For those who haven’t attended one, FutureDecoded is an annual UK Microsoft event held in London where they take a deep look at where they see the world going. This February for the first time they also ported the Future Decoded format to Mumbai. Following hot on the heels of Ignite, it’s not so much about new product announcements, but showcasing recent realworld case studies where Microsoft technologies have pushed the boundaries as well as future-gazing at the sorts of technologies that are going to be relevant in a few years’ time.
In previous years they’ve had Brian Cox on quantum computing & Stephen Hawkins on AI. This year, on that front, they had Karin Strauss from Microsoft research future-gazing about data storage on DNA, yes you heard me correct. The premise being that data growth has been accelerating past our ability to build it, and the only way out is to store it much higher density: for example, with new materials such as synthetic DNA strands. When the technology hits production, with DNA, we could store every piece of data in the world in one small room.
Personally I love these events as it lets me be inspired about what others have been up to, gives me head space to reach outside the boundaries of the projects I’m working on – and imagine greater things for my future work. I like writing up my notes because it gets my analytical brain racing and because giving others the benefit of a 10 minute summary is my way of gifting you a double whammy: of both time & knowledge.
These are some broad themes I could pull out from the sessions I attended on the Technical Day: –
Digital Experiences for First Line Workers through Mixed Reality
Microsoft under Nadella want us very much to think of them as the company to help everyone on the planet achieve more. And they want to spread their wings beyond the ‘information workers’ in office, who by in-large they have covered; to a group they term ‘first line workers’, of which they estimate there are about 1.7 billion in the workplace. These are people who work in the realworld as opposed to digital versions of it. They see this as being the perfect use case for Hololens: using 3D mix-reality to visualise things before they are implemented, augmented remote collaboration or training activities where you need both hands free.
I went to the Hololens demo where I was able to try out a training programme created by Mercedes for teaching their mechanics how to change brake callipers. It was 19 steps in total, but each one was fully instructed with on screen holographic overlays. It was impressive. The technology has some way to go before it’s a no brainer, but the potential is clear. The headset is still too heavy to wear for long, and unlike VR where you get total immersion – at the moment you only get a small holo window in the middle of your field of vision – which in my opinion is too restrictive.
Azure=The Open Cloud
Microsoft want to reframe Azure as the ‘Open Cloud’. I’m not really sure where this leaves the likes of Rackspace & their Openstack. But they want customers to see Azure as the platform of choice for everything, simply by giving unrestricted choice & making it remarkably easy to get started no matter what your field of expertise: whether you’re doing advanced Machine Learning, Blockchain or Big Data; all without technology prejudice: in Azure you’ll be able to get preconfigured development & production environments up and running in a matter of minutes.
An interesting stat I saw recently in the news is that 40% of Azure VMs are running Linux; and that’s something that you could possibly have imagined a few years ago. There’s obvious overlap in positioning with AWS on this. And let’s not forget that in Q3 2017, AWS revenues were still currently more than the next 5 cloud providers combined.
But Azure is growing at a faster rate and as a result is gaining. Where Microsoft differentiate is their emphasis on the Enterprise, courting of developers by embracing open source, their focus on good developer tooling to support their stack & selling their products such as Office or Dynamics in a SaaS model
Other Cloud Trends
AI everywhere. You might have noticed recent collaborative efforts with Amazon such as allow Cortana & Alexa skills to be combined; and the launch of the Gluon framework for building Machine Learning models simpler. But Microsoft are really going all out on AI with the way that it is going be fully integrated into Office365 & Dynamics365, which for most people does put AI everywhere. But with container technology they what to soon be able to deploy your ML models wherever they’re needed in order to be performant. And this means being as close to your data as possible, for an IoT device this means being on the device.
So not just in the Cloud, which is a big step forwards, as there are plenty of situations where you want to use AI, but can’t use the Cloud. Perhaps you have data that is too sensitive for the cloud, or your connection is too unreliable.
Back in the Cloud, Mark Russinovich, Azure CTO talked about how Azure now offers GPUs & FPGAs in the cloud for much higher density compute for Machine Learning models. And then demonstrated how using an FGPA he was able to process decisions with 15 times better throughput and 50 times better latency than a GPU-based architecture. This offers in his words ‘true democratising of AI’. Beyond the sprinkling that you’ve noticed in Office, very soon you’ll be able to reference functions from your Machine Learning models in Excel just like any normal Excel functions. Consider this: very soon if you can use a spreadsheet you can do ML. That’s the true democratisation of AI. Although the prospect of Excel taking over even more of the world makes me shudder.
No writeup of mine would be complete without a mention of bots. So here it is. Bots! Darren Jefford, Microsoft Digital CTO, talked about AI Powered Service Experiences, of which bots were a big feature. But really the most interesting thing I took from this session was the forthcoming Dynamics 365 Customer Service product that allows you to mix the current scripted-bot techniques used with MS Bot Framework with Machine Learning from past conversation histories. This gives a more scalable experience. The product has been field tested with brands like Macys, HP and their own Microsoft Support service, which is currently handling around 40% of support queries automatically.
As you can see from the screenshot below, it’s all configured from Dynamics, which gives a very nice all-in-one solution.
I’ve signed myself up for an advanced preview, so watch this space for a further update.
As I hinted above, Microsoft are looking beyond the Cloud: to allow central management but migration of some cloud capabilities – such as analytics & AI – to the IoT devices on ‘the Edge’, which may comprise devices that are intermittently connected but still have synchronised states.
This is still not available for public preview, so the details I have of this are sketchy.
Blockchain is going to be BIG. That’s the consensus at the moment. I don’t have space here to delve into how blockchain works here & why you might want it in your Enterprise. But you know that we are right at the top of the hype-cycle when your mum starts asking you about Bitcoin and you’re talking about ICOs down the pub. But it’s still interesting to see how Microsoft have a dedicated blockchain engineering team to help companies through the challenges of deploying blockchain into Enterprise environments. And the good news is that all of this work is shortly going to filter down to Partners so that they can help their clients crack the hard problems they have been experiencing.
In Azure, you can get all the infrastructure you need for your blockchain project: Secure Compute Enclaves created through SGX hardware, a Consortium-friendly Coco wrapper around your preferred blockchain protocol, and even a blockchain-ready App Builder kit. To date blockchain has been at the vintage Kristal & caviar-filled truffles end of the software development menu, as with any areas where there are few established patterns, you were typically looking at 6 figures just for a proof of concept running. And for this reason, despite all the hype, there have been precious few real-life success stories other than the obvious application in cryptocurrencies.
Craig Hajduk from the Blockchain engineering team spoke about how it took a whole team of top engineers weeks to get their first environment running in Azure. Now you can have a stable blockchain production environment in Azure in an hour. And all you have to do is select & fire it up from your Azure portal.
I’ve not yet field-tested these out for myself yet, so I have to take their word that it really is as easy as pie. I’ll update you again soon when I have.
The final trend is Serverless.
Overtime, with IaaS, then PaaS, then containers the server has been abstracted further and further away from developers. Not because developers aren’t capable of handling a server or two, but merely it’s acknowledging that if you want to build systems that are scale-ready, its much simpler if you don’t have to spend effort thinking hard about the DevOps required when you have to horizontally or vertically scale your services.
With serverless: Azure Functions & Logic Apps as your glue; you can outsource those concerns to Azure and concentrate on the logical architecture.
About the Author
Ed Yau is a Solution Architect that works at cloudThing’s UK office.
If you want learn more about implementing solutions in Microsoft technologies, cloudThing are a Microsoft Gold Partner who can help you to bootstrap your technology initiatives.
Here are some related links for further reading.