WebCoRE without SmartThings


#1

Given today’s nearly 20hour and counting SmartThings outage, I’m curious if WebCoRE is architected in such a way that it can be ported to another platform?

Samsung’s lack of progress with SmartThings is becoming harder and harder to deny, and after this egregious outage one has to wonder just how much longer until they give up and pull the plug. I love WebCoRE and would hate to see it sink with the SmartThings ship.

Thoughts?


#2

Agreed 100%.

I think @ady624 has a bright future if he wants it, no matter what happens to ST. :slight_smile:


#3

I got the email from ST this morning about the outage but did not worry about it because my important pistons were working. Then around 4:30p MT, a mission-critical piston stopped working but I did not know it until 9:00p. Then suddenly ST started working again around 9:40p. If I had known that I was impacted I could have manually done something. Is there a way to detect when ST has gone south and then is back working again? Maybe using IFTTT or some other method to send a text, but I’m not sure how I would detect the failure. My mission critical piston uses an appliance module. Any ideas on how to solve this problem, short of not relying on ST for mission-critical tasks?

I agree that having WebCore running on another platform is a good idea. That would at least give us an alternative when ST has the inevitable major hiccup again.


#4

WebCore DOES run on its own independent platform.
The issue is communication with devices…
Currently said comms is via ST and that little white box you have somewhere in your house…
Yes, WebCore is tied to ST but only in as much as it relies on the ST white box for device comms.
Replace said ST box with say Wink; I’m sure Andy can get an integration going - that is not the real issue.
The real issue is you are again tied to “that liitle black box” that does the DEVICE (ZWAVE & ZIGBEE) comms.

As long as any of these boxes rely on the cloud for processing, you are hooped. Even devices that execute locally need the cloud integration with ST to be functional for the WebCore hook…

Only way around it is a LOCAL device with 100% local processing and only uses the cloud for backup and / or other integration, say something like IFTTT / Alexa.

Now there are “black boxes” that do do local processing; they’re expensive and difficult to master / implement, and may not be as open in architecture as we would like… (the old Staples box comes to mind)

Yup, I’m with you on this one… All I want for Christmas is… reliability.

J


#5

apparently ST is back (I can control via the app) but none of my webcore pistons are firing


#6

I personally don’t think cloud execution is hopeless. The ideal back end would be a run where you want container. Run it locally on your black box if you feel you need that, or in the cloud if that’s what you want.

For me the issue with cloud execution as implemented by ST seems to be that my workloads aren’t sufficiently segmented from everybody else’s. Processing my events slow yours down and vice versa.


#7

The webCoRE dashboard and piston editor runs on independent servers BUT all pistons are saved in and run 100% on the ST cloud… that’s what the storage app is for.

WebCoRE architecture is very specific to working with ST and would need to be rebuilt ground up to work on other platforms.

Sorry to be the bearer of bad news.


#8

Even if it was feasible to run webCoRE on other platforms, why would Adrian want to.


#9

That’s not terribly surprising. I’m just curious at this point about how much of the ST back end could be reverse engineered.


#10

Why? Well perhaps you weren’t affected by the 20+ hour outage yesterday, but I was.

I haven’t dug that deeply into ST internals, but for me having an open source backend container that I could run on my own hardware locally or in my own amazon or GCP account would be a dream. It’s a vague dream, but I’m envisioning something like the way dd-WRT basically open sourced the old Linksys WRT54G but with the samsung hub.


#11

LOL

SmartThings needed $1.2m from kickstarter just to get up and running… Samsung have invested many more times that since buying them out… I don’t think you understand the scale so good luck with that.


#12

Its not magic dude.


#17

My point being that I’m sure a tremendous amount of their efforts are going into scale.

A standalone solution with a WebCoRE front end wouldn’t need to scale in the same way if each instance were autonomous.

Plenty of larger projects than SmartThings have been recreated in Open Source. I don’t see anything to LOL about.


#22

I get that, but from my point of view, the system has been flawless for 99.9% of last year… A system built DIY wouldn’t ever give me that.

Being on the EU shard I was unaffected, but 20h downtime certainly sucks… I’m just saying, for the most part I find ST super reliable. My system is setup such that everything still works manually via wall switches if required, so downtime has less of an impact as well.


#23

I’m just guessing here but my guess is that ST is made up of something like a bunch of microservices that publish and subscribe to events from a message broker and probably some other ancillary services around that.

It’s becoming much more feasible to package up a microservices architecture like that as a single containerized “distribution”. Much more feasible than when ST started.

Yeah its work to recreate that backend, maybe a lot of work, but its not hopelessly undoable.


#24

I can’t imagine anyone hacking the ST hub to run their own code on it anyway… you’d need to crack one hell of a lot of security and encryption.

There are open source raspPi setups and webCoRE is open source (it’s all on GitHub including the dashboard side of things).

Maybe one day someone much cleverer than us will have a stab at it.


#25

99.9% is interesting. I’d pay for an SLA that offered, even that level of guarantee, but they don’t offer one.

I still don’t understand the SmartThings business model. It seems a bit like the underpants gnomes to me:

  1. Build ST
  2. ??
  3. Profit!

That’s fine for now if you’re just playing with ST, but I suspect that more and more of us are beginning to rely on it in ways that when it goes away leave us seriously in the lurch.

If I understood how they planned to make money, or if I had some contractual relationship with them I’d be more comfortable. Right now how can you really complain about a service that’s basically free? You get what you get. If it goes down for 20 hours well what do you expect? It’s free!

Also, who knows if/when samsung is going to lose interest? What if they take their code and go home?

It’s these contingencies that make me wary of relying too much on them, and yet I have tasks that I very much want to automate!


#26

If you look at the last Samsung financial report you’ll see that ST is making a huge loss.

The bigger picture is IoT for all the regular Samsung products that do make a profit.

ST is the only user friendly HA system I know of that gives such a powerful API allowing for things like webCoRE to run… the more reliable alternatives are less functional… so it depends what people want I suppose.


#27

I would gladly pay a fair monthly service fee for Smartthings and Webcore to ensure they stick around. This stuff is far from free to maintain.


#28

All we want is speed and reliability. As long as the cloud isn’t cloudy, I’m happy with it.

I see this cloud business about as reliable as our local internet connections were in 2002. It’ll take a little while to solidify, but eventually we’ll get to that 99.9% number.

I’m still enamored by the idea of fully localized home automation, not reliant on anything except power. But might not be necessary.