earlier replies
Replying to @clemensv
@clemensv Sure. So is the Web of Things, e.g. @connectevrythng. You gonna bet against the Web, again? ;)
1
@distobj @ConnectEvrythng I'm not betting, I'm observing what our manufacturing industry customers are doing and what their requirements are
1
@clemensv Early days. I wouldn't put much faith in adoption by anybody who doesn't grok network effects, or why pubsub sucks for this
1
@distobj I see you're trying to guess a few elements like MQTT or pubsub. It'd be neat if things were that straightforwardly wrong.
1
@clemensv I'm firmly with @distobj here: Even if the Web sucked for this purpose, the incentive to just stick with it is going to be huge
1
@stilkov @clemensv @distobj Don鈥檛 agree that one protocol should be applied to everything. There鈥檚 cases when its weaknesses hurt the prob
1
@kellabyte True, but the greater the network effects, the more tolerable the inefficiencies. Compare MQTT value prop with WAP circa 1998
3
@distobj I disagree to a point. If I'm requiring high performance it makes no sense to take the hit of an inefficient protocol.
1
@kellabyte Totally agreed, but I think high perf is a tiny part of this application space
2
@distobj @kellabyte but high-scale persistent bidi connections with tiny overhead are
3
Replying to @clemensv
@clemensv @distobj @kellabyte I'm betting on HTTP/2 for that

May 15, 2013 路 3:52 PM UTC

1
Replying to @stilkov
@stilkov @distobj @kellabyte "we'll multiplex all the world's TCP traffic over 443" isn't all that unlikely, but I wouldn't call that HTTP
2
@clemensv @distobj @kellabyte I agree if we're talking about something like Websocket, but HTTP/2 very much *is* HTTP