Presentation is loading. Please wait.

Presentation is loading. Please wait.

On Efficient On-line Grouping of Flows with Shared Bottlenecks at Loaded Servers by O. Younis and S. Fahmy Department of Computer Sciences, Purdue University.

Similar presentations


Presentation on theme: "On Efficient On-line Grouping of Flows with Shared Bottlenecks at Loaded Servers by O. Younis and S. Fahmy Department of Computer Sciences, Purdue University."— Presentation transcript:

1 On Efficient On-line Grouping of Flows with Shared Bottlenecks at Loaded Servers by O. Younis and S. Fahmy Department of Computer Sciences, Purdue University Presented by Felix Lam

2 Content Introduction System Design Performance Evaluation Conclusions Comments and Discussions

3 Introduction Bottleneck Sharing is common among different connections Ordinary TCP is simple but NOT efficient. Internet FTP/ Web/ Streaming Server Clients

4 Introduction Coordinated Congestion Management  Flows sharing the same bottleneck(s) are grouped together for congestion control  Each flow in a group is given different quality of service by the server (e.g. different sending rates, different error protection… etc.)  Collective performance improved

5 Introduction How to group flows that share the same bottleneck(s) ? FlowMate  Make use of packet delay information to group flows sharing the same bottleneck(s)

6 Basic Architecture The FlowMate module is embedded into the TCP implementation of the sender to collect delay samples for correlation test.

7 Delay Computation Packet delay information can be collected in two ways  Timestamping ACKs Option fields supported in TCP implementations in most operating systems. TS Value – time at which the data packet / ACK is sent TS Echo Reply – the previously received TS Value, only present in ACK. TS Recv. Time – extension proposed by the paper Clock-skewness does not affect the correlation values as long as it is constant. Delay = TS Value (Sender) – TS Recv. Time (Receiver)

8 Delay Computation  Round Trip Time (RTT) samples For each ACK received, TCP sender get a RTT sample. However, using RTT samples instead of forward delay may degrade grouping accuracy because:  Bottlenecks of reverse path and forward path may be different  Delayed acknowledgement affect RTT The reduction in grouping accuracy < 5%

9 Correlation Test Correlation test is performed after each flow has got N (e.g. N=6) delay samples Apply Pearson’s correlation function on the delay samples Packet Sent Time 1s2s3s4s5s6s8s11s12s15s x1x1 x2x2 x3x3 x5x5 x6x6 y1y1 y2y2 y3y3 y4y4 y5y5 x4x4 y6y6 7s14s

10 Correlation Test Cross-measure M x  Select delay samples from the two flows that a packet of flow x must precede a packet of flow y immediately  The correlation of the 3 consecutive pairs of delay samples is computed Packet Sent Time 1s2s8s12s x1x1 x5x5 y1y1 y4y4 x4x4 y6y6 7s14s

11 Correlation Test Auto-measure M a  Select pairs of delay samples from one of the two flows that the spacing between each pair is larger than the average spacing between the pairs used to compute M x (e.g. 1.33) Packet Sent Time 1s5s6s12s x1x1 x2x2 x3x3 x5x5

12 Correlation Test If Mx > Ma Then the two flows are grouped together (i.e. they share the same bottleneck(s))

13 Partitioning If more than one groups succeed in the test, join the one with highest M x Each flow will be repartitioned after it gets enough new delay samples G1G1 G2G2 Representative Flow A new flow Correlation tests G3G3 form a new group if fail

14 Performance Evaluation k j is the number of splits of the correct group j. correct: {1,2,3},{4,5,6} result: {1,2},{3,4,5},{6} Accuracy Index =0.67                    flowsTotal k flowsTotal flowssharedfalse IndexAccuracy Groups j j # )1( # # 1 # 1     

15 Performance Evaluation

16 Accuracy Index

17 Performance Evaluation Effect of Buffer Size

18 Performance Evaluation Effect of packet drop policy

19 Performance Evaluation Effect of background traffic load

20 Performance Evaluation Performance degradation with bursty Telnet traffic

21 Conclusions FlowMate is an on-line partitioning algorithm, requires no active-probing Robust under heavy background traffic, and different router buffer sizes High burstiness degrades the performance Useful for many applications that require inter-flow coordination

22 Comments and Discussions The high accuracy, robustness under high background traffic load makes FlowMate quite useful in practical network environments. Although burstiness degrades performance, it does not affect its usefulness in the scenario of TCP video streaming.

23 Comments and Discussions Wonder if varying background traffic load will affect the performance Any better way to choose the representative flow in each group?


Download ppt "On Efficient On-line Grouping of Flows with Shared Bottlenecks at Loaded Servers by O. Younis and S. Fahmy Department of Computer Sciences, Purdue University."

Similar presentations


Ads by Google