No one wants to see their Robotic Process Automation project fail. Check out when and where RPA can go wrong – and learn from common mistakes.
What comes after Moore's Law?
Today's short-lived software applications and cloud services open up possibilities for more specialized processor designs
3. Cloud platforms
Finally, cloud providers offering services such as machine learning further abstract hardware from the consumers of a service. If you’re running an analytics or deep learning workload on a cloud provider, you probably don’t really care what processor or accelerator it runs on. You just care about how fast it runs and how much it costs. (Of course, you also need it to run correctly, but that’s a given no matter whatever hardware is used.)
This last trend may ultimately be the biggest driver for hardware proliferation. Large cloud providers have both the scale and the financial incentives to optimize their hardware around the specific services that they offer to a degree that will be hard for end users to do on a case-by-case basis. The downside is that this, in some respects, looks like a return to a proprietary hardware and software stack.
4. Open source software
There’s an alternative, though.
Due to the widespread use of open source software in operating systems, new types of infrastructure such as containers, and new workloads such as machine learning, software users have more flexibility in their choice of platform.
In the past, their choice of processor and platform would have been largely determined by their software vendor. You effectively chose your application vendor and your application determined the platform you’d run on – or at least which of a small number of systems you could choose from.
Large independent software vendors like to minimize the number of platforms they support. While open source doesn’t eliminate hardware dependencies, it does result in software that isn’t as tightly linked to a given software vendor’s desire to support or not support specific hardware platforms.
Users can no longer depend on being able to just wait for next year's faster processor to put in an appearance. Instead, open source software gives them more flexibility to chart their own path with whatever platforms are optimal for their particular workloads. Increasingly, these will be specialized platforms that focus on running a particular class of tasks most effectively.
[ Learn the do's and don’ts of cloud migration: Get the free eBook, Hybrid Cloud for Dummies. ]