Kilton Hopkins has spent his career creating technologies for real-world use, most recently around IoT and the edge He was IoT adviser to the most tech-forward city of them all, San Francisco. Farah Papaioannou jumped out of the enterprise IT world (from Hitachi and HPE) into the world of venture capital (yes, frying pan to fire is a saying that springs to mind) and while a professional investor was involved in go to market initiatives for a number of companies with great technology ideas but sometimes poor go to market knowledge. As an aside, she was heavily involved in everyone’s favorite new-age storage company SolidFire, before it was acquired by NetApp.
Anyway, back to the subject at hand. He has deep technology smart while she has an innate understanding about how to take those products to market and wrap investment around them. Put the two together and you have a strong team that has now focused its attention into the emerging edge computing space via recently emerged-from-stealth startup Edgeworx.
For those who need a quick primer, over the fast decade or more we have seen a concentration of computing into a small number of massive-scale providers. The cloud has driven efficiency and cost-effectiveness and has certainly democratized computing, but it has also concentrated that computing in a handful (OK, a couple of handfuls) of locations. That’s good from the efficiency perspective, but kind of sucks when you’re looking for the lowest possible latency, when communications aren’t reliable, or when you have some pressing need to compute “at the coal face.” Which is where edge computing comes in – the idea of edge is to enable discrete and contextual computing to happen at the edge (hence the term, OK?) – a good example would be image recognition built into a camera or real-time traffic management built within traffic sensors on highways.
But here’s the thing. Edge computing, at least in some ways, goes against the big computing vendors demand for ever higher levels of centralization. Beyond the efficiency question, however, it also shifts the balance of control. If an end-user organization is able to process a piece of data at the edge and doesn’t need to send it back to the cloud, then that is one piece of data that is unable to be molded, mined and monetized by the big players. While not a conspiracy theorist or overly concerned that Amazon, Google or Microsoft are Big Brother-ing me, that’s an attractive proposition to some people.
Edgework wants to do things differently
So. Back to Edgeworx. The company, which is coming out of stealth today and announcing that they’ve gained some filthy Silicon Valley lucre to boot, is releasing its platform, ioFog. ioFog is an open source (and since the announcement at the weekend of IBM acquiring Red Hat, everyone is even more excited than usual about open source) platform that aims to be agnostic – that is, users will be able to use ioFog on any device. And in a world where pretty much every device at the edge has a different operating model, normalizing all of those is an attractive proposition.
But Edgeworx isn’t a one trick pony, rather it has a few value propositions – that of being the “AppStore for the edge”,” one around helping create mesh networks, and a security proposition to boot. Depending on your perspective, that triumvirate of things could either be seen as complementary and positive, or confusing and distracting – you be the judge.
As the company puts it, these three propositions can be articulated thusly:
- Intelligent: Edgeworx ioFog Engine turns any compute hardware into an intelligent edge device capable of remotely deploying and managing microservices. Edgeworx monitors the health and resources, so users can operate their edge effectively at scale.
- Connected: Edge-To-Edge Mesh Network creates a private connection between all of a user’s edge devices. No trips to the · cloud required and no more VPNs and NAT layers to deal with, so communication and data migration become simple.
- Secure: Pure Edge Security, designed from day zero for the edge, utilizes everything from hardware root of trust to microservice verification to create a peer-to-peer secure edge. Now every device secures the edge network. Instead of increasing the attack surface, users get a dramatically increased defense surface. The more devices at the edge, the more secure it becomes.
The open source angle
Edgeworx has actually been a bit of a work in progress for a number of years and part of this is an open source story. ioFog has been released under the banner of the Eclipse Foundation and has built up a developer community already – something that helps when you’re trying to create a platform that will work on such a diverse range of devices. Papaioannou, co-founder, and president of the company speaks to this open source opportunity saying that:
We believe that our open-source strategy is a differentiator at the edge. We break down the traditional IT/OT vertical silos, empowering a wave of domain experts to build and deploy novel edge applications. We are already seeing wide-scale traction with both customers and OEM partners in industries such as telecommunications, smart home, manufacturing, autonomous vehicles and oil and gas
Edgeworx strategy is to build the broad horizontal technology platform and to let their developer community (not to mention partners and customers) develop the particular vertical offerings that go on top of that. That is a relatively difficult opportunity to progress as customers want total solutions and Edgeworx is likely to get dragged into. For his part, Hopkins is bullish and ebulliently states that:
A long time in the making, we are pleased to announce the availability of our ioFog platform. With the Eclipse brand behind us and its five-million developer community, we are now able to provide a horizontal application platform that works in any industry, for any solution, on any device, while developers build the vertical applications. ioFog makes it so simple that any developer can write and deploy microservices for the edge in an afternoon.
How much will happen at the edge versus centrally?
A core tenet of Edgeworx’ proposition is that lots of the analytical operations that organizations need to run don’t require centralized processing. There has been the traditional view that it is only through the sort of mas computing that a centralized cloud enables, that deep analytics can occur. Edgeworx fundamentally disagrees with this viewpoint and has what I see as a three-level approach:
- If the analytics is reasonably compute-light, the algorithm can be deployed, across the network, directly to the device and the analysis can happen there
- IN a case of exception-identification, the edge device can do enough processing to pick out the data that needs more analysis and can facilitate sending that (and only that) back to the cloud – thereby greatly reducing communication throughput
- By developing and edge-to-edge peer network, Edgeworx enables “big software” to be run at the edge. In the same way that SETI thought of millions of distributed computers coming together like a giant, distributed whole, so to does Edgeworx offer a “cloud at the edge” paradigm
Regardless of how much processing actually needs to occur at the edge, one thing is sure and that is that the plethora of different SDKs makes building software for the edge a nightmare. This is where the “IoT AppStore” idea comes in – by normalizing all the different stacks running on the edge and providing a platform overlay, Edgeworx allows developers to build once and deploy across the edge – not having to think about firmware updates or physically flashing memory on site is a huge time saver. If you’re a telco looking to leverage your existing physical infrastructure and do some compute on random cell towers somewhere, “set and forget” is a very compelling offer.
The security thing
The final string to Edgeworx bow is a security one. As they see it, security at the edge is fundamentally different from in the cloud. While in the cloud the vendor doesn’t need to think about bad players having physical access to the hardware, at the edge physical access to infrastructure is the norm – how do you think about security when it is easy for someone to walk up to an edge device and plug in a keyboard or USB drive. This is where Edgeworx has done foundation work around edge security – dynamic keys, hardware root of trust and secure boot all being an example of their security thinking.
MyPOV
There’s a lot to think about here, and Edgeworx is trying to break into a space with lots of competition. That said, they’ve already seen success in one area that will, I think, be huge for them – telco. Edgeworx has a couple of interesting customer case studies in telco – one around using existing LTE base stations as edge computing points and another around turning dumb old WiFi modems in peoples’ houses into edge computing devices – both of these examples point to a very important market dynamic that Edgeworx is sure to leverage – the fact that telcos are petrified about being cast into obsolescence and are doing everything they can to “smart their dumb pipes.”
This opportunity is one which is also more difficult for the large cloud players to get over the line since telcos have a natural mistrust of these players. Combine a desire to do new things with a mistrust for the large incumbents and you have a real opportunity for Edgeworx – it will be interesting to watch how the team executes.