Re: Representative sample of industry protocols

> On 24 Oct 2016, at 17:28, Benjamin Francis <bfrancis@mozilla.com> wrote:
> 
> On 24 October 2016 at 15:51, Dave Raggett <dsr@w3.org <mailto:dsr@w3.org>> wrote:
> To show the challenge we’re facing with integrating the Web of things with existing protocols, here is a representative sample of industry protocols from a report from Aruba Networks (Connect and protect: Building a trust-based Internet of things for business critical applications).
> 
> See: www.arubanetworks.com/assets/wp/WP_ConnectProtectIoT.pdf <http://www.arubanetworks.com/assets/wp/WP_ConnectProtectIoT.pdf>
> 
> The shear number of such protocols suggests that we need to look for a scalable approach to working on protocol bindings where the work is largely done by other organisations. Picking them off one by one for the IG plugfests isn’t going to get us there, although we will learn a lot on the way.
> 
> Exactly. Why would you even try to standardise bindings to all of these protocols?
> 
> Why not define a single protocol binding for the Web of Things (HTTP, upgradeable to WebSockets for events)? Device or gateway implementations can map HTTP to non-web protocols on the back end wherever necessary.
> 
> As an example, my team is currently working on a REST+WebSockets API for a gateway which uses ZigBee and ONVIF/WiFi on the back end. Another team previously created bindings to Z-Wave.

I love HTTP, and initiated its standardisation in the IETF in the nineties, but the wide variation in requirements across different application domains means that the Web of things can’t be limited to HTTP and WebSockets.  That said, I agree that defining protocol bindings to HTTP should be a high priority given its widespread use for connecting gateways to the cloud.

—
   Dave Raggett <dsr@w3.org <mailto:dsr@w3.org>>