Postdiction 2014-2016: containers, semantics and cloud will grow together, not separately
The winner will not be a single technology stack, but companies that can combine several waves into one platform.
Year-end is the traditional time for predictions. I am not going to name specific products or companies that will "win". Nobody knows that. But I can reason about structural tendencies that are already visible in 2013 and will most likely strengthen over the next two or three years.
The main point is simple: the three technology waves that currently look independent will reinforce each other. And the winners will not be those who catch one wave first, but those who learn to combine several.
Why riding a single wave does not solve the problem
Take containers. Packaging an application, speeding up deploy, isolating dependencies - that is real value. But a container by itself does not solve the data question: where does data come from, how is it structured, how do different services agree on what the same thing means.
Take cloud. Scalability, flexibility, lower capital expenditure - real advantages. But cloud by itself does not solve the manageability question for complex systems: how do you coordinate dozens of services, how do you make their behaviour predictable.
Take improvements in data semantics - shared models, ontologies, exchange standards. These reduce integration costs and improve analytics quality. But without a data delivery infrastructure and reliable deployment, this still remains an academic discipline.
Each wave individually solves part of the problem. Together they begin to address a systemic one.
What this will look like in practice
I expect that by 2015-2016, companies that moved in several directions simultaneously will have a visible advantage over those who invested in one.
Concretely: an organisation that is simultaneously building managed cloud infrastructure, working on semantic consistency of data, and introducing reproducible deployment environments will have a speed of change that is otherwise unachievable.
This does not mean doing everything at once. It means understanding the connections between directions and moving in a coordinated way.
Where strategy goes wrong
The typical mistake is to optimise one layer in isolation from the others.
A company spends a year migrating to cloud but the data stays in the same chaos it was before. Result: they pay more for infrastructure and speed has not improved.
Another variant: they invest in data unification but the deployment infrastructure stays manual and slow. The data is clean but the product still ships rarely.
A third: they automate deployment via containers but every service works with its own copy of data, without shared semantics. Deployment is fast; integrations are a nightmare.
The layers are connected. The bottleneck migrates.
What to check in your company now
Three questions that will give you a bearing:
- Where is the main bottleneck today - in infrastructure, in data, or in the development and deployment process? This determines where to start.
- Does the team understand how these three layers are connected in your specific architecture?
- Does the plan for next year include movement in at least two of these directions - not necessarily simultaneously, but in a coordinated way?
If the answer to the third question is "no" - chances are that in 2015-2016 you will again find yourself in the situation where improving one layer does not produce the expected result.