That time of year isn’t it? Everyone becomes a fortune teller and starts writing about things that will happen in the year ahead. Last year I wrote about IoT, and we did see a lot of development in that area - including (sadly) the security watchpoint I mentioned in the article.What about this year? Well, 2017 to me looks like we’re going to see another big leap forward in the world of cloud, although of course with a year being a somewhat arbitrary unit of measure, we’ll still be talking about most of the trends I’ll highlight for a few years to come.
Ready? Let’s go:
The continuing rise of Serverless computing
I wrote a year ago about the basics of Serverless and why we’re excited about the opportunity here at Cloudy Towers.
We’ve seen some really interesting projects in this space in 2016, and are expecting to be involved in many more this coming year. It’s not going to replace most enterprise IT in 2017, but we’re going to see significantly more adoption as further use cases and design patterns are documented and understood. Whatever market you’re in, you’ll see new entrants trying to compete with you using frameworks like AWS Lambda to do smart things quickly, ever abstracting away that thing called “servers” that people used to worry about….
Taking this chain of thought further out into the future towards 2020, I expect to see this growing and growing and growing. More bullish authors, including the excellent Simon Wardley, note that for organisations slow to react, by the time you’ve got your head around DevOps and cloud basics, it will be irrelevant. Personally I still think there’s a huge part to play for DevOps and containers due to sheer volume of legacy IT that exists in the world - but I do agree we’ll be seeing more companies bypassing that container stage of their lifecycle completely and going straight to Serverless as a model for deployment and operations.
And yes, I know the name “Serverless” isn’t great, but it’s the mostly widely adopted right now. Perhaps at the end of 2017, we’ll all be using “Function as a Service”. Or perhaps just “Lambda”…..
AWS have stolen a march on the world here, and it’s going to be tough for anyone to catch them. I’m hoping for some really interesting developments this year.
Cloud moves to the Edge
This builds on the previous point, as it links heavily to Serverless frameworks, and indeed the rise of IoT in general.
Everyone knows IT moves in cycles, although fewer are sure why. The current cloud models are based on centralised computing at scale. The next logical development therefore is for significant distributed processing to emerge.
Why? Well, look at the rise of IoT and what it means for data and processing. There are insane quantities of data being generated locally by the “things”, whether that’s a self driving car or a machine press in a factory. The sheer volume of data - in some cases tens, or hundreds, of GB per thing per day - often needs to be processed in real time for quick decision making and feedback loops. You simply cannot upload that back to a central system and query it in the time scales needed for many applications (like your Tesla brakes….). Ergo, you do it locally on the thing.
There’s a fascinating video (seriously, worth a watch) from the guys at Andreessen Horowitz on this topic - do watch it if you want to understand more in this area.
Will this revolution happen entirely in 2017? Nope. As with Serverless though, we’ll see a jump forward in this area, specifically the Edge and IoT objects in general becoming more sophisticated in terms of processing ability. And similarly, if you thought Serverless was a bad name….how do you feel about Fog Computing for this design pattern? <runs to be sick>
We’ve seen this year a glimpse of the data security challenges we’ll face with more sensors and data at the edge. These challenges will continue. What will speed up development in general in this area is, as usual, AWS. As I mentioned in my last blog on re:invent, the release of Greengrass and indeed Lambda@Edge, makes it easier to get started with this distributed model of computing than was previously possible. One can only imagine the investment Microsoft are putting into Azure Functions - and hopefully some competition will drive aggressive innovation.
Back in the summer of 2015, I wrote on this topic, and we’re seeing more and more developments as people see use cases which resonate with their business, and Microsoft (the leaders in ML right now) and the other major players developing their offerings. Expect a lot more in 2017.
People normally write (as I did!) about vending machines, elevators and the like when not thinking about autonomous cars. I think a really interesting recent example is this one from Aberdeen Asset Management. Know any fund managers starting to sweat?
Now, ML of course links to the previous point about computing at the edge. If data for quick decisions is held at the edge, then a lot of ML will also take place there. BUT, think about the autonomous car example again. While some decisions have to be made at the edge, the data volumes required to get driving training sets to work effectively is going to have to be centralised due to the sheer scale of it. There are, as a result, opportunities around networking and compression for those smart enough to tackle them - how on earth do you get all that data back to a central cloud efficiently in real time?
Data loss and security
Sadly, this is going to be a huge trend. The world learns lessons slowly and you can expect breach after breach, and hack after hack to be publicised this year. Security is one of my favourite topics and I write semi-frequently on it, so I won’t labour the point other than to say that there’s a huge skills gap here that needs plugging, and it isn’t going to be filled for some time. In the meantime, for the love of god, please a) encrypt everything, b) two-factor anything that can be, c) invest in some form of proactive threat management.
If you want a more bold title, Forbes recently wrote that 2017 will be the “year of cyber warfare”. Sleep well folks....