Throughput comparison of one TCP stream with and without ECN

 

• Equipment:
  • Pool of Linux host, v. 2.9.4: ECN and SACK capable
  • One IBM workstation, AIX v. 5.1: ECN and SACK capable (Amsterdam)
  • chain of intermediate ecn-capable routers (Cisco 7500 and C7200), IOS: 12.1(4)E (experimental version): in this test only the first router is ECN capable.

  Test description:

   

• Topology:
  topology consists in a chain of ecn-capable routers connecting a pool of TCP senders (on the left of the chain) and a pool of TCP receivers (on the right), like in this map
  Bandwidth bottleneck: 50 Mbps

   

• Parameters
  • min threshold constant =100 packets
  • max threshold constant = 300 packets
  • drop/mark probability : [1/50,1/5,1/2]
  • test duration: 120 sec

   

• Traffic profile
  • 20  TCP streams generated by Iperf:

  • one reference TCP stream between a Linux host in Italy and AIX workstation: ECN e SACK;
  • 19 background TCP streams between two Linux host: ECN e SACK;

Summary:

• with ECN, we can see a significant increase throughput for every mark probability value. 

Comments:

• Observing figure 1 we can notice that throughput with ECN enable is always higher than throughput without ECN  for every mark probability value. The larger gain is with a very low mark probability (0.02), and when it grows the gain with ECN is reduced.

Figure 1: Throughput comparison of one TCP stream, with and without ECN

Serena Alessandrini and Tiziana Ferrari, August 13 2002