On Wed, Feb 4, 2015 at 3:15 PM, Martin Thomson <martin.thomson@gmail.com> wrote: > On 4 February 2015 at 21:24, Mike West <mkwst@google.com> wrote: > > My justification for allowing IPv4 is not IoT in itself, but the fact > that > > IPv4 is being used today, by the > > internet-of-things-that-happen-to-be-webservers-in-datacenters. > > Can you explain how those iotthtbwid devices might benefit from CSP? > I don't want to be obtuse, but I'm not seeing a case there. > Beyond the data center, I was thinking in terms of the many sensors I've seen that just export a rest or web interface. I can totally picture some management device that loads and possibly aggregates content directly from a big pile of sensors. It seems reasonable that there are applications where you'd know ip address data for your sensors, but maybe not want to generate hostnames for them due to low resources in a larger sensor network. Or equally, some similar web management/data display utility for a home user with a few sensors who isn't going to know how to set hostnames for individual devices unless their router does it for them. Those management utilities are going to be written by folk who won't know what the internal network topography will look like, and if your device discovery/pairing routines have the potential to return 192.168.1.103 instead of thermostat1.myhouse.com because someone's on an older home router, then no CSP. I agree with Mike that this maybe isn't something we want to encourage, but I do suspect it's something we're going to see.
Received on Thursday, 5 February 2015 18:51:25 UTC