Internet Quality of Service

Quality of Service (QoS) The best-effort model, in which the network tries to deliver data from source to destination but makes no promises about end-to-end delay, is not sufficient for real-time applications. The primary goal of Quality of Service (QoS) is to support different levels of services by provide priorities including dedicated bandwidth, controlled jitter and latency (required by some real-time and interactive traffic), and improved loss characteristics.

Real-time Applications Require “deliver on time” assurances Example application (audio) –sample voice once every 125us –each sample has a playback time –packets experience variable delay in network –add constant factor to playback time: playback point Microphone Speaker Sampler, A D converter Buffer, D A

Playback Buffer Sequence number Packet generation Network delay Buffer Playback Time Packet arrival

Application Requirements Application file transfer Web documents real-time audio/video stored audio/video interactive games financial apps Data loss no loss loss-tolerant no loss Bandwidth elastic audio: 5Kb-1Mb video:10Kb-5Mb same as above few Kbps up elastic Time Sensitive no yes, 100’s msec yes, few secs yes, 100’s msec yes and no

Taxonomy Applications Real time Tolerant AdaptiveNonadaptive Delay- adaptive Rate- adaptive Intolerant Rate-adaptiveNonadaptive Interactive bulk Asynchronous Elastic

QoS Approaches IPv4 Type of Service (TOS) –IPv6 Traffic class + Flow label Integrated Services (Intserv) Differentiated Services (Deffserv) ATM service classes –CBR, VBR-rt, VBR-nrt, ABR, UBR

IPv4 RFC1349 Precedence: the importance or priority of the datagram TOS: – minimize delay – maximize throughput – maximize reliability – minimize monetary cost – normal service MBZ: Must Be Zero (Unused) –Suggested to be used as ECN field in RFC3168 PrecedenceTOSMBZ

Integrated Services Intserv provides individualized quality-of-service guarantees to individual application sessions by per-flow resource reservation. IETF intserv working group (concluded) Resource ReSerVation Protocol (RSVP, RFC 2205) Subnet Bandwidth Manager (SBM, RFC 2814)

Service Classes Guaranteed service (RFC2212) –Provides firm bounds on the queueing delays that a packet will experience in a network element Controlled-load service (RFC 2211) –provides the flow with a quality of service closely approximating the QoS that same flow would receive from an unloaded network element

Mechanisms Traffic characterization –Tspec (RFC2210) –Rspec (RFC2215) Admission control Reservation protocol –RSVP Packet processing –Weighted Fair Queuing (WFQ)

Flowspecs Rspec: describes service requested from network –controlled-load: none –guaranteed: delay target Tspec: describes flow’s traffic characteristics –average bandwidth + burstiness: token bucket filter –token rate r –bucket depth B –must have a token to send a byte –must have n tokens to send n bytes –start with no tokens –accumulate tokens at rate of r per second –can accumulate no more than B tokens

Token Bucket If the service rate at a network element is R, then the queuing delay is bounded by b/R.

Admission Control A router decides whether to admit a flow based on the R- spec and T-spec of the flow and currently available resources on the router.

Reservation Protocol RSVP is employed in Intserv to setup path and reserve resources.

RSVP Receiver-oriented reservation Receiver heterogeneity Designed to support multicast Merge requirements in case of multicast

RSVP Support multiple senders Use soft state (refresh periodically) –Source transmits PATH messages every 30 seconds –Destination responds with RESV message Separate from route establishment QoS can change dynamically

RSVP Sender 1 Sender 2 PATH RESV (merged) RESV Receiver B Receiver A R R R R R

RSVP RSVP does not specify how the network provides the reserved bandwidth to the data flows. RSVP is not a routing protocol. RSVP is a signaling protocol that allows host to establish and tear down reservations for data flows.

Packet Processing Packet classification associates each packet with the appropriate reservation class. –IPv4: SA, DA, SP, DP, Protocol –IPv6: Flow label Packet scheduling manages queues so that each packet receives the requested service. –Guaranteed: calculate end-to-end delay –Controlled load: assign the aggregate flow with a weight based on the amount of traffic admitted

WFQ WFQ provides different amount of service among queues according to their weights.

Challenges Scalability –Per-flow states Overhead –Signaling messages –Refresh messages –Packet processing Security –RSVP deny of service

Differentiated Services Diffserv provides scalable and flexible service differentiation to handle different classes of traffic in different ways within the Internet. IETF diffserv working group DS field (RFC2474, 2475) Per-Hop Behaviors (PHB, RFC2474, 2597, 2598)

Architecture “Complex at edge, Simple at core.” –At edge: Classifying Metering Marking Conditioning –At core BA classifying Queuing and scheduling

Mechanisms Diffserv Code Point (DSCP) Behavior Aggregate (BA) Per-Hop Behaviors (PHB) –Expedited Forwarding (EF, RFC3246) –Assured Forwarding (AF, RFC2597)

DSCP Edge routers mark packets of different classes with different DSCP. Core routers treat packets with different level of services according to its DSCP. DS field: TOS(IPv4), Traffic class(IPv6) CU: currently unused

Edge Functions Meter monitors whether the incoming packet flow conforms to the negotiated traffic profile. Shaper spaces the incoming packets to the negotiated traffic rate.

PHB Per-Hop Behavior is a description of the externally observable forwarding treatment applied at a differentiated services-compliant node to a behavior aggregate. EF is intended to provide a building block for low delay, low jitter and low loss services by ensuring that the EF aggregate is served at a certain configured rate. The AF PHB group provides delivery of IP packets in four independently forwarded AF classes. Within each AF class, an IP packet can be assigned one of three different levels of drop precedence.

PHB Implementations Priority queue WFQ RED with In and Out (RIO) Weighted RED (WRED)

E2E DiffServ DS Domain Bandwidth Broker (BB) Service Level Agreement (SLA)

BB Adjacent BB User/App Interface Application Server User/Host Network Operator Inter-Domain Interface Intra-Domain Interface Edge Routers Edge Routers Data Repository Routing Information Policy Manager Interface Network Management Interface

ATM QoS Network Architecture Internet ATM Service Model best effort CBR VBR ABR UBR Bandwidth none constant rate guaranteed rate guaranteed minimum none Loss no yes no Order no yes Timing no yes no Congestion feedback no (inferred via loss) no congestion no congestion yes no Guarantees ?

Service Classes

ABR ABR Resource Management (RM) cells Virtual Source/Virtual Destination

RSVP v.s. Q.2931 RSVP –receiver generates reservation –soft state (refresh/timeout) –separate from route establishment –QoS can change dynamically –receiver heterogeneity ATM Q.2931 –sender generates connection request –hard state (explicit delete) –concurrent with route establishment –QoS is static for life of connection –uniform QoS to all receivers