Evaluation of TCP BBR v2 Congestion Control Using Network Emulation - - PowerPoint PPT Presentation

▶
evaluation of tcp bbr v2 congestion control using network
SMART_READER_LITE
LIVE PREVIEW

Evaluation of TCP BBR v2 Congestion Control Using Network Emulation - - PowerPoint PPT Presentation

Nov 19, 2023 395 likes •594 views

Chair of Network Architectures and Services Department of Informatics Technical University of Munich Evaluation of TCP BBR v2 Congestion Control Using Network Emulation Intermediate talk for the Masters Thesis by Juliane Aulbach advised by

slide-1
SLIDE 1

Chair of Network Architectures and Services Department of Informatics Technical University of Munich

Evaluation of TCP BBR v2 Congestion Control Using Network Emulation

Intermediate talk for the Master’s Thesis by

Juliane Aulbach

advised by Benedikt Jaeger, Dominik Scholz Monday 25th May, 2020 Chair of Network Architectures and Services Department of Informatics Technical University of Munich

slide-2
SLIDE 2

Congestion Control

loss-based

  • perating point

Kleinrock’s optimal

  • perating point

BDP BDP+BtlneckBufSize RTprop

RTT

BtlBw

Amount Inflight Delivery Rate Kleinrock’s Optimal Operating Point [3]

  • J. Aulbach — TCP BBR v2

2

slide-3
SLIDE 3

TCP BBR

Bottleneck-Bandwidth & RTT Congestion-based Congestion Control

  • BBR: Bottleneck Bandwidth and RTT
  • 2016: Google presents BBR [1]
  • 2017: BBR becomes part of the Linux kernel (version 4.9 or higher)
  • 2018: Google announces BBR v2

How does BBR work?

  • Maximum bandwidth is determined by a single bottleneck
  • Estimates bandwidth-delay product (BDP) to create a model of the network
  • BDP = Bottleneck Bandwidth (BtlBw) · Propagation delay (RTprop)
  • max BtlBw = windowed_max(delivered data / elapsed time)
  • min RTT = windowed_min(time acknowledged - time send)
  • Continuosly updates model according to changes in the measurements
  • J. Aulbach — TCP BBR v2

3

slide-4
SLIDE 4

TCP BBR

BBR Phases Start Startup Drain ProbeBW ProbeRTT 10 20 30 5 10 15 20 Time / s Sending Rate / Mbit

s

BBR

Startup Drain Probe BW Probe RTT

  • J. Aulbach — TCP BBR v2

4

slide-5
SLIDE 5

TCP BBR

Bottleneck-Bandwidth & RTT Issues with initial version of BBR The following list is presented by Neil Cardwell at IETF Meeting 104 [2]

  • Low throughput for Reno/CUBIC flows sharing a bottleneck with a bulk BBR flow
  • > Adapt bandwidth probing
  • J. Aulbach — TCP BBR v2

5

slide-6
SLIDE 6

TCP BBR

Bottleneck-Bandwidth & RTT Issues with initial version of BBR The following list is presented by Neil Cardwell at IETF Meeting 104 [2]

  • Low throughput for Reno/CUBIC flows sharing a bottleneck with a bulk BBR flow
  • > Adapt bandwidth probing
  • Loss-agnostic
  • > Use packet loss as an explicit signal, with an explicit target loss rate ceiling
  • ECN-agnostic
  • > Use DCTCP-style ECN, if available, to help keep queues short
  • J. Aulbach — TCP BBR v2

5

slide-7
SLIDE 7

TCP BBR

Bottleneck-Bandwidth & RTT Issues with initial version of BBR The following list is presented by Neil Cardwell at IETF Meeting 104 [2]

  • Low throughput for Reno/CUBIC flows sharing a bottleneck with a bulk BBR flow
  • > Adapt bandwidth probing
  • Loss-agnostic
  • > Use packet loss as an explicit signal, with an explicit target loss rate ceiling
  • ECN-agnostic
  • > Use DCTCP-style ECN, if available, to help keep queues short
  • Problems with ACK aggregation
  • > Estimate recent degree of aggregation to match CUBIC throughput
  • J. Aulbach — TCP BBR v2

5

slide-8
SLIDE 8

TCP BBR

Bottleneck-Bandwidth & RTT Issues with initial version of BBR The following list is presented by Neil Cardwell at IETF Meeting 104 [2]

  • Low throughput for Reno/CUBIC flows sharing a bottleneck with a bulk BBR flow
  • > Adapt bandwidth probing
  • Loss-agnostic
  • > Use packet loss as an explicit signal, with an explicit target loss rate ceiling
  • ECN-agnostic
  • > Use DCTCP-style ECN, if available, to help keep queues short
  • Problems with ACK aggregation
  • > Estimate recent degree of aggregation to match CUBIC throughput
  • Throughput variations during ProbeRTT
  • > Less restrictive constraints for sending rate if entering ProbeRTT
  • J. Aulbach — TCP BBR v2

5

slide-9
SLIDE 9

Parameters & Measurements

Mininet Setup

Sender1 Sender2 . . . Sendern Switch1 Switch2 Switch3 Receiver1 Receiver2 . . . Receivern Bottleneck Link Mininet setup with sending and receiving hosts connected via a bottleneck link [3]

  • J. Aulbach — TCP BBR v2

6

slide-10
SLIDE 10

Results

RTT Unfairness 10 20 30 40 50 5 10 Time / s Sending Rate / Mbit

s

BBR 10 ms RTT 50 ms RTT Fair Share

  • J. Aulbach — TCP BBR v2

7

slide-11
SLIDE 11

Results

RTT Unfairness 10 20 30 40 50 5 10 Time / s Sending Rate / Mbit

s

BBR 10 ms RTT 50 ms RTT Fair Share 10 20 30 40 50 5 10 Time / s Sending Rate / Mbit

s

BBR v2 10 ms RTT 50 ms RTT Fair Share

  • J. Aulbach — TCP BBR v2

7

slide-12
SLIDE 12

Results

Buffersize 10−1 100 101 20 40 60 80 100 Buffersize in multiples of BDP [log] Bandwidth

  • Avg. [%]

CUBIC vs BBR

CUBIC BBR

10−1 100 101 20 40 60 80 100 Buffersize in multiples of BDP [log] Bandwidth

  • Avg. [%]

CUBIC vs BBR v2

CUBIC BBR v2

  • J. Aulbach — TCP BBR v2

8

slide-13
SLIDE 13

Results

Buffersize 10−1 100 101 20 40 60 80 100 Buffersize in multiples of BDP [log] Bandwidth

  • Avg. [%]

CUBIC vs BBR

CUBIC BBR

10−1 100 101 20 40 60 80 100 Buffersize in multiples of BDP [log] Bandwidth

  • Avg. [%]

CUBIC vs BBR v2

CUBIC BBR v2

10−1 100 101 20 40 60 80 100 Buffersize in multiples of BDP [log] Bandwidth

  • Avg. [%]

BBR vs BBR v2

BBR BBR v2

  • J. Aulbach — TCP BBR v2

8

slide-14
SLIDE 14

Conclusion

Challenges

  • Mininet is not working properly on current Linux kernel
  • There is no official draft or paper for BBR v2

Status Task Status Find out how BBR and BBR v2 work

  • Deploy Linux with BBR v2
  • Reproduce results from [3] as ground trouth
  • Repeat all tests with BBR v2
  • Compare test results from BBR and BBR v2
  • Evaluate improvements and deteriorations
  • Extend measurements
  • J. Aulbach — TCP BBR v2

9

slide-15
SLIDE 15

Backup Slides

BBR Phases Start Startup Drain ProbeBW ProbeRTT 10 20 30 5 10 15 20 Time / s Sending Rate / Mbit

s

BBR

Startup Drain Probe BW Probe RTT

10 20 30 5 10 15 20 Time / s Sending Rate / Mbit

s

BBR v2

Startup Drain ProbeBW ProbeRTT

  • J. Aulbach — TCP BBR v2

10

slide-16
SLIDE 16

Backup Slides

Fairness: Round-trip time

0.5 0.75 1 Ftp

CUBIC vs BBR

101 102 103 25 50 75 RTT / ms [log]

Bandwidth

  • Avg. [%]

CUBIC BBR

0.5 0.75 1 Ftp

CUBIC vs BBR

101 102 103 25 50 75 RTT [log ms]

Bandwidth

  • Avg. [%]

CUBIC BBR v2

0.5 0.75 1 Ftp

BBR vs BBR2

101 102 103 25 50 75 RTT [log ms]

Bandwidth

  • Avg. [%]

BBR BBR v2

  • J. Aulbach — TCP BBR v2

11

slide-17
SLIDE 17

Backup Slides

Performance 10−3 10−2 10−1 100 101 102 50 100 Loss Rate (%) - Log Scale Throughput Mbit/s Fully use bandwidth, despite high loss

CUBIC BBR

Bottleneck Bandwidth = 100Mbit, RTT = 100ms 2,000 4,000 6,000 8,000 10,000 2,000 4,000 6,000 Buffersize / kB Latency / ms Fully use bandwidth, despite high loss

Cubic BBR

Bottleneck Bandwidth = 10Mbit, RTT = 40ms

  • J. Aulbach — TCP BBR v2

12

slide-18
SLIDE 18

Bibliography

[1]

  • N. Cardwell and Y. Cheng.

Making Linux TCP Fast: 04_Making_Linux_TCP_Fast_netdev_1.2_final, 2016. [2]

  • N. Cardwell, Y. Cheng, S. H. Yeganeh, P

. Jha, M. Mathis, and van Jacobson. BBR v2 - A Model-based Congestion Control: slides-104-iccrg-an-update-on-bbr-00, 2019. [3]

  • B. Jaeger, D. Scholz, D. Raumer, F. Geyer, and G. Carle.

Reproducible measurements of TCP BBR congestion control. Computer Communications, 144:31–43, 2019.

  • J. Aulbach — TCP BBR v2

13