Int Serv and Diff Serv School of Electronics

  • Slides: 83
Download presentation
Int. Serv and Diff. Serv School of Electronics and Information Kyung Hee University. Choong

Int. Serv and Diff. Serv School of Electronics and Information Kyung Hee University. Choong Seon HONG <cshong@khu. ac. kr> Kyung Hee University

ëQuality of Service (Qo. S) ü A major driving force in Internet evolution ü

ëQuality of Service (Qo. S) ü A major driving force in Internet evolution ü Not simply defined - means many things to many people ü Has sense of predictable network behaviour ü Central idea is provision of network resources that an application requires to perform adequately Kyung Hee University

ëQuality of Service (Qo. S) ü What is Quality-of-Service? • Quality of service (Qo.

ëQuality of Service (Qo. S) ü What is Quality-of-Service? • Quality of service (Qo. S) is a concept by which applications may indicate and even negotiate their specific service requirements to the network ü Why is this an issue? • The default service in many packet networks is to give all applications the same service, and not consider any service requirements to the network. This is called a best-effort service. Kyung Hee University

ëQuality of Service (Qo. S) ü Who needs Quality-of-Service? – Video and audio conferencing

ëQuality of Service (Qo. S) ü Who needs Quality-of-Service? – Video and audio conferencing bounded delay and loss rate – Video and audio streaming bounded packet loss rate – Time-critical applications (real-time control) bounded delays – “valuable applications” better service than less valuable applications Kyung Hee University

ëQuality of Service (Qo. S) ü How are Quality-of-Service requirements specified? • Qo. S

ëQuality of Service (Qo. S) ü How are Quality-of-Service requirements specified? • Qo. S parameters are – Delay Variation (Jitter) – Throughput – Error Rate Kyung Hee University

ëQuality of Service (Qo. S) ü What is the granularity of Qo. S? –

ëQuality of Service (Qo. S) ü What is the granularity of Qo. S? – Per-flow Qo. S Guarantees are specified and enforced for single end-to-end data flow – Aggregate Qo. S Guarantees are specified and enforced for groups of flows Kyung Hee University

ëTypes of Qo. S guarantees ü Deterministic Qo. S – Service guarantees are enforced

ëTypes of Qo. S guarantees ü Deterministic Qo. S – Service guarantees are enforced for all traffic F For example, deterministic delay guarantees have the form: Delay of a packet from flow X ≤ D (D is called a delay bound) ü Statistical Qo. S F Allows a certain fraction of traffic to violate the service guarantees F Prob [Delay of a packet from flow X ≤ D ] ≥ 1 - ε Where e is a small number (e. g. , ε = 10 -6)ε Kyung Hee University

ë Classification and Scheduling ü Routers need to be able to 1. classify arriving

ë Classification and Scheduling ü Routers need to be able to 1. classify arriving packets according to their Qo. S requirements Packet Classification 2. isolate traffic flows and provide requested Qo. S Packet Scheduling Kyung Hee University

ëQo. S is Generating a Confusing Array of Acronyms Qo. S Co. S Intserv

ëQo. S is Generating a Confusing Array of Acronyms Qo. S Co. S Intserv RSVP Diffserv MPLS GMPLS Kyung Hee University

ëWhy Do We Need Such a Revolutionary Change? ü Current ‘best effort’ technology is

ëWhy Do We Need Such a Revolutionary Change? ü Current ‘best effort’ technology is essentially a quarter of a century old ü Two factors driving the development of a new generation of multimedia applications Fcommercialisation of the Internet FIncreasing availability and decreasing cost of bandwidth ü No evidence of ‘free bandwidth’ scenario emerging Frejected in RFC 1633 (1994) - still true Fdemand always rises to meet supply Kyung Hee University

ëQo. S is Not New ü Telephone network has Qo. S Feconomics and technology

ëQo. S is Not New ü Telephone network has Qo. S Feconomics and technology based on a single application Fhighly developed engineering Fbut one size fits all ü BISDN - an attempt by telephony world to generalise network to encompass diverse applications ü ATM technology - first full exploration of Qo. S on demand concepts Kyung Hee University

ëQuality of Service and Resource Management ü Fundamental resource is output link rate ü

ëQuality of Service and Resource Management ü Fundamental resource is output link rate ü Access managed via scheduling discipline ü Bursty input traffic held in buffers Fadds delay and jitter Foverflow causes packet loss ü These factors determine Qo. S at network level ü Optimise via buffer management and scheduler parameter setting Kyung Hee University

ëQo. S in the Internet ü Internet Engineering Task Force (IETF) is evolving Qo.

ëQo. S in the Internet ü Internet Engineering Task Force (IETF) is evolving Qo. S support mechanisms for the Internet - two approaches FThe Integrated Services Internet • Qo. S for individual microflows • perhaps too complex for large networks - won’t scale easily FDifferentiated Services - more scaleable • lose sight of individual microflows - Behaviour Aggregates Kyung Hee University

ë Integrated Services (Intserv) ü Qo. S approached via end to end services Fbest

ë Integrated Services (Intserv) ü Qo. S approached via end to end services Fbest effort - current performance standard Fcontrolled load - lightly loaded network performance ‘soft’ delay bound FGuaranteed - ‘hard’ bandwidth and delay bounds ü Traffic conformance to agreed form expected F‘token bucket’ model - policing if nonconforming ü Resources reserved in routers - RSVP Fmore complex set of functions than ATM Kyung Hee University

ëRSVP is Dead! ü Earlier reports of RSVP’s death were somewhat exaggerated ü Nevertheless

ëRSVP is Dead! ü Earlier reports of RSVP’s death were somewhat exaggerated ü Nevertheless there is a major problem with Intserv - fatal in the eyes of some ü Management of router resources requires each router to maintain per flow ‘state’ ü Creates ‘state explosion’ in the interior routers of core networks - perhaps confine to edges Kyung Hee University

ë Differentiated Services (Diffserv) ü Driving philosophy of the Internet has been to minimise

ë Differentiated Services (Diffserv) ü Driving philosophy of the Internet has been to minimise complexity in the core network - push complexity and intelligence to the edge nodes. ü Differentiated Services concept strives to maintain this philosophy while recognising the need to provide some levels of Quality of Service ü First widely deployed Qo. S mechanism Kyung Hee University

ëDifferentiated Services Kyung Hee University

ëDifferentiated Services Kyung Hee University

ëContent ü Intserv/RSVP ü Differentiated Service Paradigm ü Per-Hop Behavior & Codepoint ü Premium

ëContent ü Intserv/RSVP ü Differentiated Service Paradigm ü Per-Hop Behavior & Codepoint ü Premium Service ü Assured Forwarding PHB Group ü Resource Manager : Bandwidth Broker(BB) ü Boundary Mechanisms ü Diffserv WG Kyung Hee University

ëInternet Integrated Service Model Guaranteed Quality of Service ü Motivation F for applications intolerant

ëInternet Integrated Service Model Guaranteed Quality of Service ü Motivation F for applications intolerant of late data F hard real time requirements ü End-to-End behavior F an assured level of bandwidth F a delay-bounded service with no queueing loss F firm maximum on end-to-end delay F not control the minimal or average delay F no jitter control Kyung Hee University

ë Internet Integrated Service Model ü In order for a router to invoke Guaranteed

ë Internet Integrated Service Model ü In order for a router to invoke Guaranteed Service for a specific data flow it needs to be informed of the traffic characteristics of the flow, Tspec, along with the reservation characteristics, Rspec F Tspec parameters • • • p ; peak rate of flow (bytes/second) b ; bucket depth (bytes) r ; token bucket rate (byes/second) m ; minimum policed unit (bytes) M ; maximum datagram size (bytes) F Rspec parameters • R ; bandwidth, i. e. service rate (bytes/second) • S ; Slack Term (ms) Kyung Hee University

ëInternet Integrated Service Model Controlled - Load Service ü Motivation F for adaptive real-time

ëInternet Integrated Service Model Controlled - Load Service ü Motivation F for adaptive real-time applications (today’s internet) F work well on unloaded nets but degrade quickly under overload conditions --> mimics unloaded nets F If the flow is accepted for Controlled-Load Service then the router makes a commitment to offer the flow a service equivalent to that seen by a besteffort flow on a lightly loaded network Kyung Hee University

ëInternet Integrated Service Model Controlled - Load Service (cont’d) ü End-to-End behavior F Tightly

ëInternet Integrated Service Model Controlled - Load Service (cont’d) ü End-to-End behavior F Tightly approximates the behavior visible to applications receiving besteffort service under unloaded conditions F A very high percentage of packets delivered successfully F Controlled Load has some fairly simplementations, in terms of the queuing systems in routers F It is not suited to applications that require very low latency (e. g. distributed VR systems and so forth). Kyung Hee University

ëRSVP Design Principles ü Receiver-Initiated Reservation F a receiver • choose the level of

ëRSVP Design Principles ü Receiver-Initiated Reservation F a receiver • choose the level of reservation • initiate/keep reservation F more flexible and scaleable than source-initiated reservation • heterogeneous receivers • dynamic membership change ü Separating reservation from packet filtering F reservation • amount of resources reserved for an entity F packet filtering • dynamically select packets that can use the resources Kyung Hee University

ë RSVP Design Principles (cont’d) ü Maintain “Soft-state” F dynamic status change (membership change)

ë RSVP Design Principles (cont’d) ü Maintain “Soft-state” F dynamic status change (membership change) F soft-state in switches and maintained by end users F state in switches • path state -- periodic path message from the source • reservation state -- periodic reserv. msg from the receivers F timeout driven deletion • Reservations timeout if not refreshed periodically F adaptability and robustness ü Protocol overhead F reduce refreshing frequency F merging path/reservation messages Kyung Hee University

ëRSVP Comparison of RSVP and ATM signaling RSVP ATM Receiver generates reservation Sender generates

ëRSVP Comparison of RSVP and ATM signaling RSVP ATM Receiver generates reservation Sender generates connection request Soft state ( refresh / timeout ) Hard state ( explicit delete ) Separate from route establishment Concurrent with route establishment Qo. S can change dynamically Qo. S is static for life of connection Receiver heterogeneity Uniform Qo. S to all receivers Kyung Hee University

ëRSVP Message Types 1. Path_msg S 1 2. forwarding Path_msg R 1 R 2

ëRSVP Message Types 1. Path_msg S 1 2. forwarding Path_msg R 1 R 2 D 1 S 2 R 3 3 -1. Resv_msg 4. forwarding Resv_msg D 2 3 -2. Resv_msg Kyung Hee University

ëInternet Integrated Service Model Integrated Services over Specific Link Layers(ISSLL WG) • RSVP designed

ëInternet Integrated Service Model Integrated Services over Specific Link Layers(ISSLL WG) • RSVP designed to work with any protocol - Protocol must provide Qo. S support - Examples: ATM, IP with Integrated Services • IP integrated services with RSVP over ATM – VC management ( traffic flow-VC) • Data VC, RSVP signaling VC – Qo. S translation • mapping a Qo. S from the IIS model to a proper ATM Qo. S • IIS over POTS • IIS over LAN Kyung Hee University

ëIntserv / RSVP Qo. S Approach F Scalability problem • Have to maintain forwarding

ëIntserv / RSVP Qo. S Approach F Scalability problem • Have to maintain forwarding state between receiver and transmitter Kyung Hee University

ëIntegrated Services Model ü Flow specification ü Routing ü Admission control ü Policy control

ëIntegrated Services Model ü Flow specification ü Routing ü Admission control ü Policy control ü Resource reservation ü Packet scheduling Kyung Hee University

ëRSVP Functional Diagram Host Router RSVPD Routing Process Application D A T A Packet

ëRSVP Functional Diagram Host Router RSVPD Routing Process Application D A T A Packet Classifier Policy Control Admissions Control Packet Scheduler DATA Packet Classifier Kyung Hee University Packet Scheduler DATA

ëWhat is a flow? ü Equivalent packets by some classification FRSVP: Set of packets

ëWhat is a flow? ü Equivalent packets by some classification FRSVP: Set of packets traversing a network element that are all covered by the same Qo. S request ü Packet classifier determines which packets belong to which flows FIPv 6 includes a flow label to ease classification ü ISP usage (UUNET) FMicroflow: TCP or similar bandwidth connection FMacroflow: Large aggregates of packets flowing between two superhubs Kyung Hee University

ëDescribing and Identifying a Flow ü Flowspec defines traffic parameters FTraffic parameters: bandwidth, buffering

ëDescribing and Identifying a Flow ü Flowspec defines traffic parameters FTraffic parameters: bandwidth, buffering requirements FUses token bucket specification ü Filterspec identifies packets in flow FSimplest filter: Source, Dest address/port pair FData filter: classifies packets according to contents Kyung Hee University

ëResource Reservation ü Senders advertise using PATH message ü Receivers reserve using RESV message

ëResource Reservation ü Senders advertise using PATH message ü Receivers reserve using RESV message FFlowspec + filterspec + policy data FTravels upstream in reverse direction of Path message ü Merging of reservations ü Sender/receiver notified of changes Kyung Hee University

ëRSVP UDP Reservation (1) Kyung Hee University

ëRSVP UDP Reservation (1) Kyung Hee University

ëRSVP UDP Reservation (2) Kyung Hee University

ëRSVP UDP Reservation (2) Kyung Hee University

ëClient Traffic Shaping ü Issue: Need traffic shaping to meet allocated resources ü Source

ëClient Traffic Shaping ü Issue: Need traffic shaping to meet allocated resources ü Source promises that data traffic will conform to a particular shape ü Why describe and shape traffic? FNetwork knows what to expect, can manage traffic better FBetter admission control decisions FNetwork can police flows ü Bursty traffic is costly to router, network Kyung Hee University

ëTraffic Shaping Example Data Queue Flow 1 Flow 2 Data Queue Kyung Hee University

ëTraffic Shaping Example Data Queue Flow 1 Flow 2 Data Queue Kyung Hee University

ëTraffic Shapers ü Simple leaky bucket FIsosynchronous flow: regular intervals between packets ü Token

ëTraffic Shapers ü Simple leaky bucket FIsosynchronous flow: regular intervals between packets ü Token bucket FBursty flow Kyung Hee University

ëSimple Leaky Bucket Data b b = bucket size r = rate data is

ëSimple Leaky Bucket Data b b = bucket size r = rate data is sent onto network r ü Sends data at fixed intervals onto network ü Bursts bigger than b are discarded ü Traffic never injected faster than r ü Can be used with cells or datagrams Kyung Hee University

ëToken Bucket r b b = bucket size in tokens r = rate tokens

ëToken Bucket r b b = bucket size in tokens r = rate tokens are added to bucket Data Queue Data ü Sends bursty traffic onto network ü Bucket filled with tokens at rate r ü Data transmitted when enough tokens exist ü Allows bursts, but enforces upper bound Kyung Hee University

ëRestrictions on Reservations ü Admissions FIs bandwidth available? ü Policy FService guarantees give preferential

ëRestrictions on Reservations ü Admissions FIs bandwidth available? ü Policy FService guarantees give preferential access to network bandwidth FPermissions FPricing issues ü What are the policies of nodes on the path? FPolicy data represents a scaling and security issue Kyung Hee University

ëResource Reservation Model ü Senders advertise using flowspecs ü RSVP daemons forward advertisements to

ëResource Reservation Model ü Senders advertise using flowspecs ü RSVP daemons forward advertisements to receivers, update available bandwidth, minimum delay ü Receivers reservations use flowspec, filterspec combination (flow descriptor) ü Sender/receiver notified of changes ü Reservations are merged in multicast case Kyung Hee University

ëReservation Styles ü Wildcard Filter (WF) FShared reservation, select all upstream senders FTraffic from

ëReservation Styles ü Wildcard Filter (WF) FShared reservation, select all upstream senders FTraffic from upstream senders shares a single pipe FAppropriate for audio ü Shared Explicit (SE) FShared reservation, explicit sender selection FAppropriate for audio ü Fixed Filter (FF) FDistrict reservations, explicit sender selection FAppropriate for video Kyung Hee University

ëRSVP Flowspecs Sender TSpec, Controlled Load Flowspec. . . Token Bucket Rate [r] Token

ëRSVP Flowspecs Sender TSpec, Controlled Load Flowspec. . . Token Bucket Rate [r] Token Bucket Size [b] Peak Data Rate [p] Minimum Policed Unit [m] Maximum Policed Unit [M] Guaranteed Flowspec . . . Token Bucket Rate [r] Token Bucket Size [b] Peak Data Rate [p] Minimum Policed Unit [m] Maximum Policed Unit [M] Rate [R] Slack Term [S] Kyung Hee University

ëPacket Scheduling ü Implemented in hosts/routers to control link allocation ü Queuing algorithms FWeighted

ëPacket Scheduling ü Implemented in hosts/routers to control link allocation ü Queuing algorithms FWeighted Fair Queuing (WFQ) FClass Based Queuing (CBQ) ü Queue management FRandom Early Detection (RED) Kyung Hee University

ëPacket Scheduling ü Fair Queueing F Attempts to implement a scheduler that serves all

ëPacket Scheduling ü Fair Queueing F Attempts to implement a scheduler that serves all flows with a backlog at the same rate F Emulates a bitwise Round Robin scheduling algorithm F Not completely trivial to implement Fair Queuing in a packet network Kyung Hee University

ëWeighted Fair Queuing (WFQ) ü Traffic placed into queues according to flow specification, flow

ëWeighted Fair Queuing (WFQ) ü Traffic placed into queues according to flow specification, flow filter ü Fair queuing FImplements fairness of bit by bit scheduling on a per packet basis FGives queues a fair share of total bandwidth ü Weighted FQueue are not weighted evenly for scheduling ü Proven: adequate for Guaranteed Service Kyung Hee University

ëClass Based Queuing (CBQ) ü Combines scheduling and link sharing ü Hierarchical link sharing

ëClass Based Queuing (CBQ) ü Combines scheduling and link sharing ü Hierarchical link sharing FHierarchical queues FEnables protocol, organization isolation ü Scheduling FDoes not define a particular scheduling algorithm FGeneral scheduler for low latency when no congestion FLink-sharing policing scheduler when congested FScheduling per hierarchy Kyung Hee University

ëCBQ Example LINK 60% 40% Company A Company B 30% Real. Time HTTP FTP

ëCBQ Example LINK 60% 40% Company A Company B 30% Real. Time HTTP FTP telnet IP DECnet 20% 10% 20% Video Audio 20% 10% Kyung Hee University

ëRandom Early Detection (RED) ü Random Early Detection (RED) FQueue management algorithm for congestion

ëRandom Early Detection (RED) ü Random Early Detection (RED) FQueue management algorithm for congestion control FRandom packet drops as average queue length increases FCan use Explicit Congestion Notification bit instead of dropping packet FWorks well for TCP FUseful for congested Controlled Load service Kyung Hee University

ëReservation Merging (3) 50 Kbs (7) 100 Kbs R 1 Reservations merge as they

ëReservation Merging (3) 50 Kbs (7) 100 Kbs R 1 Reservations merge as they travel up tree. (6) 100 Kbs R 3 (2) 50 Kbs (9) 60 Kbs R 4 (1) 50 Kbs Receiver #1 R 6 (8) 60 Kbs Receiver #2 Kyung Hee University (5) 100 Kbs R 7 (4) 100 Kbs Receiver #3

ëTSpecs, Ad. Specs, and RSpecs ü Traffic source sends TSpec (Traffic Specification) FConsists of

ëTSpecs, Ad. Specs, and RSpecs ü Traffic source sends TSpec (Traffic Specification) FConsists of Flow. Spec and Ad. Spec ü Ad. Spec updated to reflect network capabilities FRouters update minimum delay and maximum bandwidth FTermed One Pass With Advertisement (OPSA) ü RSpec FReceiver uses Controlled Load or Guaranteed Flow. Spec to reserve network resources Kyung Hee University

ëProblems with Merging Reservations ü Issue: who pays for service, how much? ü Merging

ëProblems with Merging Reservations ü Issue: who pays for service, how much? ü Merging different types of flows FFlow 1: Low delay, low bandwidth FFlow 2: High delay, high bandwidth FFlow with low delay, high bandwidth satisfies Flows 1 and 2, but it may cost much more than Flow 1 or 2. ü Only certain flows can be easily merged given price constraints Kyung Hee University

ëReservation Merging and Price Merged Reservation: High Bandwidth, Low Latency Bandwidth Reservation 2: High

ëReservation Merging and Price Merged Reservation: High Bandwidth, Low Latency Bandwidth Reservation 2: High Bandwidth, High Latency Reservation 1: Low Bandwidth, Low Latency Price: Darker = More Costly Latency Kyung Hee University

ëRSVP Routing Problems ü Routing is separated from admission control ü If route changes,

ëRSVP Routing Problems ü Routing is separated from admission control ü If route changes, reservation must be made along new route FNew reservation takes time to setup FNew reservation might fail FOld route could still be working fine ü Route pinning FAlways use the route where reservation is in place Kyung Hee University

ëRouting Problems (cont’d) ü Reservation failure FPrimary route has inadequate bandwidth although secondary has

ëRouting Problems (cont’d) ü Reservation failure FPrimary route has inadequate bandwidth although secondary has enough ü Telephone system has a crankback feature FAllows secondary routes to be considered if reservation on primary route fails ü ATM FRouting combined with admission control Kyung Hee University

ëUsage and Implementation ü RSVP is not widely available FBest effort delivery across links

ëUsage and Implementation ü RSVP is not widely available FBest effort delivery across links with no RSVP services FReservation flag to specify that traffic traveled over a non-RSVP link ü Some links will have guaranteed performance for some traffic, but not all FPolicy issues at boundaries of networks Kyung Hee University

ëDifferentiated Service Paradigm ü Complicated operation moves to edge, and stateless in network interior

ëDifferentiated Service Paradigm ü Complicated operation moves to edge, and stateless in network interior “ push all the state to the edges, and force all perconversation work (e. g. , shaping, policing) to the edges” ü Setting a specific part in an edge node and administrative boundaries FDS(differentiated service) field ü How to forward according to a specific field of input packet FPer-Hop Behavior ü According to service rule that is previously promised FTraffic Conditioning Kyung Hee University

ëTraffic Conditioning ü Traffic conditioning mechanisms at the network boundary need to enforce that

ëTraffic Conditioning ü Traffic conditioning mechanisms at the network boundary need to enforce that traffic from a flow adheres to its specification Policing Drop traffic that violates the specification Shaping Buffer traffic at network entrance that violates specification Marking Mark packets with a lower priority or as best effort, if the traffic specification is violated Kyung Hee University

ëTraffic Conditioning ü The most popular traffic conditioning algorithm is the leaky bucket Kyung

ëTraffic Conditioning ü The most popular traffic conditioning algorithm is the leaky bucket Kyung Hee University

ëPer-Hop Behavior & Codepoint IPv 4 Header (first 32 bits) 4 -bit version 4

ëPer-Hop Behavior & Codepoint IPv 4 Header (first 32 bits) 4 -bit version 4 -bit header length 8 -bit type of service (TOS) 6 -bit DSCP for Per-Hop Behavior 16 -bit total length (in byte) 2 -bit CU Currently Unused -- DS field in IPv 4 -Kyung Hee University

ëPer-Hop Behavior & Codepoint Default PHB ü Current best effort forwarding ü codepoint :

ëPer-Hop Behavior & Codepoint Default PHB ü Current best effort forwarding ü codepoint : 000000 Class Selector PHB ü for backward compatibility (IP precedence field) ü codepoint : xxx 000 ü relative service quality Kyung Hee University

ëPremium Service ü Providing resources according to Peak capacity Profile FStatic allocations on peak

ëPremium Service ü Providing resources according to Peak capacity Profile FStatic allocations on peak rate with no statistical sharing FSmall percentage of the total network capacity allocate for Premium service FMuch higher cost (First class in aircraft) ü Commercial applications for Premium service FVideo broadcasts, voice-over-IP, VPNs, etc. Kyung Hee University

ëPremium Service Company A Internal Router Premium packet flow restricted to r rate per

ëPremium Service Company A Internal Router Premium packet flow restricted to r rate per sec Host First-Hop Router Unmarked packet flow Border Router Packet in premium flows have bit set ISP Border Router -- Premium traffic flow from end-host to organization’s ISP -Kyung Hee University

ëPremium Service ü Forwarding Path Primitives FGeneral Classifier • A transport-level signature matching based

ëPremium Service ü Forwarding Path Primitives FGeneral Classifier • A transport-level signature matching based on a tuple in the packet header FBit-pattern Classifier • A simple two-way decision based on whether a particular bitpattern in the IP header is set or not – Ex) ‘P’ bit FBit setter • Sets the appropriate bits of the IP header to a configured bitpattern would be the most general Kyung Hee University

ëPremium Service ü Priority queues F(At least) Two levels of simple priority queuing •

ëPremium Service ü Priority queues F(At least) Two levels of simple priority queuing • The high priority queue for Premium traffic ü Shaping token bucket FForward an arriving packet if there is a token present in the bucket, otherwise the packet is enqueued until the bucket contains tokens sufficient to send it • Used in Leaf router ü Policing token bucket FNever hold arriving packets, but check token availability • Used in Border router Kyung Hee University

ëAssured Forwarding PHB Group ü N AF classes ü M drop precedence level ü

ëAssured Forwarding PHB Group ü N AF classes ü M drop precedence level ü at this point 4 classes, 3 drop precedence in each class ü Example Service : Olympic service Kyung Hee University

ëOlympic Service Gold Service Class Silver Service Class Bronze Service Class ü Packets assigned

ëOlympic Service Gold Service Class Silver Service Class Bronze Service Class ü Packets assigned to Gold service class experience lighter load than packets assigned to the silver class ü Packets within each class may be further separated by the drop precedence ü Drop precedence level by using a dual leaky bucket traffic policer : committed burst, excess burst Kyung Hee University

ëResource manager : Bandwidth Broker (BB) ü A logical entity residing in each administrative

ëResource manager : Bandwidth Broker (BB) ü A logical entity residing in each administrative domain FManaging internal demands & resources according to the policy database (who can do what when) Fsetting up & maintaining bilateral agreement with neighbor domains • bookkeeping how much traffic entering which border router & going out which border router ü Today’s BB : network administrators & operators Fwould like to automate over time Kyung Hee University

ëBandwidth Broker (BB) ü Dynamic bandwidth allocation and TCA management Kyung Hee University

ëBandwidth Broker (BB) ü Dynamic bandwidth allocation and TCA management Kyung Hee University

ëChoices for implementation ü Adequate provisioning ü Manual configuration Fnot that different from static

ëChoices for implementation ü Adequate provisioning ü Manual configuration Fnot that different from static routing ü Using some setup protocols Finter-domain : BB-to-BB Fintra-domain : RSVP as a ready candidate Kyung Hee University

ëThe TCA ü One per customer - two parts ü Constraint TCA Fprotects the

ëThe TCA ü One per customer - two parts ü Constraint TCA Fprotects the provider’s resources FDS field : metering profile : disposition of n/c traffic Fquantitative service levels will also include destination address ü Fine grain TCA Fspecifies the fine grain traffic conditioning requested by the customer Fmf class. Criteria : mark : shaping profile : disposition of n/c traffic Kyung Hee University

ëConstraints ü Resources implied by fine grain TCA are constrained by those permitted in

ëConstraints ü Resources implied by fine grain TCA are constrained by those permitted in the constraint TCA FSum of shaping profiles for each mark must be less than metering profile for corresponding DS Kyung Hee University

ëInter-domain DS ü A Service Level Agreement (SLA) includes a Traffic Conditioning Agreement (TCA)

ëInter-domain DS ü A Service Level Agreement (SLA) includes a Traffic Conditioning Agreement (TCA) ü Simplest way: an administrative issue Kyung Hee University

ëConfiguring Routers ü TCA (Traffic Conditioning Agreement) Fconstraint Ffine grain ü PHB information ü

ëConfiguring Routers ü TCA (Traffic Conditioning Agreement) Fconstraint Ffine grain ü PHB information ü Miscellaneous Finterface configuration Frouting configuration Fetc. Kyung Hee University

ëConfiguring the Constraint TCA ü Specifies agreement between provider and customer ü Relatively static

ëConfiguring the Constraint TCA ü Specifies agreement between provider and customer ü Relatively static FConfiguration via SNMP, CLI, COPS, etc. ü Dynamic FConfiguration via COPS or alternate ‘BB’ protocol FCould be triggered by high/low water marks Kyung Hee University

ëQo. S Policy Control Kyung Hee University

ëQo. S Policy Control Kyung Hee University

ëConfiguring the fine Grain TCA ü Ability to make frequent changes desirable ü No

ëConfiguring the fine Grain TCA ü Ability to make frequent changes desirable ü No need to negotiate with provider ü Potentially error prone process ü Therefore - use a signaling protocol to do this whenever possible : RSVP ü It will still be necessary to provision certain fine grain entries Ffor these, use COPS, SNMP or CLI Kyung Hee University

ëSignaled vs. Provisioned TCA Entries ü Quantitative Qo. S apps use quantitative services FTCA

ëSignaled vs. Provisioned TCA Entries ü Quantitative Qo. S apps use quantitative services FTCA entries configured by RSVP signaling or provisioned FTCA entries specify egress points ü Qualitative Qo. S apps use qualitative services Fprovisioned only ü Should use separate marks (DS-field) Kyung Hee University

ëAdmission Control ü a/c necessary to prevent over-subscription Fsum of qualitative entry shaping profiles

ëAdmission Control ü a/c necessary to prevent over-subscription Fsum of qualitative entry shaping profiles leq than metering profile for qualitative DS fields • enforced at provisioning time(SNMP, CLI or COPS) Fsum of quantitative entry shaping profiles leq than metering profile for quantitative DS fields • enforced by rejection or acceptance of RSVP signaling messages • if quantitative resources are provisioned, a/c must be enforced both at provisioning and signaling times • difficult to manage Kyung Hee University

ëDiffserv WG status ü Closed ü RFC 2474, Definition of the Differentiated Services Field

ëDiffserv WG status ü Closed ü RFC 2474, Definition of the Differentiated Services Field (DS Field) in the IPv 4 and IPv 6 Headers ü RFC 2475, An Architecture for Differentiated Service ü --- Kyung Hee University

ëDiffserv / MPLS ü Diffserv FBA (Behavior Aggregate) • BAS (BA Selector) ; forwarding

ëDiffserv / MPLS ü Diffserv FBA (Behavior Aggregate) • BAS (BA Selector) ; forwarding queue behavior • BM (Behavior Modifier) ; Dropping behavior FPHP ü MPLS FSingle LSP • Same BAS • BM Carried as part of label encapsulation header Kyung Hee University

ëDiffserv / RSVP Policy Server PDP : Policy Decision Point PDP PEP PEP :

ëDiffserv / RSVP Policy Server PDP : Policy Decision Point PDP PEP PEP : Policy Enforcement Point Diffserv Network PEP PEP RSVP Diffserv / PHP / Codepoint Kyung Hee University PEP