPA
7、NET plan)(we are still using the same stuff)1969: ARPANET commissioned by DoD for research1977: Kleinrocks paper “Hierarchical Routing for large networks; performance evaluation and optimization”(we are still using the same stuff)1980: ARPANET grinds to complete halt due to (statusmsg) virus1986: NS
8、FNET backbone, 56Kbps. NSF-funded regionals. IETF, IRTF. MX records (NAT for mail)1991: CIX, NSFNET upgrades to T3, allows .com. web. PGP. 1995: under pressure from USG, NSF transitions backbone to competitive market. no consideration of economics or security. kc proposes caida.org2005: The Economis
9、ts cover story: “How the Internet killed the phone business” (September)what have we done?we replaced a critical infrastructure with something not designed to be critical infrastructure historical context explains it but does not address incongruitiesand this decade, free markets go up against free
10、speech what have we learned?most important thing weve learn so far: society has decided IP is like water.strong implications for an industry structuring itself to sell wine. but thats what the data shows.when you want to move water, you care about 4 things: safe, scalable, sustainable, stewardship.t
11、he 4 Sssafety: is the data toxic upon arrival? scalable: can we route/name/address earths needs? sustainable: is it economically viable?stewardship: will the provisioning and legal frameworks we choose leave our children - and democracies - better or worse off?none of these are purely technical issu
12、es,but they all require deep technical (among other) understanding to get right.and theyre all connected.how have we done?how safe is the Internet? data doesnt look goodhow scalable is the Internet?data doesnt look goodhow sustainable is the Internet? data doesnt look goodhow did we do on stewardshi
13、p?data doesnt look goodfailure (to measure progress) on 4Ss poses risks to economics and democracies:that we wont learn from our own history. e.g.,not only dont we understand the economics, but we dont understand that we dont understand the economics, and thus must set policy based on unvalidated as
14、sumptionsthat we will design another architecture with no actual plan for economic sustainability (much less incenting further innovation in a competitive market!)that other forces will “code” innovation into the architecture (free markets vs free speech)there is good newswe made something so great,
15、 everyone wants it. in fact many of us want it more than once! (um)the current industry is a historical artifact of technical and (science routeviews (BGP)the 4 Ss and addressing crisissafety: no perceived increasescalable: current hacks offer local optimasustainable: IPv6 economically viablesteward
16、ship: how to get unused addresses back? none of these are purely technical issues,but they all require deep technical (among other) understanding to get right.and theyre all connected.other provisioning crisesbackbone provisioning: free marketDNS: free marketaddress space: _? what do we do in provis
17、ioning crises?and how successful have we been? backbone provisioning: profits toward (& below) zero, consolidation toward monopoly, no security, no innovation.DNS: profits toward zero, consolidation, no security, unfinished internationalization. some innovation (“sitefinder”)address space: _? how ha
18、ve free markets been to infrastructure?“The design was immediately and grievously flawed for not only was there no plan for the privatization and no criteria by which to measure its success or failure. Furthermore it held unacknowledged economic implications for what was being privatized was the onl
19、y part of the network that had no customers.” -Gordon Cook, 1992“Turning hegemony into democracy by peaceful means has been done only a few time in human history, and the outlook for this time isnt good. -Paul Vixie, 2005, reminder: there is good newswe made something so great, everyone wants it. in
20、 fact many of us want it more than once! (um)the current industry is a historical artifact of technical and (science & regulatory) policy innovations in the 60s, 70s, 80s, 90s, and 00speople are starting to study interplay, but theyre undercapitalizedin the meantime, it became global critical infras
21、tructure. oops.who pays for critical infrastructure?how is the Internet different from other critical infrastructure? (hint: who protects public interest?) what will give? how long will it take?according to history: capital will be allocated to architectural innovation(rural electrification: 48 years, finally using cooperatively owned companies and federal funding act)
