Resource borrowing is a common underlying approach in grid computing and thin-client computing. In both cases, external processes borrow resources that would otherwise be delivered to the interactive processes of end-users, creating contention that slows t
geststhatifthe“right”cycleswereused,those“inbetween”thecyclestheuserisusing,therewouldbelittlefortheusertoperceive.Thepriority-basedschedulersofmodernoper-atingsystemsapproximatethis.
Despiteindirectevidence,thereexistnoquantitative,empiricalmeasurementsthatcouldbeusedtoanswertheabovequestions.Thismayseemsurprisingtosomereaders,asthiswouldappeartobeanexcellentprobleminhuman-computerinteractionorpsychology.However,theworkinthoseareashasconcentratedontheimpactoflatencyonuser-perceivedutilityofthesystem[17,8],andonuserfrustrationtodifferentuserinterfaces[16,21].Withinthesystemscommunity,relatedworkhasexaminedtheperfor-manceofend-useroperatingsystemsusinglatencyasop-posedtothroughput[9],andsuggestedmodelsforinterac-tiveuserworkload[2].
Inresponsetothislackofinformation,wehavedevel-opedasystem,theUnderstandingUserComfortSystem(UUCS).UUCSisadistributedWindowsapplicationsim-ilartoSETI@Homeindesign.AUUCSclientemulatesresourceborrowingofCPU,memoryanddiskontheuser’smachineinacontrolledmannerencodedinatestcasepro-videdbyaserver.Theuseroperatesthemachineasnormal,butifhisinteractivitygetsaffected,hemayexpressdiscom-fortbyclickingonaniconorpressingahot-key.Resourceborrowingstopsimmediatelyifdiscomfortisexpressedorwhenthetestcaseis nished.Thepointofdiscomfort,ifany,isreturnedtotheserveralongwithcontextualinfor-mation.Byanalyzingtheresultsofapplyingaparticularsetoftestcasestoaparticularsetofusers,wecanquali-tativelyandquantitativelycharacterizeusercomfort.Notethatabackgroundapplicationhasfullcontroloverthere-sourcesthatitborrows,butamappingbetweenresourceborrowingandinteractivitymetricslikesystemlatencyorjitterisdif culttoobtain.Thisstudyisintendedtodirectlystudytheend-to-endrelationshipbetweenresourceborrow-ingandusercomfort.
UsingUUCS,whichwedescribeindetailinSection2,wehaveconductedacarefullycontrolleduserstudyatNorthwesternUniversitytoaddressthequestionsraisedearlier.Wedescribethisstudyanditsresultsinde-tailinSection3,whileSection4describesourongo-ingInternet-scalestudy.Finally,inSection5,wepro-videadvicetoimplementorsbasedonourstudyresults.MoreinformationabouttheUUCSsystemcanbefoundinaseparatetechnicalreport[13].Toexamineoursoft-ware, yersandquestionnaires,ortoparticipatevisitdoc.guandang.net.
2Systemdesign
UUCSconsistsofaserverandaclient,asshowninFig-ure1.BothareWindowsapplicationsthatstoretestcases
Local Result
Global Result


Store
Local Testcase
Store
Global Testcase

Client


Server
Store
Figure1.StructureofUUCS.
Figure2.TheAnalysisPhase.
andresultsonpermanentstorageintext les.Atestcasecontainsfunctionsthatdescribehowto“exercise”acollec-tionofresources.
Usingitslocaltestcaseandresultstores,theclientcanoperatedisconnectedfromtheserver.Therearetwointer-actionsbetweenthetwo,bothinitiatedbytheclient.Whentheclientisinitiallyrun,itregisterswiththeserver,provid-ingitwithadetailedsnapshotofthehardwareandsoftwareoftheclientmachine,andallowingtheservertoassociateagloballyuniqueidenti erwiththeclient.Subsequently,atparticularlyselectedtimeswhentheclientisconnectedtothenetwork,theclientinitiatesa“hotsync”withtheserver.Newtestcases,whichcanbeaddedtotheserveratanytime,aredownloadedbytheclient,whilenewresultsareuploadedbacktotheserver.
Hotsyncingoperatesatuser-de nedintervalsandac-quiresagrowingrandomsampleoftestcasesfromtheserver.This,combinedwithlocalrandomchoiceoftest-casesandPoissonarrivalsoftestcaseexecution,isdesignedtomakeacollectionofclientsexecutearandomsamplewithrespecttotestcases,users,andtimes.ThisisthemodeofoperationthatweuseinourInternet-widestudy.AUUCSclientcanalsobecon guredtobehavedeterministi-cally,executingaprede nedsetofcommandsfromalocal le.Weusethisfeatureinourcontrolledstudy.
Inadditiontotheclientandserver,wehavedevelopedasetoftoolsforcreating,viewing,andmanipulatingtest-cases,andforimportingtestcaseresultsintoadatabase(Figure2).Anadditionalsetoftoolsisthenusedtoana-lyzetheresultsandguideustootherinterestingtestcases.
Dinda, Measuring And Understanding User Comfort With Resource Borrowing, In Proceedings of the 13th IEEE International Symposium on High Performance Distributed...
UNDERSTANDING THEVALUE OF ACCOUNTINGUNDERSTANDING ... new partners) Firms with external borrowing ...measuring performance, resources are misallocated and...
热门文档