Proposing a Mobile Index to Measure Networks' and Apps' "Capability"Proposing a Mobile Index to Measure Networks' and Apps' "Capability"

With a standard methodology, networks can be evaluated and then applications requirements can be used to define whether acceptable performance is achieved.

With Enterprise Connect coming up in a week or so, I thought it would be a good time to start a discussion about the concept of a Mobile Index. The Mobile Index is a quantitative measure of the availability and capability of mobile device connectivity. The point of a Mobile Index is to begin to think about how bandwidth of the future will enable us to change both business and organizational behavior and how we develop and deploy applications.

As organizations look to deploy BYOD and their user devices are now on a plethora of networks, having some mechanism to define what the "requirements" are to support a specific application or class of applications could be very valuable; enterprises could then evaluate specific network capabilities against that requirement.

It would seem there are two measures of the "goodness" of mobile access: bandwidth and availability. Bandwidth can be characterized by both how much is available as a mean, and the variation distribution from the mean. In other words, if the mean bandwidth is 10 Mbps per second and the 3 Sigma standard variation is 5 Mbps, that means that the average user is getting more than 5 Mbps about 99.8% of the time. The combination of mean and distribution allows for defining the suitability of the bandwidth for a specific application.

Availability can be characterized by the percentage of time that the bandwidth is actually available at all, and can be characterized by either a percentage or by using the more telecom-centric measure of "nines", where five 9s is an unavailability of .001% or 5 minutes a year. I have not included latency, as that is more of a binary: It is either low enough or not.

If we assign numeric ranges to both of these factors, then the multiplicative result can be seen as a combined measure of the relative suitability of mobility for a class of applications. For this purpose I would propose a scale of 1-10 for each area.

For bandwidth, let's take the example of HD video transmission: For general use, traditional HD displays at 2K are common, so a sustained maximum bandwidth of about 16 Mbps for stereo (3D) transmission of HD video is probably reasonable. This could become the "10". To make the lower level changes more dramatic, an exponential scale could be used for the bandwidth. Figure 1 shows a model, with 0 being no bandwidth and 10 being 16 Mbps, with an exponential map in between. Shown for reference are the relative values of a 1 Mbps 4G connection as 2.5 and an 8 Mbps WiFi connection as 8.5.

For availability, we could use the same type of model, but use the well known "nines". If we make the ten rating as 99.9999 or six nines of availability (the 99.9999% availability is .0001% unavailability or about 30 seconds per year of average unavailability, a goal that is certainly sufficient for almost any operation); and if we make 1 a single 9 of availability (10% unavailability means that more than one month per year, the bandwidth is unavailable). We end up with a similar mapping, with the index being as shown in Figure 2. In this case the model is linear, as that applies well to the actual outcomes. The traditional five 9s of telephony is shown also shown in the figure for reference, with the resulting 8.5 value in the calculated index.

The reason for going to six 9s is the realization that some applications such as health care and financial actually have more stringent availability requirements (someone can die in the 5 minutes of unavailability of five 9s).

The result is that, by multiplying the Bandwidth times the Availability, a measure on a 0-100 scale is generated, where 100 is capable of supporting virtually any form of available endpoint today, with essentially guaranteed availability. So, if my connectivity is as shown in Figure 3, my network is three 9s of availability and I can sustain 2 Mbps of bandwidth. This yields a Mobility Index figure of 4.3 for Bandwidth and 5 for Availability, or an overall Mobility Index of 21.5. For many web-based applications, this is quite sufficient, though it may be appropriate to specify one or the other of the base factors as required, depending on the type of application.

It seems to me that having some common form of defining the bandwidth and availability for mobile devices would be very helpful in managing the emerging world of BYOD and network complexity. By having a standard methodology, networks can be evaluated and then applications requirements can be used to define whether acceptable business performance can be achieved.