Discover Canonical in 2017 on p60
Jonni Bidwell didn’t get to go to Barcelona for MWC 2017, but he did get to chat with Canonical’s EVP of Devices and IoT, Mike Bell, who did.
Mobile World Congress 2017, which this year ran from 27 February to 2 March, is the premier conference for the mobile industry. It showcases the latest mobile developments from telecommunications operators, IoT makers and innovators, and leaders in enterprise. Mike Bell manned Canonical’s stand there and was kind enough to chat with us and share his experiences. An industry veteran, he’s been CTO for Jaguar Land Rover, then director for their Global Connected Car program. Going back further, he’s been in charge of Information Systems Strategy & Architecture at United Utilities. Building on these experiences he now heads Canonical’s Devices and IoT team, and we were excited to hear what they’ve been up to.
Linux Format:
Our budget didn’t extend to sending me to Barcelona, so how was MWC 2017? What sorts of visitors did you get at your stand, and what were they mostly interested in?
Mike Bell:
From our perspective it was actually our best MWC – we had record numbers of people at our booth. It was quite an intense show for us as well, the level of interest that we had was quite diverse. We had traditional telcooriented businesses visiting the stand, but we also had the traditional players in the tech world, as well as device makers – so it was a real mixture. There was a lot of interest in areas like smart cities, robotics, those are the kind of topics that came up. The other thing that came up and generated a lot of interest, was that we were demonstrating some edge gateways via Dell and a number of other manufacturers. We also demonstrated Programmable Logic Controller (PLC) equipment, running our latest IoT operating system Ubuntu Core. The PLC marketplace is dominated by proprietary solutions and technology, whereas what we were showing was more of a democratised open platform running Ubuntu Core to deliver PLC application use cases.
LXF:
Canonical have quite a history at MWC: At MWC 2015 they presented a top of the rack switch running all manner of softwaredefined networking applications and all built on cheap, commodity hardware. Then at MWC 2016 they presented LimeSDR, which enables hobbyists, makers and dabblers to make their own wireless devices and protocols. Can you tell us a little about what you demonstrated this time around?
MB:
Both of those projects are ongoing by the way. This year we showed a LimeSDR base
On that robot “Right now in the robot space, Ubuntu is really the default platform.”
station running on an Intel-based i7-platform. This was a base station that had a range of approximately a kilometre, running on a PCI board with Lime’s chipset on it. We also demonstrated using Ubuntu Core with snaps – we had a couple of partners delivering an LTE and an Evolved Packet Core (EPC) stack and we were able to show these downloaded over the air to the base station and then being able to launch an LTE terminus from that base station. The thing that’s really interesting is that we had virtually every telco who was at MWC coming to visit that – they were very interested. Right now EE are doing a pilot with Lime. EE have made the claim that their UK strategy is to provide 99% coverage, and delivering that is something that they need to look at the overall capital expenditure investment to make that happen. So technologies like Lime are very interesting for EE and BT, Vodafone are actively looking at this as well. These are serious evaluations of the technology, and I think 5,000 of the Developer Editions of the LimeSDR are being sent out as we speak, so it’s been a very successful crowdfunding campaign. This will be something where developers can actually build applications for the base station, deploy them using our Snapcraft tools and allow an experience to be created on base stations. I think this is really interesting because it does two things: firstly it addresses the proprietary nature and high costs associated with base stations and secondly it provides a marketplace to deliver applications close to the edge as well. So by having an open platform, we can support Intel and ARM-based hardware. As we currently stand Lime has produced an Intel-based architecture, but it’s not wedded to that. We have the developer kit which uses a USB interface, so that you can develop it on your laptop and plug it in as a separate board and then when you deploy it it’s deployed on a PCI card. So this will really lower the cost of getting the base station out to market.
LXF:
Presumably this will empower independent developers too, because there’s no need to buy expensive equipment or pay for access to proprietary frameworks. Anyone can join the party.
MB:
Correct. That’s the really interesting thing and we’ve had a huge number of developers register for the development kit, which shows there’s a nascent demand to break into an area which has hitherto been a very proprietary space. I think the other thing it’ll do, because we’re not tied to any specific technology per se, is allow more conventional open source development tools to be used, rather than very proprietary embedded toolsets. We also demonstrated a couple of network switches at MWC, on based on the Facebook Wedge, which is part of their Open Compute Project (OCP), so we demonstrated snaps running on an OCP Wedge device. We also showed a Quanta switch running Ubuntu with snaps. These are active, and there’s quite a lot of opportunities which I can’t necessarily talk about, but there are some really interesting opportunities with network vendors. They’re relatively conservative I have to say, but initiatives like OCP and the Telecom Infra Project (TIP) coming in allow organisations to start thinking about how they will focus on the software and less on the hardware. Tech giants like Facebook want to drive down hardware costs and our snap infrastructure provides a great way of doing that, because we can securely isolate and containerise the applications running on that switch on a relatively efficient operating system.
LXF:
I’ve seen some interesting pictures of your stand at MWC. There appears to have been a friendly looking robot hanging around a lot. Can you tell us a little about it?
MB:
These are service robots, one of them was used in an application in Russia and the other in the Middle East. They’re running Ubuntu and their goal was to provide retail services, it was fun to get them onto the stand. But they had a serious message, that Ubuntu has a huge number of applications that can be deployed. Right now in the robot space, Ubuntu is really the default platform. If you look at the Robot OS ( www.ros.org) development kit it’s pretty much standard on Ubuntu with only experimental builds available for other OSes. So we do see a huge demand from makers and from very serious large organisations to bring robots to market on an Ubuntu platform. Again, the focus is on time to market: I can use frameworks and I can use open source to deliver a huge amount of it, and then they potentially add their own closed source proprietary components where their IP-base is, their competitive advantage.
LXF:
Virtual reality (and augmented/mixed reality) have been talked about for some time now, in many ways it’s been seen as a toy, but now it seems like there are some serious projects using it. One example is the DAQRI helmet showcased at your stand.
MB:
AR is clearly a very hyped technology, but in some cases you could argue that it’s a technology looking for a problem to solve. DAQRI is a great example of where they’ve found a good niche. What they demonstrated was an ability to allow someone repairing some machinery to have additional information put into the display, to augment their vision and help them carry out a complex task. There are a number of applications – everything from aerospace to automotive repair and manufacture, utility asset repairs, anything where there’s a complex task that needs to be done correctly. It can provide an instruction set and overlay it. Where the asset is extremely expensive, something like repairing a jet engine or a vehicle, those are things where the helmet can play a great part. These are AR applications you’ll see that come slowly, because there’s always a business case within those kind of environments. The real end customer has got to make a business case for it, which is largely quality driven. But there are very clear use
cases that can create great business cases. If you look at something like a jet engine, which are now generally rented, downtime means no revenue for that developer, so it’s a really interesting application. LXF: If nothing else it means the engineer has two hands free, rather than them being full of probes, plugs and connectors. MB: That’s right. From my previous history, working in areas like oil, gas, utilities and manufacturing, these are all areas where you could be dealing with extremely expensive assets, whether they’re being made or repaired, and the ability to do that correctly and safely is paramount. So there are many areas where having a tool like this, where having both hands free, is critical. LXF: What was the most exciting thing you saw at MWC?
MB: Nothing specific hugely jumped out at me, but here’s a couple of things that resonated with me. One was the continual push of miniaturisation, and we showed a lot of this on our own stand in terms of the size of the boards. The fact is, on both the ARM and the Intel platforms, the SoCs and the platforms that we can deliver are getting smaller and smaller and costs are coming down... That miniaturisation, both in terms of physical size and costs, and yet the compute capabilities, what you can do with something that, say, costs less than $10. That was a theme you could see running everywhere.
Being a telco-oriented show, there was a huge amount of 5G stuff. It does make me smile slightly because we’ve yet to actually get that fully standardised. But the enthusiasm from the infrastructure providers to demonstrate 5G capability is really interesting. What it does show is that there are alternative applications that we can’t address with today’s cellular technology. Specifically, it’s not just about bandwidth – it’s about latency. One of the targets of 5G is sub- millisecond latency and I think that’s something that really makes the application space very interesting, particularly at the telco edge.
Lots and lots more IoT, and I think a big focus as well was on Low Power Wide Area Network (LPWAN) technology. I don’t believe all of these technologies can continue in five years time. Right now we’ve got NarrowBand IoT (NB-IoT) backed by the GSMA, as well as LTE-M, we’ve got SIGFOX and we’ve got LoRa. These are really interesting from an IoT perspective, but I’m not convinced the telcos are going to make as huge an amount of money as they expect, because these are relatively low data-rate technologies. But for device makers they’re very interesting because we can get connectivity from a boiler sitting in a cellar or a basement, where traditionally getting normal cellular connectivity would’ve been impossible. These kind of low-power technologies are definitely much better at penetrating difficult and remote locations, so they’re pretty interesting from that point of view. Self driving cars was a theme that came up and the amount of R&D investment going into the automotive space is huge. We’re seeing automotive companies moving into the world of software, and that will lead to some of the most complex devices that are out there.
LXF: Traditional telecom providers are facing all kinds of challenges nowadays. Canonical recently announced a partnership with AT&T, which is a big deal. How will this and other Canonical partnerships help to modernise the ‘old guard’?
MB: The Cloud division at Canonical has been very successful within the telco space. They deliver technologies like OpenStack, Metal as a Service (MaaS), Kubernetes, these resonate very well with the telcos. They need to refresh their backend infrastructure, to automate it as much as possible. Previously the fact that we could virtualise was a big thing, with Network Function Virtualisation (NFV) one of the biggest challenges is actually operating it. So really what we were describing was a whole bunch of technologies, and that’s what the likes of AT&T are using, which lower the cost of operating. That’s really a big push in the cloud space right now for the telcos. In the Devices and IoT division that I head up the focus is interesting because it’s very much from the edge – we talked about the line base station – but also right into the home. At the moment several of those telcos have had problems with putting in DSL routers that are white-labelled and having never really looked at security properly. We see an opportunity for Ubuntu Core, from a security point of view within houses and small businesses, but also to be able to run a platform. Right now your typical WiFi router doesn’t do a great deal and one of the things we’re investigating with the telcos is how they can offer additional services on that hardware. So rather than thinking about it as a very specific device, how about we add in a little more compute and memory, and then we’re able to run snaps which allow additional paidfor services to be offered by the telco. Or they could be offering free services helping to minimise churn and retain customers.
LXF: Just in closing, can you summarise Canonical’s overarching strategy and goals for this post-desktop world?
MB: The whole of Canonical’s focus really is Software Defined Everything. It’s a simple phrase but it actually means something quite powerful: the fact that software is changing the world. By making it open source and making it able to run on every type of device – whether it’s a small sub-$10 piece of hardware or a GPUenabled server within a data centre – it’s software that’s making a difference. It’s software that’s transforming businesses and that’s what we want to empower.