Analysts are not employed by SDxCentral and the views, thoughts, and opinions expressed in their content belong solely to the author and do not reflect the views of SDxCentral. Note: AvidThink is a separate organization, created by Roy Chua, that is not affiliated with SDxCentral.
Last week’s blog post on containers and telco elicited a much larger than usual number of responses from readers – particularly those at telcos. Many privately reflected that they wished their companies were more nimble and did a better job of mimicking successful cloud companies like Amazon, Google, and Microsoft.
To be fair to telcos, there are many who innovate, plus building out large wireless and wireline networks is not an easy task. I’m sure we’ll see some breakout examples of telco innovation next week at MWC Barcelona. I realize many of you are busy preparing for your trek to Barcelona, Spain, and so I thought I would write a follow-on to last week’s post that, hopefully, makes us all think a little differently as we head out.
Conway’s Law and the Mirroring Hypothesis
We’ll start with Conway’s law. The more experienced of us might recognize the adage, attributed to programmer Melvin Conway, who, circa 1967, stated that “organizations which design systems … are constrained to produce designs which are copies of the communication structures of these organizations.” As an example, Eric S. Raymond of The Cathedral and Bazaar fame, made the statement, “If you have four groups working on a compiler, you’ll get a 4-pass compiler.” Or as Raymond succinctly put it, “Process becomes product.”
Evidence supporting Conway’s law, now given the fancy term “the mirroring hypothesis” – see, it’s not just telcos that use fancy names – has been found by numerous business school researchers from the likes of MIT and Harvard. If you do a search for “Henderson and Clark,” “von Hippel,” “Chesbrough,” or “Teece,” you’ll find a host of research papers from these academic researchers on organizations and product innovation and how products mirror organizations and process.
We’ll examine next how mirroring could be relevant to telcos.
NFV, SDN, and Disaggregation
Long-time readers will recognize that many of the key concepts underlying NFV and SDN stem from disaggregation. In NFV this includes breaking down historically integrated proprietary products into components and using commodity platforms to help lower the cost of the overall solution. While in SDN, separating the control plane from the data plane provides greater flexibility and innovation. The white box movement saw the use of merchant silicon to reduce the cost of expensive switches and routers. This drove value into the network operating system and networking applications on top of commodity silicon.
Organizationally, with NFV and SDN we’re seeing moves within historically integrated network equipment providers (NEP) to break up the product lines into platforms and applications. The larger ecosystem is doing the same as we organize around different companies and initiatives focused on separate parts of a disaggregated platform. For example, networking OS and switching/routing applications such as Cumulus, Big Switch, and SnapRoute versus merchant silicon like Broadcom, Marvell, and Barefoot.
Interestingly enough, in these situations it’s not the organization that’s driving disaggregation, but the open-source movement – an exogenous element from outside telco and networking product organizations that’s disrupting the ecosystem. While NFV and SDN are still works in progress in many telcos, they have significantly impacted telco infrastructure.
This leads me to wonder if the mirroring hypothesis can work in one direction, maybe it can work in the other?
The Disaggregated Agile Telco (DAT)
Telcos – and the cable multi-service operators (MSOs) – often complain about the over-the-top (OTT) players being a threat and potentially eating their lunch. They lament that unified communications and collaboration (UCC) solutions are taking away their ability to sell voice minutes. And many point to failed in-house video streaming offerings versus the much more successful offerings from YouTube, Netflix, and Amazon. They worry about devolving into commoditized connectivity, and so turn to NFV, SDN, SD-WAN, 5G, and IoT to try to remain relevant and come up with new value-added services to drive revenues.
Instead of waiting to be destroyed by the OTTs, why not disaggregate preemptively and actively manage cannibalization? In other words, run the mirroring hypothesis in reverse and have the organization mirror the disaggregation happening within telco data centers and in the telco network? For the remainder of this post, I’ll call this the disaggregated agile telco, or DAT for short.
Give Me More of DAT
Essentially, the DAT would have a division focused just on the network infrastructure and provide network services. I’ll now use the term API loosely – it’s not just programmatic function calls but the larger concept of a standard interface into the network infrastructure with pre-specified SLAs. The best way to describe this is an infrastructure-as-a-service (IaaS) approach where you can access network services – wireless service, fixed-line service, 5G network slices, etc., in prepackaged, easily provisioned, and consumable forms.
The rest of the DAT would be regrouped under different application groups, each focused on providing solutions built on top of the underlying infrastructure services. As part of this disaggregation, the DAT could learn how best to structure these “APIs” and build the right abstractions, as well as how best to provide a unified platform that could work for the different applications businesses and consumers purchase today.
There’s value in the DAT learning how to create an infrastructure platform that services applications from UCC to SD-WAN, and from IoT for home automation to IoT for autonomous vehicles. Just as Amazon Web Services (AWS) figured out how to structure and scale their IaaS, which evolved over time into EC2, SD3, RDS, Route53, VPCs, and a host of infrastructure services that power a wide variety of consumer and enterprise applications, this disaggregation will provide important insights to the telco.
Cannibalize or Be Cannibalized? Hobson’s Choice?
At the right time, the DAT could provide the same first-class “APIs” to external OTT providers as well. If the external OTTs outperform their own telco applications, then perhaps their own telco application group should be shut down. Or the DAT could decide to withhold key APIs – that same old story of OS manufacturers withholding key API calls for their own apps; Microsoft, Apple, and Google have all been accused of this – if they feel uncomfortable with the rate of cannibalization. There’s certainly much more to this idea than can be covered in a short post, and I’ve had numerous conversations over the past five years with different telcos worldwide on the topic of disaggregating telcos. Most people I talk to admit that no or few telcos would venture down this path for numerous reasons: fear of the unknown, lack of internal expertise to carry this out, denial that something this drastic was needed.
I believe though that the brave telco that takes on the DAT approach has much to gain. If done right the telcos could be in a situation to out-Twilio Twilio, out-NetFlix NetFlix, and out-UCC Zoom or UberConference or Dialpad. And if mis-executed, yes, they would have much to lose and prematurely arrive at the inevitable outcome of being commoditized. But the question is, would you rather manage your own cannibalization – with a chance to survive and even thrive – or be cannibalized by someone else?