O estado da computação de ponta

KEMP

wrote last year about how the trend towards edge computing was going to be a significant driver for change in the data generation and processing space. And that there was going to be massive investments in edge computing infrastructure over the next few years. As I wrote last year:

Edge computing is going to be a significant driver for network evolution over the next five years. Spending on edge computing is forecast to be $9 billion per year by 2024…The rollout of 5G will drive a lot of this expansion as edge computing technologies are essential in delivering the local high bandwidth, low latency network provision that 5G requires.

The article also outlined how load balancing will be crucial to enabling edge computing and data processing in edge data centers. This post will cover the current state of the edge computing landscape, as reported in the recently published State of the Edge Report 2021.

State of the Edge Report 2021

The State of the Edge organization publishes the State of the Edge report. It is a vendor-neutral platform for open research on edge computing dedicated to establishing a shared vocabulary for the industry. State of the Edge develops free in-depth research with input and content from many leading experts in edge computing. Their report is now in its fourth year. It is a premier source on the state of edge computing, and on how the next-generation Internet will evolve over the next few years.

In the sections below, I highlight some of the key findings in the latest report. I encourage you to download and read the 95-page PDF to get a complete overview of the current state of the edge. But before I get into the headlines from the report, a short aside on how they define the edge.

How does State of the Edge define Edge Computing

One of the stated goals of the State of the Edge project is to create a shared vocabulary that the industry and others can use when discussing edge, IoT, and 5G infrastructure and solutions. They define the edge via these four principles:

  • The edge is a location, not a thing.
  • There are many edges, but the edge we care about is the last mile network.
  • This edge has two sides: an infrastructure edge and a device edge.
  • Compute exists on both sides, and they work together with centralized cloud computing.

Other definitions on what constitutes edge computing exist, but these four principles encapsulate the concepts pretty well for me. Now, on to the key findings from the report.

The Key Findings in the 2021 Report

The report concludes that we are entering the hyper-connected era. As Fay Arjomandi says in the forward:

We are in the midst of a transition from the mobile internet to the hyper-connected era where nearly every object in our physical world can have computing and connectivity built in, whether it’s a simple consumer doorbell or a complicated robotic manufacturing device. As the number of these smart devices grows, they will fuel automation and personalization, thereby transforming many industries.

People have been saying this for a while now, but all the signs point to it occurring. Sectors such as healthcare, manufacturing, energy, logistics, connected cities, online & physical retail, and emerging markets are experiencing a shift to last-mile processing and extensive data mining via machine learning and AI. 

The list below summarizes the headline findings presented in the executive forward of the report, but you should consult the full report for the complete picture.

  • Deployment of new edge infrastructure and applications continued apace in 2020, despite the impact of COVID-19. Seven out of ten areas saw increased forecasts compared to the 2020 report.
  • By 2028 approximately $800 billion will be spent on new and replacement server equipment and edge computing infrastructure (an increase on previous estimates). The edge spending will be split between deployments on the infrastructure and the device edges. There will also be an increase in the power and number of mobile devices used by end-users to consume the edge computing solutions offered by Telco’s and content delivery networks.
  • The fourth industrial revolution will continue to be driven by the key enabler of the edge cloud within the wider cloud computing landscape. The continued uptake of IoT devices, the spread of the global sharing economy, and the increase of zero marginal cost manufacturing will be key drivers to deliver large economies of scale across multiple sectors.
  • The computing power needed to drive the required processing at the edge is fostering fierce competition between the major chip companies. Intel, AMD, and Nvidia are all looking to design, create, and sell the best silicon to deliver the AI, ML, networking, and other dedicated processing units needed. The competition will increase in this sector as new entrants from China and other countries enter the market to supply these nationally important classes of components. 
  • The options for edge infrastructure deployment have expanded in the last year and will continue to do so. Service providers running hyper-scale data centers, regional data center suppliers, and traditional telecoms providers all now offer colocation solutions, micro data centers, wireless tower edge infrastructure deployment, and deployments in existing and new local network edge locations.
  • Hybrid and multi-cloud solutions from major players like Amazon Web Services, Google Cloud, Microsoft Azure, VMware, IBM (Red Hat), and others are extending the cloud experience out to more locations for use in innovative ways.
  • The marketplace for off-the-shelf edge solutions is growing in size. This is making it easier for companies to build the edge computing solutions they need from preexisting components. Google Cloud now includes over 200 edge-focused applications and services from 30 software development companies. Amazon Web Services and IBM are also expanding their off-the-shelf cloud edge solutions. 
  • The availability of open-source edge computing projects is also enabling rapid expansion while at the same time helping to deliver some standardization across the industry.
  • 5G is not yet a requirement for mainstream edge applications. Existing 4G-LTE, wired, and wireless networks are adequate for most use cases. However, the report authors expect more services that require the ultra-fast local speeds delivered by 5G networks to increase. Autonomous vehicles are often mentioned as an example that will need 5G. I expect the truth to be more prosaic, with 5G networks enabling intelligent services for solutions like public transport tracking (traffic congestions, busses route trading, and similar). 
  • SD-WAN (Software-Defined Wide Area Networking) and SASE (Secure Access Service Edge) are just as applicable for edge purposes as they are for enterprise deployments. The security, resilience, and session awareness that these technologies enable are beneficial to mission-critical services deployed at the edge.
  • It’s not all good news. As the number of edge node sites increases and the need for real-time data processing of the data generated increases, the digital infrastructure used will increase its power consumption from the 1 GW used in 2019 to 40 GW in 2028. At a time when there is a push to reduce carbon emissions by 2030, there will need to be a corresponding green revolution alongside the edge revolution.