Data Gravity and Edge Computing

Data gravity sometimes is defined as the ability of a mass of existing data to attract more data, services and applications. The idea is that as datasets become massive, they are harder to move, and then also attract other apps, services and data that benefit from proximity to the critical mass of existing data. 

Think of the notions of ecosystem, economies of scale or ability to add scope. In other realms of life, data gravity is akin to the reasons businesses and people tend to cluster in cities; why amenities and specialities flourish in cities; why ecosystem partners locate near other partners. 

Perhaps data gravity is an example of the big getting bigger as more processing and storage happens at a relatively small number of global locations, driven by a shift of enterprise computing from owned facilities to use of cloud services.

Paradoxically, data gravity might grow, even as some forms of processing become more decentralized. 

source: Digital Realty

The reason is that data analytics are more likely to happen at remote and centralized facilities, even if real-time telemetry and process control shifts to edge facilities. For a growing range of applications reliant on artificial intelligence and real-time process control with huge data sets, processing must be at the edge. 

The need for ultra-low latency will drive processing at the edge, essentially counteracting data gravity that happens for other reasons. 

source: Digital Realty

Evaluating MEC Cost

What something costs can be evaluated only in relation to its value, and that applies to edge computing--including multi-access edge computing--as well. For smaller and mid-sized organizations, the cost of MEC might be higher than relying on remote computing at a hyperscale data center, but typically will be far less costly than an on-premises, owned solution, according to Ericsson. 

source: Ericsson

As always, assumptions matter.  Ericsson used figures for a connectivity provider in Sweden. Aggregation sites have power capacity installed up to 10kW, hosting an average of eight server units, each with four cores. 

With 8,000 access sites and one aggregation site per 10 access sites, there is a virtual processor (vCPUs) capacity of 25,600 (800 sites x 8 servers per site x 4 cores per server) for enterprise applications at CSP-owned edge sites. 

Servers are depreciated over three years while investments in power and cooling systems are depreciated over 14 years. 

Up to 1.6MW (800 sites x 8 servers per site x 250W per server) of incremental power is required. With an assumed power efficiency factor of 2, 3.2MW power is needed on average to power all the aggregation sites. 

The cost of compute resources at each aggregation site is estimated at USD 20,000. This cost is very similar to USD per critical watt for building a large-scale data center, according to Ericsson. 

Opex is the sum of electricity cost as well as operations and maintenance. Ericsson used a range of USD 0.10–0.15/kWh for power costs. 

All that matters only in relation to the business value of applications that either require or significantly benefit from MEC. That would include ultra-low latency use cases, but also instances where privacy concerns are high or the cost of WAN bandwidth is significant.

Another Kind of "LIving on the Edge"

Number one is my favorite. Dolphins and seals also enjoy riding the waves. Also, that experience is most similar to what regular, everyday surfers might experience, most of the time. The other nine feature talented surfers riding waves with excellent structure, at places most surfers do not ride, on good days for waves. Most people never are "in the barrel" this way, as it takes a certain type of reef (coral, not rock). Rides like that on a sand bottom are exceedingly rare. 

Meta Gaming Support Uses Edge, Orchestration, Workflow Changes, Cluster Management

By definition, streamed content and games are remote apps, as opposed to local computing operations that can be conducted directly on a device or on an organization’s own premises. 

Cloud gaming, audio and video streaming require higher levels of latency performance than many other apps, which is why edge computing is part of the fabric of the present distributed computing architecture and also will be part of the next generation of cloud computing. 

Remote and local computing each have advantages and disadvantages. Content operations typically benefit from remote storage. The downside, for highly-interactiver experiences, is performance. 

 

“Meta’s data centers alone cannot provide the level of ultra-low latency we require for cloud gaming,” say Meta engineers Qunshu Zhang and Xiaoxing Zhu. “So we rely on edge computing, where we bring the cloud gaming infrastructure closer to players by deploying in edges that are in metropolitan areas close to large populations.”

source: Meta 

But gaming itself is evolving. “From new 3D experiences like AR and VR to what will eventually become the metaverse, people all over the world want to play increasingly immersive games as seamlessly and easily as possible,” they note. 

Meta also created a hosting environment to support the graphic processor units as well as server cluster management, work flows, security and orchestration of streamed content to support gaming apps.

Cloud XaaS Market is Not Yet Fully Stable, but it is Gettting There

The latest Gartner estimate of infrastructure as a services market share is in some respects unexpected. The other major providers are growing faster, but Amazon Web Services continues to hold the lead. 

Global IaaS Public Cloud Services Market Share, 2020-2021 (Millions of U.S. Dollars)
Company	2021  Revenue	2021 Market  Share (%)	2020  Revenue	2020 Market Share (%)	2020-2021 Growth (%)
Amazon	35,380	38.9	26,201	40.8	35.0
Microsoft	19,153	21.1	12,659	19.7	51.3
Alibaba	8,679	9.5	6,117	9.5	41.9
Google	6,436	7.1	3,932	6.1	63.7
Huawei	4,190	4.6	2,681	4.2	56.3
Others	17,056	18.8	12,697	19.8	34.3
Total	90,894	100.0	64,286	100.0	41.4
Source: Gartner (June 2022)

Were IaaS the only element of cloud computing, we might conclude that the basic market structure already is set, and unlikely to change. Stablle competitive markets have a characteristic structure. 

Bruce Henderson, founder of the Boston Consulting Group is credited with a couple of foundational ideas about business, including the notion of the experience curve, which explains how the cost of products decreases with volume.

The “classic” stable pattern would have market shares where the market leader had twice the share of number two, which in turn has twice the share of provider three. Some might refer to that as a “rule of three.”

“Costs characteristically decline by 20 percent to 30 percent in real terms each time accumulated experience doubles," Henderson posited in 1968.

Some note there is a similar return on sales and market share relationship. Also, others note that market share and return on investment also seem to have similar relationships.  

source: Marketing Science Institute

Were IaaS the only part of the cloud computing as a service market, we might well conclude that the market share pattern is set, and is unlikely to change.

Of course, there are other key products, such as platform as a service, software as a service and private hosting. Even considering the fuller range of products, the market share structure is approaching a stable 4:2:1 pattern. 

source: TechTarget

Edge Computing, Private Networks Inextricably Linked

Much edge computing is inextricably linked with private network usage as well. In a salient use case, private 5G networks might support factory automation, linking sensors with local prccessing.

5G private network demand is a complicated matter. Defined as use of 5G infrastructure by an enterprise to create its own 5G connectivity within a building or on a campus, 5G private networks have been seen as a suitable support for manufacturing applications, for example, to support sensor networks and edge computing, often on the premises.

Nokia, Ericsson and other 5G suppliers naturally believe this is a growth opportunity for them, as factory automation use cases might be use 5G rather than Wi-Fi, for example.

Forecasters sometimes try to think about the total market in terms of private networks that might otherwise use Wi-Fi. But 5G does not make all that much sense for most Wi-Fi use cases where support of phones, PCs and other office or consumer equipment is the use case, generally fixed but also for “mobile” applications such as in-car connectivity.



source: Grand View Research

Grand View Research says the global 2021 market size for 5G private network services, hardware and software was about $1.4 billion. If the growth rate is 48 percent, then it is possible 2030 global revenue could be nearly $48 billion.

5G Private Network Forecast Global

Year                 2021   2022 2023    2024   2025    2026 2027   2028   2029   2030   

Revenue $B 1.4    2.1  3.1    4.5 6.7     9.9 14.7     21.8   32.2    48

source: IP Carrier estimate

Much hinges on the assumptions about growth. Also often unclear is whether we ought to be counting indoor 5G transmission infrastructure (“neutral hosts”) as 5G private networks. That has been expected to develop.

What remains unclear is the degree to which enterprises will find enough value in private 5G to replace some Wi-Fi connectivity. This is essentially a new form of the “Wi-Fi versus mobile network” debate that has arisen episodically over the past three decades or so.

Some have argued that Wi-Fi would be a replacement for mobile connectivity. The argument now seems to be about the degree to which 5G can replace Wi-Fi.

Applications that require ultra-low latency and predictability will be the use cases where private 5G will be considered, rather than Wi-Fi. But Wi-Fi proponents will not be standing still. So private 5G will still be compared to Wi-Fi that has much-better latency performance and quality.

Private networks using 5G will not displace most Wi-Fi networks, but will augment some Wi-Fi networks. No private network will displace the public networks for most outdoor, untethered connectivity.

ABBA Voyage Concert is an Example of Practical Metaverse

SaaS and Software Sales: How Much Overlap?

cloud services will represent about $454 billion in sales volume in 2022, according to Precedence Research. By way of comparison, Statista says global software will amount to about $609 billion in 2022 sales. 

source: Precedence Research  

As always, the forecasts possibly overlap, as some “software as a service” revenues might also be counted as “software” sales, where retail SaaS is counted as well as software license revenue for app companies that sell wholesale to providers of SaaS.

Software as a service might represent close to $200 billion in 2022 revenue, according to Statista.