Video and Audio Analysis through offline captures
Introduction
As a network engineer/administrator you are sometimes exposed to quality issues with your networks, like intermittent voice calls or freezing video. These kinds of problems are not always related to packet loss or network congestion and cannot be identified with the traditional tools like ping or bandwidth usage.
A complete network analysis is a complex task and requires the use of expensive tools, intrusive methods, and it takes time; even if this method has proved to be effective is not always the best option for small business or urgent situations.
The objective is to accomplish an "easy" method to perform this analysis; what steps needs to be followed; what information is the most important; and how to interpret the results. A collateral outcome will be the possible use of this method to capture traffic for spying purposes.
Tools like Wireshark to capture the traffic, and audio/video players to reproduce the data will be used. Layer 3 and layer 4 data will be analyzed to identify relevant values. There are several documents related to the analysis of video and audio traffic, as well as how to configure the different tools such Wireshark, audio and video players, and the different brands of switches, however, there is not a single document that explains the whole process.
Scenarios
I.
Relevant tools were installed to perform the analysis. Some of the tools used are: Microsoft Visio, WireShark, Oracle VM VirtualBox, and Kali Linux.
II.
The following application topology will be implement to analyze voice and video traffic.
III.
Network design and switch configuration is an important step to carry out the capture of the data and the analysis of it. The first figure shows a common network topology and the second one represents how the mirroring should take place.
Logbook
The following diagrams present some scenarios were the use of the common parameters such as delay, bandwidth, jitter and loss was not effective to identify and solve the problem:
- Scenario #1: A customer change ISP for administrative reasons, the new ISP provides the same link characteristics as the previous ISP, however, the customer complains about video performance with youtube.
- Scenario #2: A corporate customer bought a video conference solution to be placed in the main offices, once the equipment was installed the performance is deficient even when the bandwidth was higher than required.
- Scenario #3: A TV cable provider needs a solution to convert from 3x1G interfaces to a 1x10G interface, to pass TV channels over IP. An EoMPLS solution was implemented, but the customer complains of blurry channels randomly.
What would you do?
IV.
During the test, two type of scenarios where discovered. The first scenario was for the Campus network, where the connection was performed directly between the PC and the video server.
The second scenario was similar in the residence and free networks.
IV.
During the test, two type of scenarios where discovered. The first scenario was for the Campus network, where the connection was performed directly between the PC and the video server.
The second scenario was similar in the residence and free networks.
V.
Port mirroring receives different names according to the brand of the switch where it is going to be implemented, however, the functionality is similar across different manufacturers.
The following image presents relevant information on the most used brands of switches; our work will be focused on the top three.
V.
Port mirroring receives different names according to the brand of the switch where it is going to be implemented, however, the functionality is similar across different manufacturers.
The following image presents relevant information on the most used brands of switches; our work will be focused on the top three.
VI.
Video analysis was performed using Vimeo as the content provider. The list of videos tested and the characteristics are presented in the following table.
VII.
The following graph shows the different behavior observed for the videos tested, no benefit was observed from using one or another service provider, and the use of cache server from Telus and Shaw dindn't improve the quality of the service.
VIII.
Another interesting feature available to troubleshoot video issues, is the debug feature available on the different video providers, where you can identify bandwidth problems, packets dropped and video specifications.
IX.
Attached are the documents used during the poster presentation, the first one contains the slides used on the poster and the second the video used to provide additional information.
X.
Conclusions:
-
The measure that the applications(Vimeo) does at the beginning of the playback determines the recommended quality setting to play the video.
-
The initial bandwidth determination does not necessarily match the real bandwidth capacity of the channel.
-
The on – off cycles used to transmit the data are critical to make efficient use of the channel.
-
The initial buffering phase seems to be the most important part to ensure a wonderful experience to the user, it seems better to wait a few seconds that experience freezing video during playback.
-
Every video, every location and every moment could represent a different behavior on the video playback performance.
-
Wireshark is a powerful tool; however, the increasing use of encryption limits the capabilities of this tool to recreate the traffic received by the user.
-
Wireshark also provides an important way to determine how is the traffic flow and who are the real participant on the conversation.