International research through networking: an old bEditor, Anaplasmosis/Babesiosis Network, Washington State University, Pullman, WA, USA Abstract
tem of ancient Rome But no matter how remote and distinct by their nature different networks can seem, they all have something in common
That means to stay up to date and maintain your efficiency, your network devices should be replaced approximately every three years
Computer Networks, Networking History, Networking Lit- erature networking is almost as old as work on the ARPAnet 1The paper was submitted in September
Hardware-Defined Networking (HDN) explores the patterns that are common to modern net- prevent old information from circulating around the network
Standards for Networking Ancient Person-data: Digital approaches to problems in prosopographical space Gabriel Bodard, Hugh Cayless, Mark Depauw,
Keywords: Networking, entrepreneurship, business development, tourism, social capital With the background of the Hotel Groups seven years old networking
What the Chinese can teach us about networking The Chinese have been perfecting guanxi which is the equivalent of our networking and relationship building
14605_3p57_v37n1h_ammar.pdf
10NetworkingPapers:ABlastfromthePast
MostafaH.Ammar
CollegeofComputing
GeorgiaInstituteofTechnology
Atlanta,GA,USA
ammar@cc.gatech.edu
CategoriesandSubjectDescriptors
A.2[Reference]:GeneralLiterature;C.2[Networks]:Com- puterCommunications
GeneralTerms
Algorithms,Design,Performance
Keywords
ComputerNetworks,NetworkingHistory,NetworkingLit-
erature
Soitismyturntorecommendareadinglistof10network-
ingpapers.Idecidedtotakethisopportunitytorecommend papersfromthenetworkingarea'spast.Ihaveseveralrea- sonsforfocusing onsuchalist.First,asarelativelyyoung disciplinethathasseentremendousgrowthinrecentactiv- itiesandcontributions,ourpresenttendstooverwhelmour pastinsheervolume.Soourpastcanfadeawayunceremo- niouslysometimes.Tobesureweareallveryfamiliarof certainmilestonepapers.Bydenition,however,signicant milestonesdonotoccurfrequentlyenoughforthemtopro- videanaccuratesamplingofourpast.Second,readingolder paperscanbequitefascinating.Oneisoftenleftinaweof theamountofinsightandideasgeneratedbyresearchersar- rivingearlyintoanopeneldofresearc h.Alsotheresearch styleandcontexthasastrangelookandfeeltoit,re ecting amarkedevolutionincommunityresearchexpectations.Fi- nally,Indthatreadingolderpaperscangiveanimportant freshperspectiveonhowwereadandevaluatemoremod- ernliterature;especiallywhentryingtojudgelongevityof contributions.
Mylistconsistsofpapersprimarilyfromtheperiodbe-
tween1975and1984;mostlybecausethisistheperioddur- ingwhichIpersonallyarrivedintothenetworkingscene.I couldnotresistincludingonepaperfrom1962,simplyto makethep ointthatmy\past"hasitsownpast.Papers wereincludedinthelistformanyreasons.Somepapers areonthelistbecausetheydescribewhatIconsidertobe amilestonecontribution,othersareonthelisttomakea pointabouthowsophisticatednetworkingresearchwasat thatpointintime.Stillotherswereincludedbecausethey aregoodrepresentativesofresearchstylesorperiods.Some paperspossessmanyoftheseelementssimultaneously.Inthetraditionofthiscolumn,thisisahighlypersonal listandImadenoattempttobecompleteinanyway .In manyinstances,thefulleectcanonlybehadbyreading \around"mylist{lookingupreferencesorrelatedmaterial inpriororsubsequentwritings.
Sohereismylistinchronologicalorder:
\RoutingProceduresinCommunicationsNetworks-Part
I:RandomProceduresandPartII:DirectoryProce-
dures"[12,13].Thisbodyofwork1byProsserwritten over45yearsago(!)isquitefascinating.Thepapers askthenowclassicquestionofhowtobestroutemes- sagesinacommunicationnetwork,exceptthatin1962 itwasnotaclassic questionyet.Prosserposesthe questionformilitarycommunicationnetworks.This wasclearlyhismotivation.Whatisinterestingthough iswhatdistinguishedmilitarynetworksfromcivilian networksinhismindwasthatinaciviliannetwork \eachstationhasaccesstocompleteandcorrectin- formationaboutthesystemandroutesmessageson thebasisofthisinformation".Itisinterestingtosee howourviewofsuchnetworkshaschangeddramati- callyovertheyears.Aswithanyworkthatmanages toaskanimportantanddeepquestionearly,thepa- percontainsawealthofideasabouthowtodorouting whic hwecontinuetoinvestigatetoday.Forexample, aquicksearchon\RandomRouting"showsproposals foritsuseinsettingsincludingMobileAd-HocNet- works,OpticalWDMnetworks,andprocessorinter- connectionnetworks. \PacketSwitchinginaMultiaccessBroadcastChannel:
PerformanceEvaluation"[8].Irememberfallingin
lovewiththispaperwhenIreadit.Ihadalwaysheard andreadhandwavingargumentsabouttheinstability ofALOHAbutIcouldnevergetanexactanswerabout whythatwasthecase.Readingthispaperwaslikea revelation.
Whilethemathisratherstraightforward
(atleastbytoday'sstandards),theanalysisproduces tremendousinsight.Inthebesttraditionoftheday, theauthorspublishedafollowuppaperdescribingcon- trolprocedurestodealwithALOHA'sinstability[10]. \TheOrganizationofComputerResourcesintoaPacket
RadioNetwork"[7].Mystudentsareoftensurprised
toknowthatresearchintomobileandwirelessdata networkingisalmostasoldasworkontheARPAnet1ThepaperwassubmittedinSeptember1962andpublished inDecember1962{thosewerethedays! ACM SIGCOMM Computer Communication Review57Volume 37, Number 1, January 2007 itself.Thispaperoverviewsthemotivationandarchi- tectureofPacketRadioNetworks(reallyMANETsby anothername).Iincludeithereasarepresentativeof thisbodyofworkthatpre-datesthecurrentinterest insuchnetworksandasapointertoawealthofearly workinthisarea.Thesophisticationandambitionof thiseortarereadilyillustratedbythestatedgoals enumeratedintheintroduction. \AnOptimalAdaptiveSchemeforMultipleAccessBroad- castCommunication"[9].Thispaper,whichwonthe bestpaperawardattheIEEEICCin1978,describes theUrnmultipleaccessprotocol.Theprotocolman- agestooptimallyadapttothenetworkloadovera rangefromTDMAtoslottedAloha.Ihavealways foundtheideabehindthisprotocolanditsanalogyto pullingballsoutofanurntobequiteinteresting.The paperisanexcellentrepresentativeofthesignicant energythatwasdevotedtodesigningandunderstand- ingMultipleAccessProtocolsatthetime. \ReversePathForwardingofBroadcastPackets"[5].
Thisisperhapsthemostwell-knownpaperoutofmy
list,certainlytoanyonethathasworkedonmulticast- ing.Thepaper'sintendedpurposeistointroducethe reversepathforwarding(RPF)technique,aningenious compromisebetweenhighlyinecientbuteasytoim- plement oodingandoptimally-ecientbutcumber- somespanningtreerouting.RPFwouldgoontoform thebasisofsubsequentInternetmulticastroutingpro- posals(seee.g.[6]).Iamrecommendingthispaper, however,becauseintheprocessofdescribingbench- marksagainstwhichoneshouldcomparetheRPF technique,itmanagestoprettymuchcovertheentire spectrumofmulticastroutingideasthatwouldthen occupyresearchersforthenext20years. \OntheStatisticalAnalysisofQueueLengthsandWait- ingTimesforStatisticalMultiplexerswithARQRe- transmissionSchemes"[15].Ichosethispaperbecause tomeitisthequintessentialpaperofagenrethatwas verypopularinthelate70sbutisincreasinglyrare nowadays.Thepaperpresentsacarefulmodelanda highlyaccessibleanalysisofARQschemes.There- sultsprovideinterestinginsightsandthetechniqueis artful.Youwillhavetoreadityourselftondout whata\slacket"is.Aclosecontenderforthispaper slotwasanotherextremelywell-donepieceofworkan- alyzingtheHDLCprotocol[4]. \InternetworkProtocolApproaches"[11].Iincludethis paperinmylistfortworeasons.First,Iwantedto includeonepaperfromtheclassicApril1980IEEE
TransactionsonCommunicationsspecialissueonCom-
puterNetworkArchitecturesandProtocolseditedby
PaulGreen.The23articlesinthisspecialissuepro-
videanalmostcompletesnapshotofthestateofnet- workingatthetime.Second,itisinterestingtoread thispaperasitpresentsabalancedexpositionoftwo predominantinterconnectionstrategies{completewith hand-drawnpictures,presumablybyJonPostelhim- self.Withincreasedinterestinclean-slateapproaches tonetworkingitisimportanttoberemindedofthe
X.25/X.75interconnectionapproach.Itisalsoquite
interestingfromahistoricalperspectivetoseehowthetwoapproacheswereneverdiscussedinthepaperas competingandthereisneveranyattempttoshowthe
X.25/X.75approachinaninferiorlight.
\VideotexNetworks"[2].Weoftenassociatethesuc- cessoftheservicesprovidedbytheInternetwiththe successoftheInternetitself.Butwhatkindofdigi- talserviceswouldwehavehadwithouttheInternet.
Thisisanimpossiblequestiontoanswerbutthispa-
pergivesusaclue.ItdocumentstheworldofVideo- texNetworkswhichareenvisionedas\utilities[that] willoeravarietyofinformationservicesandtransac- tions,suchasretrievalfrommultipleindependentdata bases,messaging,electronicmail,conferencing,bank- ing,teleshopping,andinterestmatching."Allofthisin
1980!Thepaperdescribessystemarchitectures,trials
ofsystemsandservices,allinthecontextofdatanet- workingtechnologyexistingatthetime.Thevideotex bubblewouldburstalmostadecadelater(butbefore theinventionoftheweb)forbusinessandeconomic reasons.Itwouldleaveitsmarkinsomesystemssuch asFrance'sminitel. \WhyaRing?"[14].Thispaperwaswrittenatatime whenLocalAreaNetworktechnologywasahottopic.
Oneoftheimportantdebateswaswhetherbusnet-
works(usingprotocolslikeEthernet)orringnetworks (usingprotocolslikethetokenringprotocol)would prevail.Thispaperprovidesanexquisitesummary ofthedebateandconcludesthat\bothapproaches haveargumentsintheirfavor."Weclearlyknowwho thewinnerwas{actuallyneither.Thehubbedand switchedEthernetoftodayshareslittlewiththebroad- castbusnetworksof1981.Infactonecanargue thattheideaofusinghubsrstcameupincentral wiringtechniquesproposedtoaddressringreliabil- ityandmaintenanceproblemsasdiscussedintherst partofthispaper. \DatagramRoutingforInternetMulticasting"[1].As farasIcantellthispaperpresentstherstevercon- creteproposalfordeployingmulticastcommunication inIPnetworks.Theideaisbasedontheuseofmulti- destinationaddressing2:eachpacketcarriedalistof destinationsinitsoptionseldandunicastrouting tableswereusedtoroutethepackettoitsmultiple destinations.Ihavetoyethearasatisfactoryanswer towhythisideawasnotpickedupanddeployed.The standardansweristhattheschemedidnotscaleto beyond8or9receivers.ButtheInternetwasnotthat largein1984andaschemethatwouldallowanor- derofmagnitudesavingsofbandwidthshouldhave beenwelcomeanyway.Themysterycontinues.The multidestinationaddressingideawouldberesurrected some15yearslaterineortsonSmallGroup(a.k.a. explicit)Multicast[3].
Ihopeyoundthislistusefulasaguidetoexploringthose
yearsinthenetworkingresearchcommunity'spast.Iwould liketoencourageotherstocontributetheirownliststoCCR exploringthesameorotherperiodsinthenetworkingcom- munity'sshortbuteventfulhistory.
2Youguessedit,alsodescribedinthe1978DalalandMet-
calfepaperlistedabove. ACM SIGCOMM Computer Communication Review58Volume 37, Number 1, January 2007
1.REFERENCES
[1]L.Aguilar.DatagramroutingforInternet multicasting.ACMSIGCOMMComputer
CommunicationsReview(ProceedingsofSIGCOMM
'84,14(2):58{63,June1984. [2]A.J.S.Ball,G.V.Bochmann,andJ.Gecsei.
Videotexnetworks.IEEEComputerMagazine,
13(12):8{14,Decemeber1980.
[3]R.Boivie,N.Feldman,andC.Metz.Smallgroup multicast:Anewsolutionformulticastingonthe internet.IEEEInternetComputing,4(3):75{79,2000. [4]W.Bux,K.Kummerle,andH.Truong.Balancedhdlc procedures:Aperformanceanalysis.IEEE
TransactionsonCommunications,28(11):1889{1898,
November1980.
[5]Y.DalalandR.Metcalfe.Reversepathforwardingof broadcastpackets.CommunicationsoftheACM,
21(12):1040{1048,December1978.
[6]S.E.DeeringandD.R.Cheriton.Multicastrouting indatagraminternetworksandextendedlans.ACM
Trans.Comput.Syst.,8(2):85{110,1990.
[7]R.Kahn.Theorganizationofcomputerresourcesinto apacketradionetwork.IEEETransactionson
Communications,25(1):169{178,January1977.
[8]L.KleinrockandS.Lam.Packetswitchingina multiaccessbroadcastchannel:Performance evaluation.IEEETransactionsonCommunications,
23(4):410{423,April1975.[9]L.KleinrockandY.Yemini.Anoptimaladaptive
schemeformultipleaccessbroadcastcommunication.
InInternationalConferenceonCommunications,
pages7.2.1{7.2.5.IEEE,June1978. [10]S.LamandL.Kleinrock.Packetswitchingina multiaccessbroadcastchannel:Dynamiccontrol procedures.IEEETransactionsonCommunications,
23(9):891{904,September1975.
[11]J.Postel.Internetworkprotocolapproaches.IEEE
TransactionsonCommunications,28(4):604{611,
April1980.
[12]R.Prosser.Routingproceduresincommunications networks-parti:Randomprocedures.IEEE
TransactionsonCommunications,10(4):322{329,
December1962.
[13]R.Prosser.Routingproceduresincommunications networks-partii:Directoryprocedures.IEEE
TransactionsonCommunications,10(4):329{335,
December1962.
[14]J.H.Saltzer,D.D.Clark,andK.T.Pogran.Whya ring?ACMSIGCOMMComputerCommunications
Review(Proceedingsofthe7thSymposiumonData
Communications,11(4):211{217,October1981.
[15]D.TowsleyandJ.Wolf.Onthestatisticalanalysisof queuelengthsandwaitingtimesforstatistical multiplexerswitharqretransmissionschemes.IEEE
TransactionsonCommunications,27(4):693{702,
April1979.
ACM SIGCOMM Computer Communication Review59Volume 37, Number 1, January 2007