-
Notifications
You must be signed in to change notification settings - Fork 1
Results and Further
Prajval M edited this page Dec 10, 2018
·
12 revisions
Comparison for Fq-PIE and Fq-CoDel has been done in ns3 using a dumbbell topology with bottleneck as follows:
100mB/s, 5ms |
n0--------------|TCP
|
|
100mB/s, 5ms |
n1--------------|TCP
|
|
100Mb/s, 5ms |TCP 100Mb/s, 5ms
n2--------------| |---------------n9 TCP sink
| 10Mbps, 32ms |
n7------------------n8
100Mb/s, 5ms | QueueLimit = 100 |
| |
| |
n3--------------|TCP |---------------n10 UDP sink
| 100Mb/s, 5ms
|
100mB/s, 5ms |TCP
n4--------------|
|
|
100mB/s, 5ms |UDP
n5--------------|
|
|
100mB/s, 5ms |UDP
n6--------------|
All the n0, n1, n2, n3, n4, n5 with TCP sockets and n6, n7 with UDP sockets installed. n9 acts as a TCP sink and n10 as UDP. All measurements for throughput was measured using n8-n9 and n8-n10 links.
-
Test Suite:
- PASS : 251
- SKIP : 3
- FAIL : 0
- CRASHED : 0
- Valgrind Errors : 0
- Fq-PIE, Fq-CoDel, PIE, CoDel Comparison
( Almost all the parameters were kept similar to get a close enough comparison)
Output | Fq-PIE | Fq-CoDel | PIE | CoDel |
---|---|---|---|---|
Drops | 68197 | 84424 | 82671 | 82637 |
TCP (Mbps) | 4.01 | 4.36 | 0.52 | 0.48 |
TCP (Packets / s) | 500.35 | 543.57 | 64.87 | 60.48 |
UDP (Mbps) | 5.51 | 5.34 | 4.38 | 4.42 |
UDP (Packets / s) | 667.88 | 648.53 | 531.73 | 536.93 |
- Fq-PIE and Fq-CoDel have comparable throughputs
- PIE and CoDel are not fair when both TCP and UDP packets are present
- Fq-PIE is however better than Fq-CoDel in case of preventing buffer bloat as it reduces the number of drops significantly compared to Fq-CoDel.
- CAKE implementation and comparison with FqPIE
- Only Basic FQ-PIE has been implemented. A enhanced version can be used for better results. Enhanced PIE