Do-it-yourself

Posted on Friday, January 18, 2019

Ken Rehbehn has started a consulting practice on what is known in the business as ‘Critical Communications’ which normally is thought of as a public safety technology. But he has expanded into ‘Business Critical Communications’ and he got me interested in the subtopic of Private Mobile Broadband Networks. Hence the title.

You can see Ken’s website here which is also linked on the Macondo Telecom and Mirador Communications websites (shameless exploiting the opportunity to direct you to my other sites).

So what is a Private Mobile Broadband Network?

As Ken writes…

Wireless technology takes many forms. Most consumers are very familiar with mobile cellular service offered by large-scale mobile network operators who purchased spectrum licenses and built cell sites across a national geographic footprint. Likewise, consumers are also quite used to license-exempt Wi-Fi technology that delivers broadband data in a home or office. In contrast to a public mobile network that is owned and operated by a large service provider, Wi-Fi access points are by-and-large privately owned and operated by consumers and enterprises. With Private Mobile Broadband in the form of Private LTE or Private 5G, elements of these two approaches come together to create a wireless connectivity fabric with profound implications for industrial communications.

This is a rich area for study and we are just beginning to construct qualitative and quantitative models to understand it. We think our first report will run to tens of thousands of words so a bit beyond my usual standards for a weekly blog (750 to 2,000 words).

Instead, this week, I want to look at the general question of when enterprise customers should ‘Do-it-themselves’ versus take a Managed Services offering from a telecom operator. When I need to refer to a specific example, I will use a Private Mobile Broadband Network (P-MBB).

Enterprise networks

I realized that many readers may be more familiar with consumer telecom networks which are always run by licensed Mobile Network Operators (MNOs) so I thought I give my definition of an enterprise network.

To me, an enterprise network is a set of telecom resources dedicated to internal corporate equipment and end users. The enterprise can specify the quality of service (QoS) parameters so that performance can be optimized for its users’ specific applications.

There are two key phrases in this definition:

Dedicated, although the dedication may be virtual. Just as enterprises are shifting their servers to the cloud but still view the virtual machines as dedicated to their use, an enterprise network may be constructed with shared components but still dedicated to a single customer. Virtual Private Networks and their more advanced offspring, Software-Defined Networks are examples of dedicated networks running on shared infrastructure.

Specify the quality of service parameters so the network can be optimized for their specific applications. Unlike the public Internet or (ancient) telco voice and data networks, enterprise applications are not ‘one size fits all’ nor are enterprises usually satisfied with best-efforts quality of service. Instead, enterprise clients normally demand the ability to observe QoS in real time and adjust their network’s characteristics to achieve their objectives.

These concepts work together: for a client to have the right to adjust network performance parameters, the network would have to be dedicated to a single ‘user’.

It is easiest to think of these goals being accomplished in a standalone, DIY private network: the client owns the boxes and runs the show; the network is dedicated to them and they can fiddle with the knobs any way that they like.

But like cloud providers, telecom operators can instantiate a dedicated network virtually in shared infrastructure and give their clients control over the performance of their ‘own’ resources, although, admittedly, this is more common today in fixed networks rather than wireless ones. Network slicing – which is possible but not easy to do today in LTE – will be a core feature of later releases of 5G and so this limitation will change in the near future. Each dynamically provisioned slice in 5G could have its own unique QoS and could be dedicated to a particular enterprise client who could ‘fiddle with the knobs’ as they like without risk of harming other users.

DIY or Managed Service

If a telecom operator could slice their fixed or wireless network easily, it would seem to be a no-brainer that all enterprises would ‘rent’ just the capacity they need. That is what a Managed Service is: a way for a client to get the communications they need, paying-as-they-go rather than owning the equipment and carrying the staff on their payroll that are needed to run the network.

This next sentence will perhaps be too technical but telecom networks are generally characterized by increasing returns to scale which means the more traffic they carry, the lower the cost of carrying the next unit. Transmitting one more second of voice, one more frame of a movie, one more data packet is virtually costless if the network is already in place. Furthermore, the components that make up a network come in fixed sizes. It has to be cheaper for an enterprise to share a telecom operator’s network with other users rather than build from scratch the smallest network that can carry their traffic.

(Assuming of course that the sharing is done in such a way that the client feels they have their ‘own’ network and they can adjust QoS parameters.)

A client may also decide to contract a Managed Service for scope economy reasons. Networks are complex beasts and their management and administration may require skills or knowledge which the client does not have. In such a case, while there may be no significant cost-advantage with sharing hardware or software components, there is with sharing the human or machine skills required to run it. For the P-MBB example, the skills to run a wireless communications network are not that common nor is there much overlap with those required to run, say, a fixed LAN. In such cases, even if a dedicated resource is required, it may be more effective for an MNO to do the operating, given that they have a broader pool of such resources and established procedures to acquire, train and retain them.

Many see the economics this way, but many others do not.

Perhaps the most common non-economic reason to DIY is to have a critical component of a solution under the client’s direct control. The value of control is perceived to be higher than any cost benefit foregone by not sharing a component that could be multiuse or the cost of hiring scarce skills that might have been provided by a telco. An obvious example of this is when a telecom operator’s workers are unionized. Clients may perceive that their mission critical applications are at risk of being affected by strikes or slowdowns that have nothing to do with them.

In Industry 4.0 applications – which we see as the driver for most P-MBB networks – the devices on a client’s network may be sensors, servos and other modules interacting with their operations technology (OT) components such as autonomous mining equipment, autonomous container cranes or smart factory machines. The OT vendor may insist that the communications network be isolated, even perhaps under that vendor’s control.

Security is another consideration that might be considered a subtopic of control but one that warrants a separate discussion. Mission critical operations need above well average levels of end-to-end reliability, almost by definition. Components at a high risk of compromise require high levels of security. Clients may perceive – rightly or wrongly – that a third-party will not maintain the same level of secure access that they would if they did it themselves.

In our research we are finding that a P-MBB has many subcomponents and the arguments about control, security, economies of scale and economies of scope may be different for each. There may be regulatory restrictions (e.g. in the use of spectrum) that may require an MNO to be involved no matter what the client would prefer from a control or security aspect.

Rent versus Buy

In some sense the decision to go Managed Services or DIY for an enterprise private network is like renting an apartment versus building your own house. A Managed Services solution the renting-an-apartment alternative: everything is done for you, it may be cheaper short-term and you may find it easier to finance a monthly fee rather than an upfront purchase, but your scope for customization and control over the environment you live in is limited.

But DIY should not be though of as buying an empty lot, going to the lumber yard, ordering a bunch of raw materials and then building everything with the equipment you have on your tool belt.

Rather DIY for P-MBB is more like acting as your own general contractor: for some stuff you will hire a subcontractor; for other stuff, you will have to hire licensed tradespersons because regulation requires it (like plumbers or electricians), and some stuff you will really do yourself (hammer or paintbrush in hand).

DIY is always the way you wanted it and you have absolute control over the outcome but sometimes it can cost more, and the execution can be a little rocky.

Title Reference: Frankly I did not plan on a rock music reference; it was just the most obvious title, a term we are using in our research. But doing a quick search on Spotify, I ran across an album of the same name by the long-forgotten Punk group, Ian Dury and the Blockheads. I owned the LP at one time (or had it on cassette tape maybe) and I remember (fondly) many of the songs even though I was more of a ‘New Wave’ fan in the late 70s / early 80s. Unfortunately, Spotify has decided that the album is Explicit (that was Punk after all) and it is perhaps better for the preservation of my family-friendly-rating not to develop this note any further.

 

No Comments »

Leave a Reply

Your email address will not be published.

© 2022 Macondo Telecom