Feng Gang National Laboratory of Communication,UESTC Aug 2017 Ver 1.4 Unit 5 Buffer Management 2616009:Network Traffic Engineering 5:Buffer Management Page.1
2616009: Network Traffic Engineering Feng Gang National Laboratory of Communication, UESTC Aug 2017 Ver 1.4 5: Buffer Management Page.1 Unit 5 Buffer Management
Feng Gang National Laboratory of Communication,UESTC Aug 2017 Ver 1.4 Outline Why packet drop 。 Classification of drop strategies 。】 IP Active Queue Management(AQM) RED FRED BLUE -RIO Case study:Performance of Selective Discard 2616009:Network Traffic Engineering 5:Buffer Management Page.2
2616009: Network Traffic Engineering Feng Gang National Laboratory of Communication, UESTC Aug 2017 Ver 1.4 5: Buffer Management Page.2 Outline • Why packet drop • Classification of drop strategies • IP Active Queue Management(AQM) - RED - FRED - BLUE - RIO • Case study: Performance of Selective Discard
Feng Gang National Laboratory of Communication,UESTC Aug 2017 Ver 1.4 Packet dropping 。 Packets that cannot be served immediately are buffered Full buffers =packet drop strategy Packet losses happen almost always from best-effort connections (why?) Shouldn't drop packets unless imperative packet drop wastes resources(why?) 2616009:Network Traffic Engineering 5:Buffer Management Page.3
2616009: Network Traffic Engineering Feng Gang National Laboratory of Communication, UESTC Aug 2017 Ver 1.4 5: Buffer Management Page.3 Packet dropping • Packets that cannot be served immediately are buffered • Full buffers => packet drop strategy • Packet losses happen almost always from best-effort connections (why?) • Shouldn’t drop packets unless imperative – packet drop wastes resources (why?)
Feng Gang National Laboratory of Communication,UESTC Aug 2017 Ver 1.4 Classification of drop strategies Degree of aggregation Drop priorities Early or late Drop position 2616009:Network Traffic Engineering 5:Buffer Management Page.4
2616009: Network Traffic Engineering Feng Gang National Laboratory of Communication, UESTC Aug 2017 Ver 1.4 5: Buffer Management Page.4 Classification of drop strategies • Degree of aggregation • Drop priorities • Early or late • Drop position
Feng Gang National Laboratory of Communication,UESTC Aug 2017 Ver 1.4 Degree of aggregation Degree of discrimination in selecting a packet to drop E.g.in vanilla FIFO,all packets are in the same class Instead,can classify packets and drop packets selectively The finer the classification the better the protection Max-min fair allocation of buffers to classes drop packet from class with the longest queue(why?) 2616009:Network Traffic Engineering 5:Buffer Management Page.5
2616009: Network Traffic Engineering Feng Gang National Laboratory of Communication, UESTC Aug 2017 Ver 1.4 5: Buffer Management Page.5 Degree of aggregation • Degree of discrimination in selecting a packet to drop • E.g. in vanilla FIFO, all packets are in the same class • Instead, can classify packets and drop packets selectively • The finer the classification the better the protection • Max-min fair allocation of buffers to classes - drop packet from class with the longest queue (why?)
Feng Gang National Laboratory of Communication,UESTC Aug 2017 Ver 1.4 Drop priorities Drop lower-priority packets first 。 How to choose? endpoint marks packets - regulator marks packets congestion loss priority(CLP)bit in packet header Marked packets Policer also Switch preferentially Discards Source marks marks packets marked packets some packets 2616009:Network Traffic Engineering 5:Buffer Management Page.6
2616009: Network Traffic Engineering Feng Gang National Laboratory of Communication, UESTC Aug 2017 Ver 1.4 5: Buffer Management Page.6 Drop priorities • Drop lower-priority packets first • How to choose? - endpoint marks packets - regulator marks packets - congestion loss priority (CLP) bit in packet header Source marks some packets Marked packets Policer also marks packets Switch preferentially Discards marked packets
Feng Gang National Laboratory of Communication,UESTC Aug 2017 Ver 1.4 CLP bit:pros and cons Pros if network has spare capacity,all traffic is carried - during congestion,load is automatically shed Cons - separating priorities within a single connection is hard - what prevents all packets being marked as high priority? 2616009:Network Traffic Engineering 5:Buffer Management Page.7
2616009: Network Traffic Engineering Feng Gang National Laboratory of Communication, UESTC Aug 2017 Ver 1.4 5: Buffer Management Page.7 CLP bit: pros and cons • Pros - if network has spare capacity, all traffic is carried - during congestion, load is automatically shed • Cons - separating priorities within a single connection is hard - what prevents all packets being marked as high priority?
Feng Gang National Laboratory of Communication,UESTC Aug 2017 Ver 1.4 Drop Position Can drop a packet from head,tail,or random position in the queue Tail - easy default approach Head harder lets source detect loss earlier source destination Next packet dropped packet Previously served to arrive Creates "hole" packet ACKs 2616009:Network Traffic Engineering 5:Buffer Management Page.8
2616009: Network Traffic Engineering Feng Gang National Laboratory of Communication, UESTC Aug 2017 Ver 1.4 5: Buffer Management Page.8 Drop Position • Can drop a packet from head, tail, or random position in the queue • Tail - easy - default approach • Head - harder - lets source detect loss earlier . . . source destination Next packet to arrive dropped packet Creates “hole” Previously served packet ACKs
Feng Gang National Laboratory of Communication,UESTC Aug 2017 Ver 1.4 Drop Position (cont.) Random -hardest if no aggregation,hurts hogs most unlikely to make it to real routers 。 Drop entire longest queue easy almost as effective as drop tail from longest queue 2616009:Network Traffic Engineering 5:Buffer Management Page.9
2616009: Network Traffic Engineering Feng Gang National Laboratory of Communication, UESTC Aug 2017 Ver 1.4 5: Buffer Management Page.9 Drop Position (cont.) • Random - hardest - if no aggregation, hurts hogs most - unlikely to make it to real routers • Drop entire longest queue - easy - almost as effective as drop tail from longest queue
Feng Gang National Laboratory of Communication,UESTC Aug 2017 Ver 1.4 Router Support For Congestion Management Traditional Internet Congestion control mechanisms at end-systems, mainly implemented in TCP Routers play little role ·Traditional routers FIFO -Tail drop 2616009:Network Traffic Engineering 5:Buffer Management Page.10
2616009: Network Traffic Engineering Feng Gang National Laboratory of Communication, UESTC Aug 2017 Ver 1.4 5: Buffer Management Page.10 Router Support For Congestion Management • Traditional Internet - Congestion control mechanisms at end-systems, mainly implemented in TCP - Routers play little role • Traditional routers - FIFO - Tail drop