We’re seeing lots of wishful thinking in the industry around being able to monetize QoS (Quality of Service) for consumer services. In part its coming from the current LTE (Long Term Evolution) sales push, which has an overly complex set of QoS capabilities given the limitations of the air interface. As well as the summer madness of the “Verizon and Google proposal,” an example of what happens when people in the know go on holiday so the corporate ‘BSiticians’ get a chance to play without supervision. But let’s focus away from the technology and political shenanigans, and instead focus on what really matters: the consumer and their experience with respect to QoS.
Customers pay for internet access, so if they have a 25/25Mbit/s or a 15/5Mbit/s or a 3/0.5Mbit/s access their web services simply work, e.g. YouTube, Hulu, BBC iPlayer, etc. They do not expect to pay twice, once for internet access and then again (either directly or indirectly) for additional ‘protection’. The following pitch is unsaleable: “Your service may not work some of the time, so through buying QoS it will likely work more of the time.” Quantification is challenging at best as it also depends upon the home network, over which most operators have little control.
Looking at the situation from the content provider’s perspective. Hulu buys a CDN (Content Delivery Network) service from Level 3, it is not going to then cut a number of additional access QoS deals across the myriad of connection options, e.g. FiOS, xDSL, private circuit, MetroEthernet, Verizon, AT&T, Cincinnati Bell, Comcast, etc.? From Level 3’s perspective they provide POP (Point of Presence) interconnect, so its up to the ISP to adequately support their customers to the advertised rates. Chargeable QoS across fixed broadband for consumer services is tough both from both customers’ expectations and also from a provider responsibility perspective as it encourages lots of finger pointing between the content owner, CDN provider, POP data center, ISP (core and metro), and the different access networks within the ISP.
Then onto LTE, this weblog has discussed previously how its a small step from HSPA (High Speed Packet Access) with respect to capacity, and the fundamental problem with radio transmission is Rayleigh fading. Over fiber / copper its Gaussian so its relatively easy to design so it appears up all the time. However with Rayleigh and real-world scenarios, e.g. edge of cell coverage or being in a building or its raining / misty the signal just goes away; we’ve all experienced it. So for mobile broadband the limitations of the air interface make any QoS guarantee quite difficult, even with LTE.
In practice we’ll only likely see with LTE traffic prioritization (differential QoS), which if you’re at the edge of a cell or holding your iPhone wrong means nothing anyway. A customer may be prepared to pay for a higher priority mobile data service, which will likely be bundled with an expensive data plan as there can be no service guarantees. For example, if I get high prioritization can the operator guarantee Hulu will work? They can not, hence the value to the customer of QoS is near zero. We lived through this with the ATM and IP debate 15 years ago, unless the difference is manifest and guaranteed for the customer it has little value. So selling QoS to consumers is not likely to be successful. Enterprise is another matter, there its about network management, which has a well understood value proposition to that customer segment.