Congestion Control Algorithms = General Principles of Congestion Control = Congestion Prevention Policies = Approaches to Congestion Control * Network Provisioning + Traffic Aware Routing * Admission Control * Traffic Throttling * Load Shedding = Traffic Control Algorithm — Leaky bucket & Token bucketCongestion = Too many packets present in (a part of) the network causes packet delay and loss that degrades performance. This situation is called congestion. * The network and transport layers share the responsibility for handling congestion. Since congestion occurs within the network, it is the network layer that directly experiences it and must ultimately determine what to do with the excess packets. “capacity of subnet Desirable Congested Packets delivered Packets sent = When too much traffic is offered, congestion sets in and performance degrades sharply.General Principles of Congestion Control A. Monitor the system . — detect when and where congestion occurs. B. Pass information to where action can be taken. C. Adjust system operation to correct the problem.Congestion Prevention Policies Policies that affect congestion Layer Policies Transport ‘* Retransmission policy * Out-of-order caching policy * Acknowledgement policy * Flow control policy + Timeout determination Network * Virtual circuits versus datagram inside the subnet * Packet queueing and service policy * Packet discard policy * Routing algorithm « Packet lifetime management Data link ‘* Retransmission policy * Out-of-order caching policy * Acknowledgement policy + Flow control policyApproaches to Congestion Control * The presence of congestion means that the load is (temporarily) greater than the resources (in a part of the network) can handle. Two solutions come to mind: increase the resources or decrease the load. As shown in Fig,, these solutions are usually applied on different time scales to either prevent congestion or react to it once it has occurred Network — Traffic-aware Admission _—_Traffic Load provisioning routing control throttling shedding — SA tt tt Slower Faster (Preventative) (Reactive) * The most basic way to avoid congestion is to build a network that is well matched to the traffic that it carries. If there is a low-bandwidth link on the path along which most traffic is directed, congestion is likely. * Sometimes resources can be added dynamically when there is serious congestion for example, turning on spare routers or enabling lines that are normally used only as backups (to make the system fault tolerant) or purchasing bandwidth on the open market. = More often, links and routers that are regularly heavily utilized are upgraded at the earliest opportunity. This is called provisioning and happens on a time scale of months, driven by long-term traffic trends.Approaches to Congestion Control * To make the most of the existing network capacity, routes can be tailored to traffic patterns that change during the day as network users wake and sleep in different time zones. For example, routes may be changed to shift traffic away from heavily used paths by changing the shortest path weights. Some local radio stations have helicopters flying around theit cities to report on road congestion to make it possible for their mobile listeners to route their packets (cars) around hotspots. This is called traffic-aware routing. Splitting traffic across multiple paths is also helpful. * However, sometimes it is not possible to increase capacity. The only way then to beat back the congestion is to decrease the load. In a virtual-circuit network, new connections can be refused if they would cause the network to become congested. This is called admission control. * Ata finer granularity, when congestion is imminent the network can deliver feedback to the sources whose traffic flows are responsible for the problem. The network can request these sources to throttle their traffic, or it can slow down the traffic itself.Approaches to Congestion Control ‘Two difficulties with this approach are how to identify the onset of congestion, and how to inform the source that needs to slow down. To tackle the first issue, routers can monitor the average load, queueing delay, or packet loss. In all cases, rising numbers indicate growing congestion. To tackle the second issue, routers must participate in a feedback loop with the sources. For a scheme to work correctly, the time scale must be adjusted carefully. If every time two packets arrive in a row, a router yells STOP and every time a router is idle for 20usec , it yells GO, the system will oscillate wildly and never converge. On the other hand, if it waits 30 minutes to make sure before saying anything, the congestion- control mechanism will react too sluggishly to be of any use. Delivering timely feedback is a nontrivial matter. An added concern is having routers send more messages when the network is already congested. Finally, when all else fails, the network is forced to discard packets that it cannot deliver, The general name for this is load shedding. A good policy for choosing which packets to discard can help to prevent congestion collapse.Congestion Control in Virtual-Circuit Subnets: Admission control Congestion A —_——_* oN / Virtual 7” circuit Congestion @ o) (a) Acongested subnet. (b) A redrawn subnet, eliminates congestion and a virtual circuit from A to B.Admission Control = Just saw the resource reservation, but how can the sender specify required resources? = Also, some applications are tolerant of occasional lapses is QoS and apps might not know what its CPU requirements. * Hence routers must convert a set of specifications to resource requirements and then decide whether to accept or reject the flow. Parameter Unit Token bucket rate Bytes/sec Token bucket size Bytes Peak data rate Bytes/seo Minimum packet size | Bytes Maximum packet size | Bytes Example of flow specification,Congestion Control in Datagram Subnets: Traffic Throttling = The old DECNET and frame relay networks: = A warning bit is sent back in the ack to the source in the case congestion. Every router on the path can set the warning bit. = aug +d-ayf Unew * Each router monitors its utilization u based on its temporary utilization f (either 0 or 1). = ais a forgetness rate. = Ifwis above a threshold, a warning state is reached.Hop-by-Hop Choke Packets (in high speed nets) It takes 30 ms for a choke packet to get from NY to SF, For a 155 Mbps, 4.6 Mbps gets in the pipe. (a) A choke packet that affects only the source. (b) A choke packet that affects each hop it passes through.Dropping packets Load shedding: Wine Vs. Milk Wine: drop new packets (keep old); good for file transfer Milk: drop old packets (keep new); good for multimedia Random Early Detection When the average queue length exceeds a threshold, packets are picked at random from the queue and discarded.Jitter Control 3 3 a a 3 3 £ § £ High jitter £ =| Low iter win Delay —» Delay ay (due to speed of light) f@) (b) (a) High jitter. (b) Low jitter.Requirements Application Reliability | Delay Jitter Bandwidth E-mail High Low Low Low File transfer High Low Low Medium Web access High Medium | Low Medium Remote login High Medium | Medium | Low Audio on demand Low Low High Medium Video on demand | Low Low High High Telephony Low High High Low Videoconferencing | Low High High High = How stringent the quality-of-service requirements are shown in Fig.Traffic Control The Leaky Bucket Algorithm ae ‘The bucket eocting —-| 9. tat leaky bucket ackats Unrogulatos tow = —Rogulatod ‘ow a. o Water sips out of he oO tale 2 contort re —~ roy o (a) A leaky bucket with water. (b) a leaky bucket with packets.The Token Bucket Algorithm computor computor 1 One token a eacies | tweet aoe i ses 1am 5 (a) Before. (b) After. = Token bucket allows some burstiness (up to the number of token the bucket can hold)