The year was 2012. And it wasn’t a remarkable one for most people. But tech enthusiasts mark that year as the start of something truly revolutionary. It wasn’t a technological revolution of the type that people saw with the release of the iPhone. It was a much quieter technological revolution. One which would change things nearly to the same extent. But which was so small, in so many ways, that most people don’t even know about it.
But those who are aware of the history of technological innovation know 2012 as the date the Raspberry Pi was first released. It marked a huge turning point in many respects. Smart phones had proved that small single board systems could pack a lot of power. But they were closed to one extent or another. The hardware and usually the software alike were locked down. They were the property of a large company and couldn’t be easily leveraged to one’s own projects.
The Raspberry Pi changed everything though. It was a fully open system in terms of hardware. And the software was just a variation on Linux. Meaning that the entire thing was fully open to people’s imaginations. What’s more, the company wasn’t simply alright with people modifying either hardware or software. That was the whole point of it. Everything from the pin expansions to the fact that there wasn’t a default operating system included meant that people could easily work it into their dream projects.
Today it runs everything from tracking systems for public transit to smart mirrors. And what it’s turned into is something amazing. This humble little board turned into a whole industry of single board computers. But an industry based on inexpensive but reasonably performing devices now brings up an important question. Multicore architecture makes it more possible to actually push single board computers into the next tier of performance. For example, a single board computer quad core style is essentially four small but competent computers running as a whole. What might happen if these cores were chained to four more? And those to four more? The single board computers quad core style might be able to run as a full cluster.
And that’s just what has been attempted in more recent times. Carefully controlled experiments have shown some impressive results. This was usually done with 16 node clusters using the Pi Stack cluster construction technique. The experiment itself was as much proof of concept as anything. Part of it was to show that it is possible to leverage single board computers into a far more advanced setup. But it was also to determine what the most effective methods for doing so are.
It’s also worth noting that while the technique originated for use with the Pi, it’s far from limited to it. In fact, the Odroid C2 showed the best results in the tests. However it’s worth noting that the C2 has some significant differences from the tested Pi 3B. In particular the amount of ram on the C2 is significantly greater than Pi models before the Pi 4. However, the Pi 4 introduced a scaling amount of ram. The fact that it has a higher performance CPU and more ram makes it quite probable that performance would equal or exceed the C2.
But in any case the experiment more than proved the capacity of single board computers in a cluster setting. This means that extranet development with single board computers may be viable in even fairly CPU intensive tasks. And this might scale even further with the addition of specialty coprocessors. But even with raw hardware and lower end single board computers one can expect significant flexibility in performance with the addition of clustering.