Once in a while we see echoes of things past in new technology development, in this case, the Internet of Things (IoT) has shifted an old methodology to its zenith of maturity. In both hardware and softweare, managing the IoT requires us to think of devices as clients again, an echo of things past from the beginnings of the computer age.
Big boxes
Once upon a time, computers were huge complex machines that not only cost a lot of money, they also needed a crew of people and their own special environment to function in. A smart person recognized that since a computer, even the valve-driven behemoths of yesteryear, think way faster than people can. This enabled a company to buy one computer and put it in the basement for everyone (with access privileges) to use from their office, or a common room. Each user’s position was simply a dumb terminal, a “Thin Client” because it had no processing power on its own.
As computing progressed, power moved to the desktop (and eventually the hand), enabling companies to retire the basement mainframe and replace it with a communications, memory storage, site-hosting server. This was not thin client, as the processing power used in the software by the device was contained within it, with the server’s role being only data transfer and offsite data storage.
With the advent of the Internet of Things and the creation of cloud-dependent devices, we see a slight return to the old management paradigm of Thin Client. There is still plenty of processing power and memory in the user’s device, but now an offsite server, linked to the device by the cloud, now does some of the heavy lifting. For example, a handheld device can now use offsite processing power for a task like real-time language translation.
The complete picture is of course far more complex, with data, memory, processing, timekeeping, and a plethora of other functionality, much of which has not been invented yet. The IoT is now a community, not a server-client relationship but more and more a weave of devices, functions, and roles. These roles will be determined more by haptics considerations and product/service convenience to the user more than by technological capabilities.
Is the server dead?
As we migrate forward, the massive computing power in our devices will be harnessed to create huge intelligent mesh networks, sharing the processing needs among them. One day your “phone” (we need a new word for a personal general multimedia communications device, I wish I could think of a word as cool as “ansible”) will be a node in a massive computing entity, a “Mesh” supporting other devices when you aren’t using it (although there would have to be safeguards against “define Pi” questions).
This does not mean that the fixed site server farm is dead, far from it. The need for secure storage and redundant management and oversight, plus the need for additional functionality that may not be supported by the Mesh, demands some kind of secure facilities in the future. Not to mention that this migration could take decades, not because the tech cannot achieve it, but because the legacy systems and infrastructures would take that long to replace. However, I may be pessimistic on the time scale.
