For tech entrepreneurs and investors, one of the most valuable insights in the business is understanding where the infrastructure is heading. Building an internet company in 2012 without integrating cloud-computing would have been a giant missed opportunity.
In the digital economy, how we live, work and play is constantly evolving and unlike previous telecommunication inventions — like the television — the change is consistently dramatic.
Over the past 10 years, the internet infrastructure has basically transitioned from Black & White to Color over, and over, and over again.
Five years ago, cloud computing was ‘the future’ and even today we talk about the cloud as if it is still a nascent technology — which, to a certain degree, it is.
Cloud companies get so much funding because in modern societies, entire companies run on the cloud — and not just their tech infrastructure but also payroll, financials and internal communications.
But, the cloud is not the future and, thankfully for us, the future has already been invented, but it is reminiscent of the internet in the early 1990s — it exists, but not many people are getting on board.
The future of the internet is called ‘Edge computing’, and if engineers actually develop autonomous vehicles, security robots and full-scale AR technologies, it will become the building blocks for this new version of the internet.
What is ‘edge computing’?
Think of it this way – edge computing is a piece of hardware that is right at the ‘edge’ of the source generating data.
What people predict is that a company will buy multiple data-generating sources (like a car, security camera or even IoT running shoes). These ‘edge devices’ will be plugged into each other, creating a mini-network whereby each edge device is now seen as a decentralised computational node. In a sense, it is like a company building its own intranet system.
This is where the word ‘edge’ originates. Today, our internet is built on a wheel-and-spokes model, with individual products feeding into massive centralised data centres which then processes the information at warp speed.
In its most basic sense, edge computing is a decentralised hardware device that provides an object with an isolated node, which has a stand-alone processing and storage system. When an edge computing device is plugged into another, it creates a unique, isolated, network.
So, why do we want to do this? This seems like a huge undertaking.
It is a massive infrastructure shift, which actually would return the internet to its late-1990s/early-2000s model of decentralised networks that do not interact with one another.
Anyone who used to rip music and burn them onto CDs knows what I am talking about.
To begin with, it minimizes latency. Because edge computing is deployed right where the data is collected, it is much faster than sending the data to the cloud. This is especially suitable for real-time processing and time-sensitive decisions.
In a video presentation, Andreessen-Horowitz Venture Partner Peter Levine used the example of driverless vehicle. Imagine that typical scene in which an elderly traffic policeman is holding a ‘Stop’ sign, walks onto the middle of the road, and helps people cross the street.
A driverless car needs to recognise the man, the stop sign and the potential for danger almost instantly. Furthermore, it must make a decision to ‘stop’ in a milisecond, or less.
Those factors + decisions are data, and today they must be sent to a centralised location, processed and sent back. Then the car can stop. While today’s internet is incredible, and this can happen extremely quickly, it is not good enough. If the car is travelling 50kph, by the time the “stop” signal has reached the car, it has blown through the crosswalk and run over the poor policeman.
Edge computing creates the ability to make to such time-sensitive decisions locally in the car and not in the cloud. The data never leaves the car.
If we bring into consideration that a driverless traffic system will be producing an amount of data that is almost incomprehensible, stakeholders begin to move away from “cool” and into the realm of “necessity” when thinking of edge computing.
Edge computing unleashes the potential of the actual product — our smart cars, IoT devices and medical technology — to become the data centre themselves. By doing so, something like an autonomous vehicle can run entirely independently, without needing any supporting infrastructure from outside.
Furthermore, people think that, should projects like autonomous vehicles become a reality, edge computing will be a necessity, because the current centralised cloud infrastructure is not a tenable solution.
Other, less intense, advantages
Annabelle Kwok, a Singaporean entrepreneur who is building an end-to-end edge computing solution at SmartCow.
Kwok’s hardware board is the size of a tablet and can be installed directly into a car, boat, medical device etc. She told e27 it is priced at about S$2,500 (US$1,800), but she’s working to bring the price-point down. For people familiar with cloud pricing, S$2,500 (US$1,800) feels like a bargain.
“The other option [more data processing] is servers, and that is expensive. Not only just to purchase it, but to maintain it. So what is the in-between? That’s where our device comes in. It is closer to a laptop price range than a server,” she said.
Plus, it is adaptable with various advantages. Kwok broke down a list below:
- It Minimises Latency (the car example)
- It conserves bandwidth (key for weak-internet areas)
- It is scaleable across the globe (the technology being built on hardware decks the size of an iPad)
- Improves security (hackers need to target hyper-specific networks rather than a centralised port)
- Reliability (it works on low-battery and is highly mobile)
- Mobile data (people will be able to take edge computing to far-flung places in the world).
While the billion-dollar opportunity may lie in autonomous driving, national security and data analytics — Kwok pointed to a test-case where edge computing can do a lot to help change the world.
The example is malaria testing in poor countries.
Often, physicians hike into some of the most remote villages in the world to take the tests. They register the information, but to send the data to their facilities, they must hike out to get internet.
This process is not only slow, but it limits the amount of tests the physicians can take because of biohazard concerns.
“It defeats the purpose for early detection by the time they get the results,” said Kwok.
The reason edge computing is a solution is because of both the intranet capabilities but also because the hardware can run on minimal energy (the SmartCow system only requires a 12 volt battery).
If physicians can take the tests and analyse the data on-site, they could reasonably test, and treat, an entire village before heading home.
Problems with edge computing
Like all technology, nothing is perfect and edge computing has its share of problems.
The most obvious is that companies are unlikely to reject the cloud any time soon — and to say edge computing will replace the cloud is a bit of a stretch.
If we use ourselves, e27, as an example, the cloud is insanely convenient for internal communications, accounting and storing company documents — but we are also not building a deep-tech revolution over here. We have no need to build an entirely independent intranet infrastructure, and with a global employee-base it actually might become more inconvenient.
I would argue that most companies are like e27 and edge computing would be called a luxury by an optimist or a waste by a pessimist.
To this point, the largest barrier, according to Kwok, is cost. Pursuing edge computing is a full-scale infrastructure overhaul and it will surely need to interact with legacy systems. How this happens is unclear or has not been invented yet.
Finally, while the hardware exists and is performing admirably, these types of systems require engineers (both software and hardware) that know what they are doing. The autonomous driving vehicle requires the best-of-the-best software engineers to run the system and pinpoint problems.
Even if a company wants to migrate towards the edge, they may not have the people to run the system.
However, if edge computing becomes the future, people will learn and adapt their knowledge to this new approach to the internet (like they always do).
The internet eats everything
It seems more inevitable than questionable that in the next decade the internet will infiltrate our daily-life in such a way that it becomes almost indistinguishable from how we live offline.
Even today, if someone wanted to spend the money, they could wear an outfit that is essentially a personal data centre. The homes of wealthy people are now generating a massive amount of data, soon our transportation system will become a mobile data centre and if Elon Musk has his way, maybe even our thoughts would generate physical data points.
Our current cloud computing infrastructure would not be able to handle this type of information in a seamless, affordable fashion. Which means one of two options are available, we see major developments/improvements in the cloud infrastructure, or a completely transformational system is integrated into our lives.
For those who believe the second outcome is more likely, edge computing may very well look like the future of the internet.