11/29/20071CS640: Introduction to Computer NetworksAditya AkellaLecture 20 -Queuing and Basics of QoS2Queuing Disciplines• Each router must implement some queuing discipline– Scheduling discipline– Drop policy• Queuing allocates both bandwidth and buffer space:– Bandwidth: which packet to serve (transmit) next – Buffer space: which packet to drop next (when required)• Queuing also affects latency3Typical Internet Queuing• FIFO + drop-tail– Simplest choice– Used widely in the Internet– FIFO: scheduling discipline– Drop-tail: drop policy• FIFO (first-in-first-out) – Implies single class of traffic, no priority• Drop-tail– Arriving packets get dropped when queue is full regardless of flow or importance11/29/200724• Lock-out problem– Allows a few flows to monopolize the queue space– Send more, get more No implicit policing• Full queues– TCP detects congestion from loss– Forces network to have long standing queues in steady-state– Queueing delays – bad for time sensitive traffic– Synchronization: end hosts react to same events• Full queue empty Full empty…• Poor support for bursty trafficFIFO + Drop-tail Problems5Lock-out Problem• Priority queueing can solve some problems– Starvation– Determining priorities is hard• Simpler techniques: Random drop– Packet arriving when queue is full causes some random packet to be dropped• Drop front– On full queue, drop packet at head of queue• Random drop and drop front solve the lock-out problem but not the full-queues problem6Random Early Detection (RED)• Drop packets before queue becomes full (early drop)• Detect incipient congestion• Avoid window synchronization– Randomly mark packets• Random drop helps avoid bias against burstytraffic11/29/200737RED Algorithm• Maintain running average of queue length• If avg < minthdo nothing– Low queuing, send packets through• If avg > maxth, drop packet– Protection from misbehaving sources• Else mark packet in a manner proportional to queue length– Notify sources of incipient congestion8RED OperationMin threshMax threshAverage Queue LengthminthmaxthmaxP1.0Avg queue lengthP(drop)9Fair Queuing: Goals• How do you protect the most important packets?– How do you provide some isolation in general?– Simple priority queuing does not help• Two approaches:– Fair Queuing (in itself is sufficient)– Leaky bucket (with other techniques which we will cover next class)• FQ key goal: Allocate resources “fairly”– Keep separate queue for each flow• Isolate ill-behaved users• Still achieve statistical muxing– One flow can fill entire pipe if no contenders–Work conserving scheduler never idles link if it has a packet11/29/2007410What is “Fairness”?• At what granularity?– Flows, connections, domains?• What if users have different RTTs/links/etc.– TCP is “RTT-Fair”• BW inversely proportional to RTT of flow– Should they share a link equally or be TCP-fair?• Maximize fairness index?– Fairness = (Σxi)2/n(Σxi2) 0<fairness<111Max-min Fairness• Allocate user with “small” demand what it wants, evenly divide unused resources to “big”users• Formally:– Resources allocated in terms of increasing demand– No source gets resource share larger than its demand– Sources with unsatisfied demands get equal share of resource12Implementing Max-min Fairness• Use separate queues per flow– Round-robin scheduling across queues• Why not simple round robin at packet level?– Variable packet length can get more service by sending bigger packets• Ideally: Bitwise round robin among all queues11/29/2007513Bit-by-bit RR Illustration• Not feasible to interleave bits on real networks– FQ simulates bit-by-bit RR14Bit-by-bit RR Simulation• Single flow: clock ticks when a bit is transmitted. For packet i:– Pi= length, Ai= arrival time, Si= begin transmit time, Fi= finish transmit time– Fi= Si+Pi= max (Fi-1, Ai) + Pi• Multiple flows: clock ticks when a bit from all active flows is transmitted round number– Can calculate Fifor each packet if number of flows is know at all times• Why do we need to know flow count? need to know A This can be complicated15Fair Queuing• Mapping bit-by-bit schedule onto packet transmission schedule• Transmit packet with the lowest Fiat any given time11/29/2007616FQ IllustrationFlow 1Flow 2Flow nI/PO/PVariation: Weighted Fair Queuing (WFQ) Flows can have weightKey to QoS, as we will see in next class.17No Pre-emptionF=10Flow 1(arriving)Flow 2transmittingF=2OutputF=5F=8Flow 1 Flow 2OutputF=10Cannot preempt packetcurrently being transmitted18Fair Queuing Tradeoffs• FQ can control congestion by monitoring flows– Need flows to be adaptive to avoid congestion collapse• Complex state– Must keep queue per flow• Hard in routers with many flows (e.g., backbone routers)• Flow aggregation is a possibility (e.g. do fairness per domain)• Complex computation– Classification into flows may be hard– Must keep queues sorted by finish times– Must track number of flows at fine time scales11/29/2007719Token Bucket for Traffic Policing at the Network EdgeTokens enter bucket at rate rBucket depth b: capacity of bucketOverflowTokensTokensPacketEnough tokens packet goes through,tokens removedTokensPacketNot enough tokens wait for tokens to accumulate20Token Bucket Characteristics• On the long run, rate is limited to r• On the short run, a burst of size b can be sent• Amount of traffic entering at interval T is bounded by:– Traffic = b + r*T– Can provide a lose sense of isolation among flows.• Especially because the send rate of each flow is throttled at the source• Still need some mechanism within the network to ensure performance
View Full Document
Unlocking...