Chapter 6 Congestion Control and Resource Allocation

Chapter 6 Congestion Control and Resource Allocation

6.3 TCP Congestion Control

 Additive Increase/Multiplicative Decrease (AIMD)

o Basic idea: repeatedly increase transmission rate until congestion occurs; reduce transmission rate when congestion occurs.

o New state variable per connection: CongestionWindow

 CongestionWindow limits how much data source is allowed to have in transit in order to not cause congestion.

 MaxWindow = MIN(CongestionWindow, AdvertisedWindow) is the maximum number of unACKed bytes allowed.

 EffeciveWindow = MaxWindow – (LastByteSent – LastByteAcked) o The AIMD algorithm

 Increase CongestionWindow by one MSS per RTT when all packets sent during the last RTT have been ACKed (additive increase)  Divide CongestionWindow by two when a timeout occurs

(multiplicative decrease)

 Source interprets timeouts as a sign of congestion

 CongestionWindow not allowed to fall below MSS

 In practice, CongestionWindow is incremented by a fraction of MSS when an ACK is received.

 Increment = MSS x (MSS/CongestionWindow)

 Slow Start

o Slow start is used at the beginning of a connection to increase the congestion window rapidly.

o The slow start algorithm:

 Begin with CongestionWindow = 1 MSS  Double CongestionWindow each RTT

o Slow start is also used after a timeout occurs. Upon timeout:

 Threshold is set to 1/2 of CongestionWindow just before timeout.  CongestionWindow set to 1 MSS.

 CongestionWindow grows exponentially to Threshold (slow start), then grows linearly (additive increase).

 Fast Retransmit and Fast Recovery o Fast retransmit

 Goal: trigger retransmission of a dropped packet sooner than the regular timeout mechanism.

 Trigger of retransmission: three duplicate ACKs  Timeouts cannot be eliminated because

 There may not be enough packets in transit to cause 3 duplicate ACKs when the window size is small.

 A large number of packets may be lost.

o Fast recovery: skip the slow start phase when fast retransmit detects a lost packet

 Threshold set to CongestionWindow/2.

 CongestionWindow set to Threshold, window grows linearly.

 Summary of TCP congestion control

o When CongestionWindow is below Threshold, window grows exponentially.

o When CongestionWindow is above Threshold, window grows linearly. o When timeout occurs, Threshold is set to CongestionWindow/2 and

CongestionWindow is set to 1 MSS.

o When a triple duplicate ACK occurs, Threshold is set to

CongestionWindow/2 and CongestionWindow is set to Threshold. 6.4 Congestion-Avoidance Mechanisms

 Congestion avoidance

o Predict when congestion is about to happen.

 Two congestion avoidance approaches

o With router assistance: DECbit and RED. o Without router assistance: TCP Vegas.

 DECbit

o Add a congestion bit to packet header. o Router actions

 Monitor average queue length over last busy + idle cycle, plus the current busy cycle.

 Set congestion bit if average queue length 1 when the packet arrives.

 A queue length of 1 maximizes throughput/delay. o End host actions

 Destination host echoes congestion bit back to source host.  Source records the fraction of the last window’s worth of packets

that resulted in the congestion bit being set.

 If fraction < 50%, increase congestion window by 1 packet.

 If fraction ≥ 50%, decrease congestion window to 0.875 times previous value.

 A threshold of 50% maximizes throughput/delay.

 Random Early Detection (RED)

o Implicit notification: router drops packets when congestion is imminent.  Designed to be used in conjunction with TCP.

o Early random drop: router drops each arriving packet with some drop probability whenever the queue length exceeds a threshold.

o RED details

 Average queue length computation

 AvgLen = (1 - Weight) x AvgLen + Weight x SampleLen, 0<Weight<1.

 Two queue length thresholds

 If AvgLen MinThreshold, queue the packet.

 If MinThreshold < AvgLen < MaxThreshold, calculate

probability P and drop the arriving packet with probability P.

 If MaxThreshold AvgLen, drop the arriving packet.  Computing probability P

 TempP = MaxP x (AvgLen – MinThreshold) /(MaxThreshold - MinThreshold).

 P = TempP/(1 – count x Temp), count = number of newly arriving packets that have been queued since the last drop.

o Using count in the calculation ensures a roughly even distribution of drops over time.

 RED provides fair resource allocation because the probability of dropping a particular flow’s packets is roughly proportional to the share of the bandwidth the flow is currently getting at the router.

 TCP Vegas

o Goal: maintain the right amount of extra data in the network.

 Extra data – data that the source would not have transmitted had it been trying to match exactly the available bandwidth of the network.

 Too much extra data: long delays and congestion.

 Too little extra data: cannot respond rapidly enough to transient increases in the available network bandwidth.

o The algorithm

 Define BaseRTT to be the RTT of a packet when the flow is not congested.

 BaseRTT is set to the minimum of all measured RTTs.  Calculate expected throughput

 Calculates current sending rate, ActualRate, once per RTT.

 ActualRate = K/(t1-t2), where t1 is the time a packet is sent, t2 is the time the ACK is received, K is the number of bytes transmitted between t1 and t2.

 Set congestion window based on ActualRate and ExpectedRate.

 Diff = ExpectedRate – ActualRate (Diff ≥ 0)

 If Diff < α, increase CongestionWindow linearly during next RTT.

 If Diff > β, decrease CongestionWindow linearly during next RTT.

 If α < Diff < β, leave CongestionWindow unchanged.  Use multiplicative decrease when a timeout occurs.

6.2 Queuing Disciplines

 Router implements queuing discipline that consists of

o Scheduling discipline: determine the order in which packets are transmitted. (allocate link bandwidth)

o Drop policy: determine which packets get discarded. (allocate buffer space)

 FIFO Queuing

o The packets are transmitted in the order they arrive. o Tail drop.

 Priority Queuing - a variation of FIFO Queuing o Each packet is marked with a priority.

o Routers implement one FIFO queue for each priority class. o Queues are serviced in strict order of queue priority. o Starvation problem.

 Fair Queuing (FQ)

o Maintain a separate queue for each flow, queues serviced in round-robin.

 Segregate traffic so that an ill-behaved source does not interfere with well-behaved sources.

o Approximate bit-by-bit round robin to fairly allocate link bandwidth.  Imagine a clock that ticks once each time a bit is transmitted from

all of the active flows.  For a single flow

 Let Pi denote the length of packet i.

 Let Si denote the time when the router starts to transmit

packet i.

 Let Fi denote the time when the router finishes transmitting

packet i.

 Let Ai denote the time when packet i arrives at the router.

 Fi = Si +Pi and Si = MAX (Fi-1 , Ai), so Fi = MAX (Fi-1 , Ai) + Pi .

 For multiple flows

 For each flow, calculate Fi for each packet.

 Treat all Fi ’s as timestamps, next packet to transmit is one

with lowest timestamp.

 Not perfect: can’t preempt a packet that is currently being transmitted.

o Properties of FQ

 Work-conserving: the link is never left idle as long as there is at least one packet in the queue.

 Fair bandwidth allocation: when there are n flows, each flow gets 1/nth of the link bandwidth.

 Weighted Fair Queuing (WFQ) - a variation of FQ o A weight is assigned to each flow (queue)

 Weight specifies how many bits to transmit each time the router services that queue

o Bandwidth received by a flow depends on its weight and the number of flows that are sharing the link.