90% of New Operational Processes Will Happen on the Edge

The number of new operational processes deployed on edge infrastructure will grow from less than 20 percent today to over 90 percent in 2024, IDC analysts predict. 

Also, through 2023, reactions to Covid-19 pandemic challenges will be the dominant accelerators for 80 percent of edge computing investments and business model changes in most industries, says IDC. 

In 2021, 65 percent of organizations will have shifted to “digital first”  automated operations and contactless experiences, IDC says. 

By 2022, 80 percent of organizations that shift to a hybrid business by design model will boost spend on AI-enabled and secure edge infrastructure by four times to deliver business agility and insights in real time. 

By 2024, 25 percent of organizations will improve business agility by integrating edge data with applications built on cloud platforms. 

Will Edge Computing be a Functional Substitute for Network Slicing?

Among possible new revenue sources created by 5G, network slicing is expected to support advanced connectivity use cases for enterprises and governments that are primarily, though not exclusively, driven by untethered end points. 

In fact, many observers now believe 5G revenue growth will be disproportionately driven by new use cases and value for enterprise, business and government, with network slicing, internet of things connectivity and edge computing among the lead candidates for significant new revenues. 

Compared to today’s private enterprise networks, which are isolated and private, network slices represent virtual private networks that are public (wide area) and isolated. In other words, a network slice provides end point networking similar to that provided by any 4G or 5G mobile network, but with features optimized for a use case, and logically separated, as would be the case for a virtual private network. 

For enterprise or government end users, network slicing can enhance security and improve robustness. In the former case, any attacks are limited to a specific slice. In the latter case, network issues in other slices, or the public network, do not affect a particular slice. 

The issue is how much additional 5G revenue lift might be possible, from network slicing, for example. Some believe the connectivity portion of the market is fairly small, despite the hype. 

How small? Less than $1 billion globally by 2026. That is hardly a “market” at all, by global telecom standards. That noted, value sometimes is monetized in other ways, such as higher new account rates, lower churn, higher gross revenue, lower capex, lower opex or some combination of those benefits. 

And, if that happens, it will in large part be because business buyers of networking solutions have found some other way of solving those problems. Edge computing, for example, is a functional substitute for network slicing as a means of reducing application latency or bandwidth availability. 

source: GM Insights 

A network slice is a virtual network created through a 5G core network able to maintain private network features all the way to the end point. In principle, a network slice should be easier and faster to create and tear down, with advantages for temporary networks or use cases including disaster recovery, large events or other instances where unexpected--but temporary--demand emerges.  

 source: Nokia

Low-latency use cases also are candidates for network slicing. Mobile gaming is one consumer use case while internet of things applications are the best examples of enterprise or government use cases. But edge computing also offers solutions to such problems. 

Network slicing allows creation of virtual networks whose characteristics (bandwidth, speed, latency, reliability, and security) can be optimized. Network slicing offers a way to create virtual networks with performance characteristics optimized for lead applications. But buyers also can use dedicated private virtual or physical networks to solve many of the same problems network slicing is touted to address, including availability guarantees or other quality of service levels.

Processing speed also can be assured using private networks or edge computing. 

Mobile gaming using virtual reality or artificial reality might benefit from latency and bandwidth assurances provided by network slicing. Other use cases in the financial industry might benefit from enhanced security and network availability guarantees. Again, the issue is whether network slicing is the way to do so with lowest total cost of ownership and effectiveness. 

In the fleet management industry, different core functions might benefit from use of distinct network slices. Low latency and ultra-reliable performance might be best for  traffic notification, where infotainment requires higher bandwidth but less need for low latency or robustness. 

It is conceivable network slicing protects core transport bandwidth for fixed wireless operations, since fixed wireless peak bandwidth usage might be an order of magnitude higher than typical mobile data requirements. 

Quality of service also has been a driver of product pricing differentiation. Business services with QoS guarantees, support services reaction times and financial advantages for customers if the policies are violated, have been a staple of communications product pricing for decades, and network slices might be a new way to provide such services. 

The ability to support 5G network slicing in the core network, all the way to the mobile edge, is supposed to create new product capabilities, such as networks optimized for particular parameters. That might be among the strongest drivers of network slicing value, as QoS measures are assured to the actual device level (technically the access radio level). 

It would be fair to note that connectivity providers often have developed and offered  many features and services they  believed customers would want. The catch is that revenue for connectivity providers is the cost for enterprises, businesses and consumers who are asked to buy and use the features and services. 

If ultra-low-latency applications are those which could benefit from network slices, one alternative is to do commuting at the edge, and not sending data across wide area networks that are optimized for low latency. In other words, edge computing might often be a functional substitute for network slicing.  

In many use cases, the value of ultra-low-latency computing is supplied by edge computing services, with non-real time backup across wide area networks. 

Perhaps ironically, consumer customers who have few other alternatives might be good candidates for internet access with quality of service features a network slice offered by a connectivity provider, providing governments deem this lawful. Network neutrality rules often var the offering of consumer services with quality of service guarantees, for example. 

Gaming services, work-from-home conferencing and ultra-high-definition video are among potential consumer use cases for network slicing, where lawful.

Edge is in Early Days

As valuable and important as edge computing services and software remains a small market at the moment. 

The global edge computing market  for software will reach $1.72 billion by 2026, a  study by Mind Commerce predicts. The market for multi-access edge software in support of IoT applications will reach $546 million globally by 2026. 

Markets for edge hardware might be the most substantial markets of all, based on significant shift of infrastructure to the edge by computing-as-a-service suppliers. By 2023 or so, the large hyperscale cloud computing as a service providers might be deploying half their capex at the edge. That might mean annual spending of perhaps $34 billion by hyperscaler cloud computing giants alone. 

Connectivity revenue might not be much more substantial. Some estimate global edge revenue earned by connectivity providers in the $2 billion to $4 billion range. 

All of that suggests we are in the the early days of edge computing.

Edge Investments are Mostly "Change or Grow the Business" Decisions

As valuable as edge computing might become, for many firms in many industries, its value is not likely going to be primarily in the area of cost savings. More likely, the value will come from ability to enable changes in the business model, allowing firms to do things they could not do before.

So edge computing investments are mostly "grow the business" or "change the business" decisions rather than "maintenance capital" decisions.

Only secondarily, at first, will decisions about how to source edge computing be strategic. As always, the cost of achieving the desired capabilities will matter most. Matters can change over time, as scale changes.

Scale matters where it comes to many computing use cases. As useful as hosted voice services are, total cost of ownership is higher for large enterprises than an “owned infrastructure” approach, primarily because costs are directly related to the number of lines supported. 

Some companies with high reliance on computing facilities, such as software firms, will often find they can cut costs dramatically by shifting computing operations back to their own facilities, and off computing as a service platforms, partners at Andreessen Horowitz argue. 

That appears to be especially true for public software companies, where the computing infrastructure is an unusually-high percentage of total costs, Andreessen Horowitz argues. 

“When you factor in the impact to market cap in addition to near term savings, scaling companies can justify nearly any level of work that will help keep cloud costs low,” say Sarah Wang and Martin Casado, Andreessen Horowitz partners. 

“A billion-dollar private software company told us that their public cloud spend amounted to 81 percent of cost of revenue,” they say. Also, “cloud spend ranging from 75 to 80 percent of cost of revenue was common among software companies.”

Citing the example of Dropbox, the partners note that “when the company embarked on its infrastructure optimization initiative in 2016, they saved nearly $75 million over two years by shifting the majority of their workloads from public cloud to ‘lower cost, custom-built infrastructure in co-location facilities’ directly leased and operated by Dropbox,” 

Dropbox gross margins increased from 33% to 67%  from 2015 to 2017, which they noted was “primarily” due to Infrastructure optimization and an increase in revenue. 

Thomas Dullien, former Google engineer and co-founder of cloud computing optimization company Optimyze, estimates that repatriating $100 million of annual public cloud spend can often result in nearly $50 million in annual total cost of ownership from server racks, real estate, cooling, network and engineering costs, they note.

Other sources estimate savings ranging from 33 percent to 50 percent, they add. “A director of engineering at a large consumer internet company found that public cloud list prices can be 10 to 12 times the cost of running one’s own data centers, the partners say. 

Smaller firms, earlier in their growth cycles, almost always benefit from cloud computing, rather than building their own computing infrastructures. But that can reverse when firms grow larger and growth slows. 

“It’s becoming evident that while cloud clearly delivers on its promise early on in a company’s journey, the pressure it puts on margins can start to outweigh the benefits, as a company scales and growth slows. Because this shift happens later in a company’s life, it is difficult to reverse,” say Sarah Wang and Martin Casado, Andreessen Horowitz partners. 

source: Synergy Research 

“You’re crazy if you don’t start in the cloud; you’re crazy if you stay on it,” the paradox becomes. 

The issue is how cloud spending versus owned facilities plays out over time for firms in other lines of business that might not have computing cost as such a large driver of total cost of revenue. 

For most connectivity providers, computing cost is not likely a huge driver of total cost of revenue. Information technology costs vary by industry from 1.4 percent in manufacturing  to 11.4 percent in the financial services industry, for example. 

 

source: Computer Economics 

Others estimate information technology spending in about the same ranges, though spending varies by firm even within each industry. Some might argue maintenance levels of spending could range about two percent of revenues, while growth spending might range up to perhaps six percent of revenue. 

Perhaps most firms spend between two and three percent of revenue on all IT. 

 

source: Statista 

Few industries have computing costs as such a large driver of cost as do software firms. Few firms in most industries can improve equity valuations or profits as much as software firms can by optimizing computing cost. 

For most firms in most industries, the balance of owned versus “computing on demand” spending might not be crucial in that regard. 

In principle, if the same savings software firms obtain by optimizing computing spend were to hold universally, most firms in most industries could save relatively little by optimising computing choices. Saving 33 percent, if possible, on an item that represents two percent to three percent of total cost of revenue is helpful, but arguably not all that significant as a driver of cost and profit.

Cloud Ultimately is Derivative

Sometimes cloud computing is defined with reference to the way it is sourced--as a service--rather than where the servers are located, namely at remote locations. Other definitions emphasize how services are delivered (using the internet). 

The context for cloud computing might be directly related to the emergence of internet computing, allowing access (with proper credentials) to all resources using a web browser. In other words, cloud computing is derivative of web-based computing, application and content delivery. 

We sometimes forget that the term cloud originated in network schematics that abstracted “all other resources” not relevant to a particular local deployment of computing resources. That, in turn, became important in the client-server context, where local area networks introduced the use of external computing resource access within the office. 

The “cloud” then represented all other network elements any local PC or server could communicate with. 

source: Thinglink 

These days, browser-enabled access to content underpins most content delivery and application functions. A decade ago, the “internet” had emerged as a significant contributor to economies globally, the epitome of the information age  and arguably was the forerunner of all present efforts at digital transformation. 

So cloud is derivative of the internet; the use of the browser as the content interface and the shift of computing from “work”  to content delivery.

Multi-cloud trends in 2021

As seen by Telefonica. 

Cloud Services Revenue Reaches $312 Billion in 2020

The global public cloud services market, including Infrastructure as a Service (IaaS), System Infrastructure Software as a Service (SISaaS), Platform as a Service (PaaS), and Software as a Service (SaaS), grew 24.1 percent year over year in 2020 with revenues totaling $312 billion, according to the International Data Corporation. 

Worldwide Public Cloud Services Revenue and Year-over-Year Growth, Calendar Year 2020 (revenues in US$ billions)
Segment	2020 Revenue	Market Share	2019 Revenue	Market Share	Year-over-Year Growth
IaaS	$67.2	21.5%	$50.2	19.9%	33.9%
SaaS – System Infrastructure Software	$49.2	15.7%	$40.2	16.0%	22.4%
PaaS	$47.6	15.2%	$36.1	14.4%	31.8%
SaaS – Applications	$148.4	47.5%	$125.2	49.7%	18.6%
Total	$312.4	100%	$251.7	100%	24.1%
Source: IDC Worldwide Semiannual Public Cloud Services Tracker, 2H20

source: IDC 

The largest five public cloud service providers (Amazon Web Services, Microsoft, Salesforce.com, Google, and Oracle) captured 38 percent of global market share, growing 32 percent year over year, says IDC.

Microsoft now shares the top position with Amazon Web Services in the whole public cloud services market with both companies holding 12.8 percent revenue market share for the year (Keep in mind that Microsoft includes its end user software sales in its cloud revenues). 

source: IDC

Even Some SMBs are Cloud Computing Lead Users

No category of connectivity services customer is more varied than the “small and medium business.” That category can include home-based businesses, organizations with few employees or sometimes organizations with up to 1,000 employees. 

But employee count also varies. Software-based businesses can generate much more revenue than other businesses. So even some “smaller” businesses can produce enough revenue that they spend as much as enterprises do on connectivity or computing services and infrastructure, as shown in the Flexera 2021 State of the Cloud report, for example. 

The report includes 15 percent SMB respondents, including SMBs spending as much as $12 million on cloud services annually, and 38 percent spending at least $1.2 million a year on cloud services. 

source: Flexera 

Most firms spend between 1.2 percent to 5.9 percent of revenue on information technology, the exception being financial services firms that can spend as much as 11.4 percent of revenue on IT, according to Computer Economics. 

 

source: Computer Economics 

The Flexera study is led by firms in IT, financial services and healthcare. Some 64 percent of respondent organizations are in those fields. As such, the survey is disproportionately representative of leading users. 

source: Flexera 

The other important observation is that revenue--rather than firm size--likely is the better way to classify the firms considered to be “SMBs.” They might fit the definition in terms of employee count, but not in terms of revenue. 

A financial services firm of 500 people, spending eight percent of revenue on IT, and $21,000 per employee might generate $131 million in annual revenue. The point is that a financial services firm might generate far more revenue than its employee base suggests. Such a firm might be “small” by employee count, but at least mid-size by revenue. 

source: Flexera 

source: Flexera

Colocation and Private Cloud Might be Growing Faster than Cloud Computing

‘Edge computing’ is much harder to define than “cloud computing. AWS defines cloud computing as “the on-demand delivery of IT resources over the Internet with pay-as-you-go pricing.” 

All the clauses in that definition might not always apply to edge computing. 

Edge computing can occur on a device, in which case internet delivery is not necessary. It might occur on an organization’s premises, in which case delivery is by a private network. 

Also, when the computing resources are owned by the enterprise, there are no usage fees or payments. 

And one might not consider a device’s own processor and memory or a premises-located or remote location based user-owned server to provide “on-demand” delivery. In such instances workloads are invoked as needed, but not “on-demand,” when that term refers to use of a third party’s compute infrastructure. 

So edge computing uses the notion of computing close to the location where it is needed. 

That might always have been the case for personal computers, smartphones and other devices; enterprise and other organization computing on owned equipment, on the premises. In the case of remote computing it often means moving the resources geographically closer to the use point. 

Others might add that edge computing involves local computing or onboard computing; computing analytics on the device or premises; or happens wherever the digital world and physical world intersect.

Such notions suggest why hybrid computing has emerged. Workloads already are a combination of edge and remote processes. 

source: 451 Research, S&P Global Intelligence

Comparing NB-IoT, Cat-M1 and Cat-1 Access

No network platform is best for every application and use case and that appears to be the case for internet of things use cases as well. That applies both to the type of edge computing as well as the untethered access platform.

Connectivity can use fixed networks; unlicensed wide area networks or mobile networks optimized for internet of things use cases.

Connectivity cost per device also matters: some networks cost more, per device. Also, some use cases require mobility; others do not. Battery life often is a key consideration, and each network might have different strengths in that area as well. Consider three 4G-based standards, Cat-M1, NB-IoT and Cat-1. 

In practice, it is possible to buy devices using a dual-mode connectivity option, including both Cat-M1 and NB-IoT. Also, many original differences in features have lessened, or been eliminated. 

Cat-M1 often is necessary for wearable use cases, such as fitness bands and smartwatches,  asset tracking and health monitors, largely because Cat-1 supports mobility. It also is essential if voice communications are a requirement. Cat-M1 latency is 10 to 15 milliseconds. 

It also is used for automated teller machines, alarms, metering applications, security monitoring and building monitoring systems.

NB-IoT is more commonly used for lower-bandwidth use cases that also are stationary. Smart gas, water, and electricity meters, smart street lighting and parking sensors provide examples. One of the characteristics of these use cases is that data transmission is infrequent and small amounts of data are transferred. 

Heating, ventilation and air conditioning use cases, industrial monitors and agricultural sensors that monitor irrigation systems and detect leaks are use cases suited to NB-IoT networks. 

Cat-M1 supports 1.4 MHz bandwidth and also supports over-the-air firmware, software and security updates, as well as being designed to support Linux operating systems.

Cat-M1 supports full and half-duplexing, which can lower power consumption and increase their battery life when sensors are operated in half-duplex mode

NB-IoT supports uplink speeds of 66 kbps and download speeds of 26 kbps in half-duplex mode. Latency ranges between 1.6 to 10 seconds.

But NB-IoT occupies a very narrow bandwidth of 180 kHz and can be deployed in the guard band portion of an LTE network, meaning NB-IoT uses unoccupied portions of the LTE network.

NB-IoT also tends to offer better coverage inside buildings and wider coverage of outdoor areas as well. 

Cat-1 was designed for IoT devices with low and medium bandwidth needs, supporting bandwidth of 5 Mbps for uploads and 10 Mbps for downloads. Latency is low, at 50 to 100 milliseconds, and Cat-1 supports tower handoff, making it a viable option for asset tracking deployments. 

However, it consumes more power and its signal range is a bit shorter than NB-IoT and Cat-M1. 

As always, choices are a compromise between cost, battery life, coverage, bandwidth and other requirements such as voice support or mobility.