With these IoT discussions I am like a dog with bone. I find it maddening that most commentators make what I consider to be one or more of the following four very erroneous assumptions: 1) the IoT and IPv6 are part and parcel of the same thing and 2) said in a slightly different way, that IPv6 was the mandate for the IoT and 3) the IoT is just another name for the evolving, growing internet and 4) the entire IoT all needs to have the same architecture.

So my latest reply to a EETimes article was as follows. (Oh – the article was entitled ‘Namesdropping: The Many Names of the Internet of Things’).

“Entertaining article – thanks, but I feel as if it were somewhat misleading.

First, IPv6 was conceived  Рas I recall Рnot to provide the addressing scheme for the coming IoT but rather as an end all/be all answer to the limitations of IPv4.

Second, the IoT – to my way of thinking – was not just another dreamed up name to speak to the evolution of the internet as imagined through the lense of IPv6 but rather was to speak to the upcoming (100’s of billions) of things in the form of [mostly commercial/industrial] sensors that would take their places – in some yet to be determined fashion – on the internet.

Third, these things – these, sensors things – do not (if taken in a commercial/industrial setting) necessarily fit the same communication paradigm as those smart computing devices utilizing TCP/IP. I refer to daCosta saying that most sensors will/could use mostly unidirectional ‘chirps’.

You didn’t speak to the following but I would like to finish my argument and say in conclusion:

Fourth, these ‘chirping’, mostly unintelligent sensor devices will not have (aka not need) the requisite memory or processing power to carry the IP protocol stack or need to/want to participate in processing packets where the addresses are 128 bits.

And fifth, it seems myopically illogical to architect the [industrial/commercial side] of the IoT based on IPv6 just because it is easier to envision everything – industrial sensors and all – on the same flavor of internet.

I mean think about it in the following context. An agro-business with thousands of moisture sensors. Sporadic chirps consisting of a few bits (the size of the data payload) of moisture information; this as related to a data header size that might be a 100 times larger.

Packets on the network where the data headers consistently have sizes something like two orders of magnitude larger than the data payload? That’s not good engineering.”

Advertisements