EECS 122: Introduction to Computer Networks Resource Management and QoSMid-TermMid-Semester ResolutionsToday’s Lecture: 14Quality of Service (QoS)Today’s LectureWhat’s the Problem?Three Basic ProblemsA Different TaxonomyThree Basic QuestionsBack to Thee Basic ProblemsLink SharingDifferentiated ServicesTraffic LimitationsDiffServ “Code Points”“Expedited Forwarding”Is Delay the Problem?“Assured Forwarding”ExampleAdvantages of DiffServDiffServ PeeringDisadvantages of DiffServIntegrated ServicesKey DifferencesExample: VideoCircuit-Switched NetworksInternetRouter Effect on DelayRouter Effects on TrafficNetwork Effects on TrafficSlide 31Slide 32Network Effect on DelayChoicesPlayback BufferPlayback PointAdaptationApplication Taxonomy (Oversimplified and Fanciful)Key PointsTwo Service ClassesProcessProblemTSpec: The Token BucketRequired ElementsControl Plane versus Data PlaneControl Plane: Resource ReservationSlide 47Slide 48Slide 49Control Plane: Admission ControlSlide 51Data PlaneSlide 53Slide 54Resource Reservation Protocol: RSVPPATH and RESV MessagesThe Big PictureSlide 58Soft StateRoute PinningAdmission ControlIntServ Node ArchitectureAdvantages of IntServDisadvantages of IntServWhat You Need to KnowFactors Limiting QoS DeploymentQoS DebatesKatz, Stoica F04EECS 122: Introduction to Computer Networks Resource Management and QoSComputer Science DivisionDepartment of Electrical Engineering and Computer SciencesUniversity of California, BerkeleyBerkeley, CA 94720-177601/15/19 2Katz, Stoica F04Mid-TermWill be returned in discussion sectionGrading appeals:-must be submitted within one week of getting your test back (no exceptions)-regrades will be over entire test, not over single question (score could go up or down)Overall: you did very well!-mode: 85 median: 85 mean: 8001/15/19 3Katz, Stoica F04Mid-Semester ResolutionsTimestamps on slides (starting next week)Scott will try to slow down-more interactive-two hands up means stop!Clearer delineation between required material and optional materialMore focused reading assignmentsRevising future topics (more security?)01/15/19 4Katz, Stoica F04Today’s Lecture: 14Network (IP)ApplicationTransportLinkPhysical27, 8, 910,1117, 18, 1914, 15, 1621, 22, 2325601/15/19 5Katz, Stoica F04Quality of Service (QoS)The Internet’s most contentious subjectThe Internet’s most embarrassing failure-almost nothing was accomplished-the research community was dishonest and ineffective-my worst experience as a researcherThe textbook’s worst chapter-a rosy description of bad work01/15/19 6Katz, Stoica F04Today’s LectureWill be about what “could be”, not what is-today’s Internet does not have, nor will soon have, a reasonable QoS solutionFocus will be on what one could accomplish with simple (and not-so-simple) mechanisms-you will only be expected to know basic conceptsI will not discuss current deployed mechanisms-an ugly hodge-podge of hacks01/15/19 7Katz, Stoica F04What’s the Problem?Internet gives all flows the same “best effort” service-no promises about when or whether packets will be deliveredNot all traffic is created equal-different “owners”, different application requirements-some applications require service “assurances”How can we give traffic different “quality of service”?-Thus begins the problem of QoS01/15/19 8Katz, Stoica F04Three Basic ProblemsWant to control how a link is shared:-Link sharingWant to give some traffic better service-Differentiated serviceWant to gives some flows “assured” service-Integrated service (and perhaps differentiated service)01/15/19 9Katz, Stoica F04A Different TaxonomyGiving better service can differ along three dimensions:-relative versus absolute-dropping versus delay-flows versus aggregatesEach of these choices requires different set of mechanisms-router scheduling and dropping decisions-signaling protocols01/15/19 10Katz, Stoica F04Three Basic QuestionsHow does a router service this packet?-scheduling (various forms of priority and RR)-dropping (fancy versions of RED)How did the router know what to do with this packet?-bits in packet header or explicit signalingHow can one control the level of traffic?-service level agreements (SLAs) or admission control01/15/19 11Katz, Stoica F04Back to Thee Basic ProblemsLink sharing (one slide)Differentiated Services (long)Integrated Services (even longer)01/15/19 12Katz, Stoica F04Link SharingTwo organizations share an access link and want to share it equallyOne approach: partition the linkSecond approach: use FQ, with one queue for each organization’s packetsWhich is better?01/15/19 13Katz, Stoica F04Differentiated ServicesSome traffic should get better treatment-application requirements: interactive vs bulk transfer-economic arrangements: first-class versus coachWhat kind of better service could you give?-measured by drops, or delay (and drops)How do you know which packets to give better service to?-bits in packet header01/15/19 14Katz, Stoica F04Traffic LimitationsCan’t give all traffic better service!Must limit the amount of traffic that gets better serviceService Level Agreements (SLA)-source agrees to limit amount of traffic in given class-network agrees to give that traffic “better” service•for a price!-economics play an important (fatal?) role in QoS01/15/19 15Katz, Stoica F04DiffServ “Code Points”Use six of the ToS bits in IP packet headerDefine various “code points”Each code point defines a desired per-hop behavior-a description of the service the packet should get-not a description of the router implementation of that service01/15/19 16Katz, Stoica F04“Expedited Forwarding”Give packet minimal delay and loss service-e.g., put EF packets in high priority queueTo make this a true “absolute” service, -all SLAs must sum to less than the link speed-unlikelyMore likely, a way to assure relatively low delay01/15/19 17Katz, Stoica F04Is Delay the Problem?With RED, most queues are smallPackets are dropped when queue starts to growThus, delays are mostly speed-of-light latencyService quality is mostly expressed by drop-rateWant to give traffic different levels of dropping01/15/19 18Katz, Stoica F04“Assured Forwarding”Packets are all serviced in order-makes TCP implementations perform wellBut some packets can be marked as
View Full Document