CSIT 560 Internet Infrastructure Switches and Routers Active

CSIT 560 Internet Infrastructure: Switches and Routers Active Queue Management Presented By: Gary Po, Henry Hui and Kenny Chong

Agenda u Why AQM? u In Considerations u AQM Algorithms u Commercial Effort u Conclusions

Why AQM? u Two Classes of Router Algorithms for Congestion Control u What is Congestion? u Congestion avoidance in TCP, is it good enough? u Our Choice : Active Queue Management

What do we consider when implementing AQM? u Qo. S u Link Utilization u Fairness u Implementation – Keep Average queue size small – Bounded Delay – Avoid Global Synchronization – Absorbs bursts without dropping packets – Punishes misbehaving flows – Prevent bias against bursty connections – – Ease of Configurations Buffer Size Requirement (Large or Small) Per-flow State Information Computational Overhead

AQM Algorithms

FIFO + Drop Tail u Problems: – No isolation – No policing – Large queues for high utilizations – Synchronization problem – Lock-out problem

RED (Random Early Detection) u FIFO scheduling Max thresh Min thresh Average queue length Case 1: Make Define Use Admit of Two Average the Threshold New Queue Packet Values Length Average Queue Length < Min. Thresh Value

RED (Cont’d) Max thresh Min thresh p p 1 -p Average queue length Case 2: Average Queue Length between Or Admit Dropthe the. New. Packet. With. Probabilityp… 1 -p Min. and Max. Threshold Value

RED (Cont’d) Max thresh Min thresh Average queue length Until. As the New noaverage new packet orpackets being queue Case could dropped can 3: length be be admitted, drops below New Packet will be dropped Avg. the Queue average the with Length max aqueue aprobability threshold > Max. length Threshold 1 -p… value p… decreases. Value

RED Flow Diagram

RED (Cont’d) u Queue Size versus Time RED: Queue Size Delay is bounded Global Synchronization solved

Unfairness of RED An unresponsive flow occupies over 95% of bandwidth Unresponsive Flow (such as UDP) 32 TCP Flows 1 UDP Flow

CHOKe (CHOose and Keep) u Based on RED u Simple u Designed for fairness u Penalize the unresponsive flow

CHOKe (Cont’d) u Mechanism

CHOKe (Cont’d) Max thresh Min thresh Average queue length Case 1: Admit the New Packet Average Queue Length < Min. Thresh Value

CHOKe (Cont’d) Max thresh Min thresh p 1 -p Average queue length A If packet Case they If 2: they are is Avg. randomly from are Queue from different chosen Length the same flows, from is between flow, the same queue to compare Min. both and logic packets with Max. inthe RED Threshold will new applies be arrival dropped Values packet

CHOKe (Cont’d) Max thresh Min thresh Average queue length AIf If random they are packet from Case will different the 3: be same chosen flows, flow, for Avg. Queue the both new Length packets packet comparison >will Max. willbe be Threshold dropped Value

Evaluate CHOKe’s performance using NS-2

Simulation Scenario source destination 10 Mbps UDP 1 Mbps TCP router • Topology: Dumb-bell • Metrics: throughput and queue size TCP

Performance of CHOKe Unresponsive Flow (UDP) Fair Share Level Bandwidth is evenly shared 32 TCP Flows 1 UDP Flow

Parameters u Number of responsive/unresponsive flows u Transfer rate of different flows u Number of random candidates chosen for comparison

CHOKe Simulation u Different Parameters, different performance CHOKe-1 32 TCPs 1 UDP CHOKe-2 32 TCPs, 3 UDPs of different rate CHOKe-2 32 TCPs, 1 UDP of high rate CHOKe-2 32 TCPs, 3 UDPs of same rate

Evolutions of AQM Algorithms FIFO+ Drop. Tail RED SRED FRED BLUE REM, AVQ, LDC PI Controller u u CHOKe – – – SRED FRED Early congestion Low Sensitivity delay and to traffic small load Stabilized Good Simplicity protection queue from detection queues and drain rate flows occupancy u misbehaving High throughput SFB u No bias u Independence Low delayagainst of the Drawbacks u Protection from Drawbacks bursty traffic of users u number Target delay achieved misbehaving flows u Per-flow state global u early congestion Drawbacks u No Intuitive parameters, Drawbacks synchronization (to Pdrop u RED disadvantages u detection Some meaningful complexity users due to SAC updated only onstate u (target Some per-flow parameters delay) Drawbacks overflow or. Web (zombie list) Drawbacks u Low throughput for u queue Difficulty in link events) parameter setting u Some complexity due to u traffic REDidle disadvantages u Inconsistency with TCP u parameters Slow response Insensitivity to and traffic mechanism; loadthroughput and drain rates u sender Low some dependence onin works cases historybest with ECN u u u BLUE RED – – Merits

Commercial Efforts & Conclusion

Commercial Efforts & Conclusion (Cont’d) u “Applying AQM over 3 G wireless network” – a paper supported by Motorola Canada Ltd. (Mar. 2003) u 3 G network, real-time applications have hard time deadlines for packet delivery at the receiver. u Use AQM to avoid long queuing delay and prevent expiring packets.

Commercial Efforts & Conclusion (Cont’d) u AQM improves overall system performance by increasing throughput and reducing end-to-end delay.

Commercial Efforts & Conclusion (Cont’d) u “Effect of AQM on Web Performance” – a paper supported by Cisco Systems and IBM. (Aug. 2003) u Proportional Integrator (PI) controller Random Exponential Marking (REM) controller u Adaptive Random Early Detection (ARED). u IETF proposed standard : Explicit Congestion Notification (ECN) u

Commercial Efforts & Conclusion (Cont’d) u ECN has significant impact with AQM scheme in web performance. u Many researches and efforts are going on in the field of AQM. u Simple and Easy to implement