Lab 2: diffserv simulation

1. Set up below network topology. E1 and E2 are edge routers and C is core router. Link C-E2 has bandwidth 2Mb and 10ms delay; other links have have bandwidth 5Mb and 5ms delay. Use the RED parameters used in lecture to config diffserv queues.

   			
        S1
         \
          \
           E1------C------E2------D
          /
         /
        S2
   
   

2. Set up UDP connection from node S1 to node D, add CBR traffic (packet size: 1000B, rate: 2Mbps).
3. Add token bucket policy for the above CBR traffic (cir=1000000bps, cbs=5000B, codepoints: 00 and 01)
4. Enable packet statistics (for links E1-C and C-E2) and run simulation. What do you observe? Is the stats what you expect?
5. Add another CBR traffic from node S2 to node D (packet size: 1000B, rate: 3Mbps), and corresponding token bucket policy for it (cir=1000000bps, cbs=5000B, codepoints: 10 and 11). You should use two physical queues on link C-E2 so that packets from S1 are put into physical queue 0.
6. Rerun simulation. What do you observe? Is the stats what you expect?
7. Add priority scheduling between two physical queues on link C-E2. Configure physical queue 0 with higher priority (max rate: 2Mbps)
8. Rerun simulation. What do you observe? Is the stats what you expect?
9. If you are required to give "IN profile" packets higher priority, how should change above configurations?

1. Try out other scheduling schemes such as WRR.

References:

• The main ns page
• The tcl documentation page
• The otcl tutorial page