[ih] History of IoT

[Michael Thomas]

On 2/13/23 12:31 PM, Toerless Eckert wrote:
> So, what is the proper definition then for an IoT device ?
>
> Sure: "got to have a proper host IP stack".
>
> But how "embedded" into the device does that stack have to be ?

I doubt there is a proper definition, but my take is that it's a 
(mostly) headless host using embedded software which isn't involved in 
the control plane of the net. That is, it's host with its own embedded 
IP stack.

>
> We still have today likely the mayority of supposedly "IoT" devices
> such as supposedly "Smart" meters that effectively are composed
> modularily from a "stupid" device with some form of serial
> interface connecting to a "gateway" device with the host-stack
> (and web server and etc. pp.) running on that gateway.
> And in most other "IoT" devices you may not see this modular
> design, but they are still built in the same way. That's what
> arduino, rasperry style SBC are popular for.
When I first heard about "IoT" is was in the context of ipv6 and how its 
gigantic address space made enabling them much easier. I want to think 
it was Jari Arkko that gave the talk, but could easily be misremembering.
>
> My bachelor thesis in ca 1987 was to implement color postscript
> rendering for a printer, but i quickly gave up on trying
> to get the whole postscript rendering software to run on the
> printer itself because there was absolutely no diagnostics
> (download 20MByte source code.. wait, nothing happens - doh !),
> so i ended up using a Sun workstation with lpr and the renderer
> connecting via the printers serial port to only download and
> print the rendered page image. Was that combination of printer
> and Sun workstation an IoT device ? Are medical systems such
> as XRay machines that pretty much continue this approach
> until today IoT devices or not ?

The company that contracted us to design the printer and software 
managed to get postscript on it, but not the Adobe stack partly because 
of that, but probably more likely because of the licensing fees.

Was your printer an IoT device? I'd say no: it was just using normal 
host protocols to do what it did locally in the first place (ie, lpr). 
And shoe-horning LPRd into the printer itself was painful because lpr 
expected a disk to buffer everything. As I mentioned, I almost trotted 
up to ISI because of it.

Mike