Page 154 - ITU Kaleidoscope 2016
P. 154
transfer throughput results have not been used for After completely received the video file in the
calculating file transfer duration results. middle-box VM, the middle-box VM serves as a VLC
In order to calculate the file transfer duration streaming server for CHULA OpenStack VM. This
results, the average actual file transfer rate results as experiment is in order to evaluate the performance of
shown in Table 8 and Figure 5 have been used. According downloaded video file and to stream out smoothly within
to the actual file transfer results, the highest actual file CHULA SmartX box network. The Big Buck Bunny H.
transfer rate achieves in the case of using 400Mbps target 264 video codec with video and audio mean bit rate of total
file rate with multi-path scenario which combines the 11133kbits/sec with resolution 3840x2610 and duration of
bandwidth of two paths periodically. In the cases of 100, 10-minute as shown in Table 9 has been used for
200, 300Mbps, we recommend to use only Path 2 (GIST- streaming from the middle- box VLC server to the video
B>CHULA) in order to reduce the cost of multi-national client in CHULA OpenStack VM. The VLC player from
network links. Comparing to the file transfer rates of the the video server has been configured to stream out video by
cases 100, 200, 300Mbps and the cases 400, 500Mpbs, the using RTP mode. The VLC GUI interface has been used in
single-path file transferring function outperforms the order to stream out with original video tracks. The video
multi-path function in the former cases due to the packets are captured at eth1.111 of the middle-box VM
maximum file transfer limit of Tsunami application. and eth1 of CHULA OpenStack VM by using tcpdump
However, multi-path function is recommended to use in packet analyzer. To investigate the performance of video
order to obtain the highest file transfer rate when the streaming, we have run three times for the same video
single-path alone is not enough to carry out the whole streaming session. For the detailed dynamics of packet
video file traffic. We can see that trend from the case of generation arriving at the video server and the received
target file transfer rate 400, 500Mbps. packet stream departing at the video client, we include
here the arrival and departure curves in Figure 6. The
Table 8: Results of average actual file transfer rate number of packets shown in the graph is the number of
over OF@TEIN packets transmitted during video streaming. The number of
Actual File Transfer Rate packet losses is zero in all tests. This curve confirms that
Testing Scenarios
(Mbps) 4K RTP video streaming within CHULA SmartX box
Target Transfer Rate (Mbps) 100 200 300 400 500 network can be streamed out without any significant extra
Multi-path(1:2sec) 93.4 187.3 264.8 288.2 255.7 delay.
Single-path(GIST-B>CHULA) 94.9 188.8 271.4 224.2 193.9
Single-path(GIST-B>MY 91.2 134.5 268.7 169.8 207.4 x 10 4K RTP Video Streaming within CHULA SmartX Box Network
5
>CHULA) 7
6
5
Number of Packets 4 3
2
1 No. Packets at Video Server
No. Packets at Video Client
0
0 100 200 300 400 500 600 700
Time (s)
Figure 6: 4K RTP video streaming within CHULA SmartX box
network.
4. CONCLUSION AND FUTURE WORK
Figure 5: Actual file transfer rate over OF@TEIN SDN cloud
playground. In this multi-path chunked video exchanges
experiments, we have tested the combination of traditional
Table 9: Video streaming experimental parameters
Video Streaming and Transcoding Detailed Tsunami protocol and proposed multi-path file-transfer
Parameters function. In addition, we have implemented the X11
Video Format H. 264 desktop environment and access method for OpenStack
Duration 10 minutes VMs in order to use the GUI applications with fast access.
Total File Size 843Mbytes According to the tested results with three scenarios: multi-
Resolution 3840 x 2610 path (1:2sec), using Path 1 alone (GIST-B>MY>CHULA)
Audio and Video Mean Bit Rate 11133kbits/sec and using Path 2 alone (GIST-B>CHULA), the proposed
multi-path file-transfer method can transfer with the
– 136 –