Presentation is loading. Please wait.

Presentation is loading. Please wait.

WB-RTO: A Window-Based Retransmission Timeout Ioannis Psaras Demokritos University of Thrace, Xanthi, Greece.

Similar presentations


Presentation on theme: "WB-RTO: A Window-Based Retransmission Timeout Ioannis Psaras Demokritos University of Thrace, Xanthi, Greece."— Presentation transcript:

1 WB-RTO: A Window-Based Retransmission Timeout Ioannis Psaras Demokritos University of Thrace, Xanthi, Greece

2 COMputer NETworks Group (COMNET)2 Outline of the presentation Contributions of this work The current Retransmission Timeout Algorithm Recent Related Work on the subject The proposed algorithm: WB-RTO  The algorithm  Expected Behavior Evaluation Plan Experimental Results

3 COMputer NETworks Group (COMNET)3 Contributions of this work Our perspective: When contention increases, the timeout becomes the scheduler for the link. Our observations: TCP-RTO should not be solely based on RTT estimations. Congestion events cause retransmission synchronization. Wireless burst errors should not be interpreted as congestion events. Our solutions: Approximation of the current level of network contention Estimation of the contribution of each flow to congestion Allowance for asynchronous retransmissions when timeout happens.

4 COMputer NETworks Group (COMNET)4 The current Retransmission Timeout Algorithm Upon each ACK arrival, the sender:  Calculates the RTT Variation:  Updates the expected RTT prior to calculating the timeout:  Calculates the Retransmission Timeout value:

5 COMputer NETworks Group (COMNET)5 Recent Related Work on the subject Eifel RTO Algorithm  Uses the timestamp option to detect a spurious timeout. Forward RTO Algorithm  Uses the first 3 ACKs after the timeout to decide if the timeout was spurious or not. Peak-Hopper RTO Algorithm  Uses 2 timers: one is aggressive and one is conservative. Each time it decides which one to follow. CA-RTO: A Contention-Adaptive RTO  Integrates a contention-adaptive parameter and introduces retransmission randomness.

6 COMputer NETworks Group (COMNET)6 Window-Based RTO (1/4): Proportional Timeout Estimation of the contribution of the flow to congestion: c = f(cwnd, max cwnd ) Compare the current cwnd_ with the max_cwnd_:  If cwnd_ < max_cwnd_ / 2, c = 1: minimal charge  If max_cwnd_ / 2 < cwnd_ < (3/4)* max_cwnd_, c = 1,5: medium charge  If (3/4)* max_cwnd_ < cwnd_ < max_cwnd_, c = 2: major charge

7 COMputer NETworks Group (COMNET)7 Window-Based RTO (2/4): Contention Estimation Flow classification according to its cwnd_ history (awnd_): ai = g(awnd, Thresholdi) where Thresholds 1 to 4 represent different levels of network contention:  Threshold 1 corresponds to very high contention,  Threshold 4 corresponds to low contention. 3. 10 < awnd_ < 30: a3 = 3 4. 30 < awnd_ < 50: a4 = 2 1. awnd_ < 5: a1 = 10 2. 5 < awnd_ < 10: a2 = 5

8 COMputer NETworks Group (COMNET)8 Window-Based RTO (3/4): Timeout Adjustment Calculation of the Window-Based RTO: WB − RTO = random(rtt, c × ai) or WB − RTO = random(rtt, f(cwnd, max cwnd ) × g(awnd, Thresholdi)) 1. rtt, avoids timeout expiration prior to the estimated RTT measurement 2. Parameter c captures the contribution of the flow to congestion 3. Parameter a approximates the current level of flow contention 4. Randomization guarantees asynchronous retransmission attempts

9 COMputer NETworks Group (COMNET)9 Window-Based RTO (4/4): Expected Behavior High penalties result in high timeout values. As awnd_ increases timeout settles to smaller values. but Large windows do not always mean large timeout values. WB-RTO vs awnd_

10 COMputer NETworks Group (COMNET)10 Performance Evaluation Plan WB-RTO is implemented in TCP-Reno Topologies used: Evaluation Scenarios  Simple Wired Scenario  Interaction with Active Queue Management (i.e. RED)  Satellite Environment  Traffic Diversity (Mice with Elephants)

11 COMputer NETworks Group (COMNET)11 An Important Note… WB-RTO does not improve the Goodput performance of TCP significantly, but here we focus on the Retransmission Timeout Algorithm of TCP hence we pay more attention on the combination of the retransmission effort and the Goodput performance, rather than on the Goodput performance alone.

12 COMputer NETworks Group (COMNET)12 Simple Wired Scenario (1) Topology: Dumbbell Queuing Policy: DropTail DBP = Buffer Size = 10 pkts 5 participating flows 1500 sec total simulation time Flows ideal rate = 2 pkts/wnd We trace: Seqno progress, RTT, RTO TCP-RTO WB-RTO

13 COMputer NETworks Group (COMNET)13 Simple Wired Scenario (2) For the TCP-RTO performance, we observe:  RTT stabilization  Similar timeout values  50% more retransmitted pkts  Less Goodput RTT (in secs) RTO (in secs)

14 COMputer NETworks Group (COMNET)14 Interaction with AQM (i.e. RED) (1) Topology: Dumbbell Queuing Policy: RED DBP = Buffer Size = 40 pkts min_thresh = 4 pkts max_thresh = 12 pkts 1500 sec total simulation time Goodput (in B/s) Retransmitted Packets Number of Timeouts

15 COMputer NETworks Group (COMNET)15 Interaction with AQM (i.e. RED) (2) We observe:  Similar Goodput  Significant difference in Retransmission Effort (50%)  WB-RTO results in 66% less timeout expirations  TCP-RTO causes inefficient queue utilization  The average queue length always overcomes the max_thresh, when using TCP- RTO TCP-RTO WB-RTO

16 COMputer NETworks Group (COMNET)16 Satellite Scenario (1) Topology: Cross-Traffic Bottleneck Queuing Policy: RED The rest of the buffers use DT bw_bottleneck = 20Mbps bw_delay = 300ms Buffer Size = 200 pkts min_thresh = 20 pkts max_thresh = 60 pkts 150 sec total simulation time PER = 0,0001 3 blackouts on the backbone link No blackout After 3 blackouts

17 COMputer NETworks Group (COMNET)17 Satellite Scenario (2) We observe that:  TCP-RTO interprets the second timeout as a congestion signal  WB-RTO does not extend the timeout, due to low contention and hence exploits bandwidth faster  TCP-RTO still waits for the extended timeout to occur, while  WB-RTO resumes transmission immediately TCP-RTO WB-RTO

18 COMputer NETworks Group (COMNET)18 Traffic Diversity (Mice and Elephants) (1) Topology: Dumbbell Bottleneck Queuing Policy: RED bw_bottleneck = 10Mbps bw_delay = 30ms Buffer Size = 40 pkts Goodput (KB/s) Retransmitted Packets Goodput per flow (KB/s)

19 COMputer NETworks Group (COMNET)19 Traffic Diversity (Mice and Elephants) (2) We observe:  Simultaneous timeout events for TCP-RTO All flows timeout during the Slow-Start Flows 7-9 timeout simultaneously 10 times during the experiment Short flows: 83 vs 50 timeouts Long flows: 43 vs 12 timeouts We conclude that:  most of the timeouts are spurious  WB-RTO achieves an important goal: it reduces the number of timeouts TCP-RTO WB-RTO

20 COMputer NETworks Group (COMNET)20 Conclusions RTT measurements cannot always reflect the level of network contention TCP-RTO should not be solely based on RTT samples A contention-aware RTO proves to be more efficient, since it is aware of current network conditions. A randomization factor in the RTO schedules retransmissions in a fairer manner.

21 COMputer NETworks Group (COMNET)21 WB-RTO: A Window-Based Retransmission Timeout Thank you!! Presented by Ioannis Psaras e-mail: ipsaras@ee.duth.gripsaras@ee.duth.gr URL: utopia.duth.gr/~ipsarasutopia.duth.gr/~ipsaras


Download ppt "WB-RTO: A Window-Based Retransmission Timeout Ioannis Psaras Demokritos University of Thrace, Xanthi, Greece."

Similar presentations


Ads by Google