A proposal to extend SDN to the Data Plane

Position paper appearing at HotNets 2013

Instructions for reproducing results from the HotNets paper

Prerequisites

C++11 compiler, git

Steps

Download an allinone version of ns-2.35 from here
Run
```
 tar zxvf ns-allinone-2.35.tar.gz 
```
Run
```
 cd ns-allinone-2.35 
```
Run
```
 rm -rf ns-2.35 
```
to get rid of the original version of ns-2.35

Run

 git clone https://github.com/hbatmit/ns2.35 -b sdn-data-plane-hotnets13-reproduce ns-2.35

to checkout our modified version of ns-2.35

Run
```
 ./install 
```
to install ns-2.35 with all its components.

Run

 cd ns-2.35/tcl/ex/congctrl/result_scripts

Run
```
 ./generate_all_results.sh 
```

You should see the following output with results for all 6 tables:

     Table 1a: CoDel+FCFS,       average     throughput:  1.9955  Mbps  
     Table 1a: CoDel+FQ,         average     throughput:  1.8985  seconds  
     Table 1b: CoDel+FCFS,       bulk        throughput:  9.48102702703  Mbps
     Table 1b: CoDel+FCFS,       tail        FCT:         22.25  seconds
     Table 1b: CoDel+FQ,         bulk        throughput:  9.47583783784  Mbps
     Table 1b: CoDel+FQ,         tail        FCT:         18.705  seconds
     Table 2a: Bufferbloat+FQ,   bulk        throughput:  11.2208378378  Mbps
     Table 2a: Bufferbloat+FQ,   tail        FCT:         20.935  seconds
     Table 2a: CoDel+FQ,         average     throughput:  3.91908108108  Mbps
     Table 2a: CoDel+FQ,         tail        FCT:         43.72  seconds
     Table 2b: Bufferbloat+FQ,   interactive throughput:  7.469  Mbps
     Table 2b: Bufferbloat+FQ,   interactive RTT:         62240.4  ms
     Table 2b: CoDel+FQ,         interactive throughput:  6.5545  Mbps
     Table 2b: CoDel+FQ,         interactive RTT:         151.5  ms
     Table 3a: Bufferbloat+FQ,   bulk        throughput:  11.957  Mbps  
     Table 3a: CoDel+FQ,         bulk        throughput:  4.35  Mbps
     Table 3b: Bufferbloat+FQ,   interactive throughput:  11.957  Mbps
     Table 3b: Bufferbloat+FQ,   interactive RTT:         46103.0  ms
     Table 3b: CoDel+FQ,         interactive throughput:  4.35  Mbps
     Table 3b: CoDel+FQ,         interactive RTT:         158.2  ms

N.B. generate_all_results.sh can take upto 3 hours to finish on an
80-core machine. Reducing the number of simulation runs can speed
up this process.