Re: [tcmtf] Answers to possible questions in the BOF

[Mirko Sužnjević <[e-mail]>]

Hello all,
Here i have just a small addition:
By using TCM-TF we aim maintain the QoE of the service, or at least minimize the degradation.

1)      Recommended limits of multiplexing period are given in order not do significantly degrade the service (on applicable applications and studies we aimed at MOS over 4).

2)      1st example of a dynamic scenarios TCMTF is actually reducing latency by preventing congestion (rush hour scenario)

2)      2nd example of the launch of the game scenario TCM-TF provides significant savings, and increase of the QoE as service unavailability is much much more QoE degrading for gamers than slight latency increase
Also i would emphasize that the added delay is very minimal on links with high packet rate, and that the maximal introduced delay can be controlled with the multiplexing period depending on how much tradeoff is needed or allowed.
Cheers!
Mirko
From: [e-mail] [mailto:[e-mail]] On Behalf Of Jose Saldana
Sent: Tuesday, June 25, 2013 10:27 AM
To: [e-mail]
Subject: Re: [tcmtf] Answers to possible questions in the BOF

Question 3: You are talking about real-time services (VoIP, online games, remote desktop). *Why adding a new (multiplexing) delay?* Would it harm the user’s experience with the service?

Answer

1) We have had that question in mind from the very beginning. We want to save bandwidth and pps, but always preserving subjective quality. In fact, a specific draft has been written including a set of recommendations of the maximum delays that can be tolerated by users of different services. The draft includes references to studies where subjective quality has been evaluated with real people.


2) In addition, our idea is that TCM-TF can be something to be used dynamically when required. Some examples:

- I am in the rush hour and I notice a lot of traffic in my network. In order to avoid the collapse, I optimize traffic. I add a small delay, but the bandwidth savings can make it possible that the service still works.

- I am an online games provider and I am releasing a new title. The first days, a lot of people will want to play. So, how can I dimension my network? Should I dimension it for the worst case?

TCM-TF provides flexibility: there is a tradeoff: a small additional delay as the price for maintaining the service. It should be taken into account that for certain services the bandwidth savings are 50 or even 60%.


3) If I have a high number of flows, I can achieve significant bandwidth savings while adding small delays. Some examples:
- 10 VoIP flows G729 can be optimized including one packet from each flow (average additional delay 10ms) and 50% of bandwidth can be saved
- if I have 100 flows of a TCP-based online game, I can achieve 45% bandwidth saving with a multiplexing period of 30 ms (average additional delay 15ms)


4) TCM-TF is not only for real-time services. Other small-packet services can be optimized, and in this case the additional delay would not be significant.


Any other thoughts?

Jose

De: [e-mail]<mailto:[e-mail]> [mailto:[e-mail]] En nombre de Jose Saldana
Enviado el: lunes, 24 de junio de 2013 13:00
Para: [e-mail]<mailto:[e-mail]>
Asunto: [tcmtf] Answers to possible questions in the BOF

I would like to start a thread about possible questions people may ask in the BOF. Obviously, we also need answers, so we should cooperate.

This is different from the “questions to ask in the BOF”. This will be discussed separately.

Thanks!

Jose