Telco future: To QoS or not to QoS, that is the question

One would think that with the overwhelming adoption of the Internet and IP communications in general they would have died out. Yet they still exist, bell-heads and net-heads, and nowhere is the distinction more clear than around the debate around QoS. If you are a bell-head you think that QoS means providing good quality for your VoIP service while trying to stifle your (free) competition. If you are a net-head you think that QoS is an evil plot of the inherently evil phone company to rob you of your bandwidth. Of course both of these extremes are wrong but these extremes provide a good, and hopefully amusing, demarcation of the field. The answer lies somewhere in the middle and is the thing that might save the telcos and provide their customers with what they want if only the telcos want to be saved.

What is Quality of Service (QoS)? Despite the polarized descriptions of QoS above, QoS is a rather poorly defined term. It describes what is a measure of the quality of the services one achieves. If you have to wait 2 hours in a fast food restaurant it is also a quality of the service, yet a very poor one. Things become interesting for real-world application if one makes QoS guarantees; “every customer served within 5 minutes of walking in the door or get your meal for free”, this will attract customers. This offer also provides an opportunity for differentiation, “every customer served within 10 mins, but at half the price”.

Now back to the Internet. Everyone who has ever used VoIP has found out that despite the total lack of QoS on the Internet, it tends to work. However it also often does not. If one tries to do a video call at any picture size over a postage stamp, one finds that it mostly does not work and sometimes does. Yes one may surf the web without any problem.

This behavior is because of traffic jams on the information superhighway, this results in too many packets holding the audio and video being lost or overly delayed. Why not add lanes on the Internet and avoid congestion? Like on the regular freeway, these upgrades happens often and invariably lead to more traffic than the added capacity can take. So adding capacilty to solve the congestion leads to more congestions. In politics they have not learned this lesson, regarding roads. We in telecommunication lands need a different solution going forward.

QoS control on the Internet will allow us ot break out of this vicious circle. It will allow the network to split the time-sensitive voice and video traffic from emails, web browsing, and peer 2 peer downloads.

Why is this useful? Because real-time streamed applications such as IP telephony and video calls are time-sensitive and can not handle loss all that well. You need practically dedicated bandwidth to make this work. If your media stream has to contend with other data streams on the Internet at a busy time, it will not work. This is why the voice quality of the traditional phone network is so good, you get a dedicated circuit for your call.

But dedicating a channel for a user is expensive. This is where the traditional phone companies missed the boat back in the 90ies is that things get much cheaper if you do not need a dedicated channel for every web request or email, in other words the Internet model. ATM failed, IP won.

So while all the telcos are now changing over to IP technology for their voice network, they may end up losing the baby with the bathwater. They need a way to ensure the quality to today’s, (voice) and tomorrows (video) services along with the best-effort Internet trafic.

The IETF have been working on the QoS issue for at least the past decade had have produces a lot of paper (DiffServ, RSVP etc.) but very little of it to any use for the internet user in general. Together with some others, I have been trying to standardize the technology to allow a user or a service provider to indicate the type of data to the network, in the MIDCOM group. That went nowhere fast because (1) at the time Cisco was dead against anyone talking like that to their routers, and in those that mattered a lot in the IETF (2) because the telcos had not figured out that this is where their future lay.

Today Cisco has other troubles (Juniper, HUAWEI) but the telcos still have not figured out how to get out of their self-dug hole. Now the telcos want to start blocking access to other people’s VoIP service and they want to address the load of P2P file sharing on their network. It has been repeated over and over; “You can not fight your customer”.

So what is the way out? The value of a broadband connection is in the kinds of (application) services one can get out of it. QoS will increase this value by allowing more applications to run well on the network. So if you run the broadband access network with QoS rather than blocking your customers from getting the value out of what they bought from you, you may give them more.

Before this can happen, telcos need a mind-shift;

• Step 1. Telco, stop feeling sorry for yourself that you do no longer sell all of your customers their voice service! Be happy they still buy their broadband from you.
• Step 2. Stop thinking that the only way you can recoup your investment in your access network is by tying your broadband into a vertical column with services. Others will do those services better and cheaper than you ever can, so you will lose by raising the stakes!
• Step 3. Capitalize on your assets. Your access network is the one asset competitors may not easily duplicate. Make sure that your customers get the most value out of it and are happy with it. Otherwise you are creating a fertile ground for a competitor. (municipal fiber anyone?)
• Step 4. Provide several kinds of service, basic best-effort, streaming media, interactive streaming media and provide QoS assurances and accompanying prices.
• Step 5. Next allow your users and their service providers to request chunks of this service (e.g. a session of interactive streaming media for a low latency&low loss stream, in chunks of 80 bytes once every 20ms between these two addresses.)

When the price is right, consumers and on-line service providers will like it. I do not guarantee double digit growth. But I do guarantee a steady revenue stream for anyone who operates broadband like this. Income like a utility, and hence you can borrow money at a low rate like a utility.

This ties in nicely with the recent attention to the Utilicom view as post-telecom. Where the parallel breaks is that with the water and power utilities they do not care (all that much) how I use the power or the water as long as I do not use more than my connection is contracted for. Now with a QoS broadband utility one has more than one grade of service over the same line.

This is not a sexy proposition that would appeal to marketers and similar people looking for get-rich-quick schemes. However a steady path upward seems like a great step forward for a former incumbent telco in steady decline. The question is of course do today’s telco management want to be rescued?

PS: I am still sitting on the technology to deliver QoS cheaply, I would use it if telcos would buy infrastructure gear from a small company from the Netherlands.