ECN Experiment
Explicit Congestion Notification (ECN) enables queues to mark packets instead of dropping them in case of congestion. Congestion Control Algorithms (CCAs) typically react to marked packets in a comparable way to packet loss (via duplicate acknowledgements): the CCAs decrease their sending rates to avoid congestion. Thereby, a CCA may slow down and avoid congestion without recovering from congestion losses. ECN is widely used in the data center and has been popularized by Datacenter TCP (DCTCP).
Scenario
In the ECN experiment, one or more flows operate in a static dumbbell network. The flows generate greedy source traffic, use CCAs and have ECN enabled. At the bottleneck, a Controlled delay (CoDel) queueing discipline (qdisc) is used.
The experiment has two parameters: the number of flows k
and the CE threshold ce_threshold of the CoDel qdisc.
To summarize the experiment setup:
Topology: Dumbbell topology with static network parameters and a CoDel qdisc at the bottleneck
Flows: One or more flows that use CCAs
Traffic Generation Model: Greedy source traffic
Experiment Results
Experiment #24
Parameters
Command: ns3-dev-ccperf-static-dumbbell-default --experiment-name=ecn --db-path=benchmark_TcpNewReno.db '--parameters={aut:TcpNewReno,k:1,ce_threshold:1ms}' --aut=TcpNewReno --stop-time=15s --seed=42 --bw=16Mbps --loss=0.0 --qlen=20p --qdisc=CoDelQueueDisc --rtts=15ms --sources=src_0 --destinations=dst_0 --protocols=TCP --algs=TcpNewReno --recoveries=TcpPrrRecovery --start-times=0s --stop-times=15s '--traffic-models=Greedy(bytes=0)' --ns3::TcpSocketBase::UseEcn=On --ns3::CoDelQueueDisc::UseEcn=true --ns3::CoDelQueueDisc::CeThreshold=1ms
Flows
| src | dst | transport_protocol | cca | cc_recovery_alg | traffic_model | start_time | stop_time |
|---|---|---|---|---|---|---|---|
| src_0 | dst_0 | TCP | TcpNewReno | TcpPrrRecovery | Greedy(bytes=0) | 0.00 | 15.00 |
Metrics
The following tables list the flow, link, and network metrics of experiment #24. Refer to the the metrics page for definitions of the listed metrics.
Flow Metrics
Flow metrics capture the performance of an individual flow. They are measured at the endpoints of a network path at either the source, the receiver, or both. Bold values indicate which flow achieved the best performance.
| Metric | flow_1 |
|---|---|
| cov_in_flight_l4 | 0.38 |
| cov_throughput_l4 | 0.36 |
| flow_completion_time_l4 | 15.00 |
| mean_cwnd_l4 | 15.07 |
| mean_delivery_rate_l4 | 10.42 |
| mean_est_qdelay_l4 | 1.14 |
| mean_idt_ewma_l4 | 1.23 |
| mean_in_flight_l4 | 14.64 |
| mean_network_power_l4 | 643.12 |
| mean_one_way_delay_l7 | 2730.71 |
| mean_sending_rate_l4 | 10.43 |
| mean_sending_rate_l7 | 12.56 |
| mean_srtt_l4 | 16.14 |
| mean_throughput_l4 | 10.43 |
| mean_throughput_l7 | 10.43 |
| mean_utility_mpdf_l4 | -0.11 |
| mean_utility_pf_l4 | 2.27 |
| mean_utilization_bdp_l4 | 0.76 |
| mean_utilization_bw_l4 | 0.65 |
| total_retransmissions_l4 | 0.00 |
| total_rtos_l4 | 0.00 |
Link Metrics
Link metrics are recorded at the network links of interest, typically at bottlenecks. They include metrics that measure queue states. Bold values indicate which link achieved the best performance.
| Metric | btl_forward |
|---|---|
| mean_qdisc_delay_l2 | 0.07 |
| mean_qdisc_length_l2 | 0.15 |
| mean_sending_rate_l1 | 10.82 |
| total_qdisc_drops_l2 | 0.00 |
Network Metrics
Network metrics assess the entire network as a
whole by aggregating other metrics, e.g., the aggregated throughput of
all flows.
Hence, the network metrics has only one column named net.
| Metric | net |
|---|---|
| mean_agg_in_flight_l4 | 236.50 |
| mean_agg_throughput_l4 | 3.08 |
| mean_agg_utility_mpdf_l4 | -118.08 |
| mean_agg_utility_pf_l4 | -18.51 |
| mean_agg_utilization_bdp_l4 | 0.09 |
| mean_agg_utilization_bw_l4 | 0.07 |
| mean_entropy_fairness_throughput_l4 | 2.30 |
| mean_jains_fairness_throughput_l4 | 0.67 |
| mean_product_fairness_throughput_l4 | 0.00 |
Figures
The following figures show the results of the experiment #24.Time Series Plot of the Operating Point
Time series plot of the number of segments in flight, the smoothed round-trip time (sRTT), and the throughput at the transport layer.
Mean Operating Point Plane
The mean throughput and mean smoothed round-trip time (sRTT) at the transport layer of each flow.
Distribution of the Operating Point
The empirical cumulative distribution function (eCDF) of the throughput and smoothed round-trip time (sRTT) at the transport layer of each flow.
In Flight vs Mean Operating Point
The mean throughput and mean smoothed round-trip time (sRTT) at the transport layer of each flow. The optimal operating point is highlighted with a star (magenta). The joint operating point is given by the aggregated throughput and the mean sRTT over all flows
Comparison of Congestion Control Algorithms (CCAs)
Figures
