Jose Saldana, Julian Fernandez-Navajas, Jose Ruiz-Mas "Can We Multiplex ACKs without Harming the Performance of TCP?," Consumer Communications and Networking Conference, CCNC 2014. Las Vegas, January 10, 2014, pp 921-922. ISBN 978-1-4799-2356-4.

1. CCNC 2014, The 11Th Annual IEEE Consumer Communications & Networking Conference
January 10-13 Las Vegas, Nevada USA
Can We Multiplex ACKs
without Harming the
Performance of TCP?
Jose Saldana, Julián Fernández-Navajas, José Ruiz-Mas

2. Index
1. Introduction
2. Tests and results
3. Conclusions
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014

3. Index
1. Introduction
2. Tests and results
3. Conclusions
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014

4. Introduction
 Increase of emerging real-time services
 They use small packets
 This is modifying the traffic mix present
on the Internet
 Inefficiency of the packets
 IPv6 makes the problem even worse
VoIP packets
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014

5. Introduction
 TCRTP (RFC4170) improves the efficiency
of VoIP. It uses three layers:
 Header compression
 Multiplexing
 Tunneling
IP network
.
.
.
MUX
RTP
DEMUX
RTP multiplexing
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014
RTP
.
.
.

6. Introduction
 Advantage: Bandwidth and pps savings
 At the cost of an additional multiplexing delay
Native VoIP
traffic
...
Inter-pkt time
Inter- pkt time
Inter-pkt time
...
...
...
...
...
Optimized
traffic
...
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014
...

7. Introduction
 TCM-TF*: Proposal for multiplexing other traffic
flows, including UDP (non-RTP) and TCP
payload
payload
RTP
payload
payload
RTP
UDP
TCP
UDP
UDP
IP
IP
IP
IP
ECRTP
Compression layer
No compr. / ROHC / IPHC / ECRTP
PPPMux
Multiplexing layer
PPPMux / Other
L2TP
Tunneling layer
GRE / L2TP
MPLS
IP
a) TCRTP
Network Protocol
IP
b) TCMTF
*draft-saldana-tsvwg-tcmtf-05
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014

8. Introduction
 TCP video traffic: 69% of all consumer Internet traffic
in 2017.
 When downloading a video, a computer may
generate some hundreds of ACKs per second, during
some tens of seconds.
 In some scenarios (e.g. the aggregation network of an
operator) high numbers of long-term flows of ACKs
share a common path.
 Header compression ratio of ACKs: from 40 to 7 or 8
bytes (savings of 80%).
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014

9. Introduction
 Is it a good idea to compress and multiplex these
flows?
 Would the multiplexing delay degrade the
performance of TCP?
 Sawtooth-shaped delay
PE
Added
delay
PE
time
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014

10. Index
1. Introduction
2. Tests and results
3. Conclusions
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014

11. Tests and Results
 Dumbbell scenario in ns2
 A sawtooth-shaped delay is added to the ACKs B-B’
A
O
ACK
FTP
N
B
P
ACK mux (PE)
A’
M
FTP
B’
 What is the effect? We use TCP Tahoe (the most
basic one) in order to more clearly see the effect
 First tests: separate A-A’ and B-B’
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014

12. Tests and Results
Bandwidth [Mbps]
Throughput (RTT = 80 ms)
10
8
6
4
2
0
30
35
40
45
50
55
60
55
60
simulated time [s]
Bandwidth [Mbps]
Throughput (RTT = 80 ms, mux period 50 ms)
10
8
6
4
2
0
30
35
40
45
50
simulated time [s]
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014

13. Tests and Results
Bandwidth [Mbps]
Throughput (RTT = 80 ms)
10
8
6
avg 9.24 Mbps
window reset
every ~7 sec
4
2
0
30
35
40
45
50
55
60
simulated time [s]
Bandwidth [Mbps]
ACKs arrive
Throughput (RTT = 80 ms, mux period 50 ms) in bursts
10
8
6
window reset
every ~9,5 sec
4
2
avg 8.04 Mbps
(12% reduction)
0
30
35
40
45
50
simulated time [s]
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014
55
60

14. Tests and Results
Window size
250
no PE
PE=50ms
Window size
200
150
100
50
0
0
10
20
30
40
50
60
simulation time [s]
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014
70
80

15. Tests and Results
Window evolution. One period
140
no PE
PE=50ms
120
Window size
100
Slow start
ends later
80
60
Window size
increases
more slowly
40
20
0
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014
The period
between
window
resets
increases

16. Tests and Results
 Second tests: A-A’ and B-B’ share the bottleneck
 Are multiplexed flows in clear disadvantage?
 We will use four different TCP variants:
 Tahoe
FTP
A
O ACK
A’
 Reno
N
M
 New Reno B
FTP
P ACK mux (PE)
B’
 SACK
 Results: Throughput difference between
multiplexed and non-multiplexed flows
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014

17. Tests and Results
 Results: Throughput difference between
multiplexed and non-multiplexed flows
Multiplexing Period PE [ms]
TCP
5
10
15
20
25
Tahoe
4.91 %
10.05 %
31.67 %
7.88 %
49.74 %
Reno
5.95 %
17.78 %
48.62 %
24.29 %
61.92 %
New
Reno
4.82 %
12.95 %
30.52 %
16.70 %
52.87 %
SACK
2.27 %
12.70 %
20.62 %
14.75 %
50.90 %
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014

18. Tests and Results
 Results: PE=5 ms
With Throughput difference between
the difference
multiplexed and non-multiplexed flows
is small
Multiplexing Period PE [ms]
TCP
5
10
15
20
25
Tahoe
4.91 %
10.05 %
31.67 %
7.88 %
49.74 %
Reno
5.95 %
17.78 %
48.62 %
24.29 %
61.92 %
New
Reno
4.82 %
12.95 %
30.52 %
16.70 %
52.87 %
SACK
2.27 %
12.70 %
20.62 %
14.75 %
50.90 %
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014

19. Tests and Results
 Results: Throughputms
With PE=10 difference between
the difference
multiplexed and non-multiplexed flows
becomes higher
Multiplexing Period PE [ms]
TCP
5
10
15
20
25
Tahoe
4.91 %
10.05 %
31.67 %
7.88 %
49.74 %
Reno
5.95 %
17.78 %
48.62 %
24.29 %
61.92 %
New
Reno
4.82 %
12.95 %
30.52 %
16.70 %
52.87 %
SACK
2.27 %
12.70 %
20.62 %
14.75 %
50.90 %
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014

20. Tests and Results
 Results: Throughput difference between
Below 10 ms the
difference may
multiplexed and non-multiplexed flows
become
unacceptable
Multiplexing Period PE [ms]
TCP
5
10
15
20
25
Tahoe
4.91 %
10.05 %
31.67 %
7.88 %
49.74 %
Reno
5.95 %
17.78 %
48.62 %
24.29 %
61.92 %
New
Reno
4.82 %
12.95 %
30.52 %
16.70 %
52.87 %
SACK
2.27 %
12.70 %
20.62 %
14.75 %
50.90 %
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014

21. Tests and Results
Throughput (SACK)
no PE
PE=5 ms
10
Throughput [Mbps]
8
6
4
2
0
900
910
920
930
940
950
960
970
980
Simulation time
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014
990
1000

22. Tests and Results
Throughput (Reno)
no PE
PE=25 ms
10
Throughput [Mbps]
8
6
4
2
0
900
910
920
930
940
950
960
970
980
Simulation time
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014
990
1000

23. Index
1. Introduction
2. Tests and results
3. Conclusions
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014

24. Conclusions
 Suitability of traffic optimization, based on header
compression and multiplexing, to the flows of ACKs
 The expected bandwidth savings are huge because of
the absence of payload
 Counterpart: throughput reduction when an
optimized flow shares a bottleneck with a nonoptimized one
 The impairments can be maintained in tolerable
limits, by setting an upper bound on the period
 Further study this trade-off between bandwidth
saving and throughput reduction
Can We Multiplex ACKs without Harming the Performance of TCP? - CCNC 2014

25. Thank you very much!
Jose Saldana, Julián Fernández-Navajas, José Ruiz-Mas